WELCOME FROM NIMH AND NIDA
JENNIFER HUMENSKY: This meeting was developed by NIMH to showcase health economics research in the areas of mental health and substance use disorder research, funded by NIMH and NIDA.
This meeting is led by three health economists at NIMH, Leonardo Cubillos is the Director of the Center for Global Mental Health Research at NIMH. Sarah Duffy is the Associate Director for Economics Research in the Division of Epidemiology Services and Prevention at NIDA. And I’m the Chief of the Financing and Managed Care Research Program in the Division of Services and Intervention at NIMH.
This meeting includes over 700 registrants, with academic researchers, federal and state and local policy makers and advocates from all over the US and across the world. We are excited to share with you the research that will be highlighted today.
We have three panels that will highlight three key topic areas: payment and financing, behavioral economics, and the economic implications of social determinants of health. Each of these panels include three presentations, each representing the research areas of our three respective divisions in mental health, US domestic, and international, and substance use disorder.
In addition, we have three plenary talks which will also go in a little more detail on some topics that may be of particular interest to this audience. Darrel Gaskin is a Professor at Johns Hopkins University and will be discussing the economics of health equity. Varun Gauri is a lecturer of public and international affairs at Princeton University, who will be discussing biases and mental models in healthcare. And Mark Fendrick is a professor at the University of Michigan and will be discussing value-based insurance designs.
We hope this NIMH health economics collaboration will grow to include future meetings and other opportunities to help contribute to the growth of the health economics community.
While unfortunately a virtual meeting doesn’t give us as much opportunity for interaction as an in-person meeting would, we hope you can interact with speakers through Q&A to help facilitate a dynamic discussion. After the meeting the presentations will be made available on the NIMH YouTube page in about six to eight weeks. In addition, a special issue of the Journal of Mental Health Policy and Economics will highlight speakers from this meeting, so please stay tuned for that issue.
We will start with brief introductions from Josh Gordon and Nora Volkow, the Directors of NIMH and NIDA respectively. Josh and Nora are both enthusiastic for this health economics collaboration, but they were unfortunately unable to join us in this time slot, but they still wanted to welcome you to this meeting and share their enthusiasm for the work presented here. We will share these two short videos from Josh and Nora, and after that Leo, Sara and I will each describe our respective divisions and health economic areas of interest for each. So let me go ahead and introduce Josh Gordon.
JOSHUA GORDON: It is my distinct pleasure to kick off this first ever NIMH/NIDA collaborative meeting on health economics. NIMH has a long history of investing in health economics research, and these investments have since transformed mental health service delivery.
Just one example, beginning in the 1980s Dr. Agnes Rupp, an NIMH Program Officer, who many of you know who has since retired began developing an extramural research portfolio aimed at examining the economic feasibility of parity for mental healthcare and private insurance plans.
That program of empirical research eventually provided the economic rationale for the Mental Health Parity and Addiction Equity Act of 2008, legislation that has increased access to care for millions of people living with mental illness in the United States.
Today NIMH funded health economics research continues to address questions that are vital to improving mental health treatment, service delivery, and outcomes, such as structuring effective telemental health service delivery in the era of COVID< establishing and testing payment models to incentivize delivery of high-quality care and promoting economic wellbeing for people with serious mental illnesses.
Our health economic research impacts mental health both domestically and globally, particularly in low and middle-income countries, by ensuring that our services, research, and research training in sub-Saharan Africa, South Asia, South America, and elsewhere, results in scalable mental healthcare solutions appropriate for countries attempting to make the most out of scarce resources.
Today’s meeting represents a true collaboration between NIMH and NIDA. At NIMH our Division of Services and Intervention Research and our Center for Global Mental Health Research highlight research in the US and around the world that can help to improve mental health service delivery.
And we are thrilled to be joined by our NIDA colleagues to examine research addressing key issues in the delivery of substance use disorder treatment. I hope that you will take full advantage of this virtual opportunity to engage with speakers, ask questions, and connect with your fellow health economists, policy researchers, and others who are doing the best work to improve the value and efficiency of mental health and substance use disorder care. We look forward to continuing to work with you to ensure that your research has a maximal public health impact. Thank you for joining us.
NORA VOLKOW: Hello everyone. I am Nora Volkow, and I direct the National institute of Drug Abuse. Like NIMH, NIDA has a long history of supporting rigorous health economic research, including tools and methods for economic evaluation of substance use prevention and treatment services.
And as he mentioned in his introductory remarks, the work has enabled for example the parity laws that now provide full reimbursement in screening and treatment of individuals with mental illness, but also with substance use disorders. And that would have not been possible if it were not for the health economic research done in both of our institutes.
We are also very interested and have been funding research on cost effectiveness and cost benefit for novel interventions, because ultimately, they will determine the extent to which they are sustainable. We also welcome research on economic simulation models to help inform domestic as well as international interventions that are going to be likely to have an impact.
And that pertains in general at NIDA for our investments on substance use in general, but also very specifically in terms of injection drug use and HIV. And all of these continue to be extremely important areas of research, at NIDA we all continue to be investing.
However, in the meantime, the substance use disorder field has been shifting pretty dramatically, and there is need for more health economic research to help bring that knowledge into products that have been applied, services, and also for models that are more sustainable.
What are some of those changes? Certainly we have advanced an enormous amount in our understanding about the changes that happen in people who become addicted to drugs, or the factors that make someone vulnerable to take drugs. And those in turn are informing us in how to optimally prevent and treat.
Also what has happened I would say over the past two decades but accelerated since 2016, is the diversification of the types of drugs that are available for people to consume, extremely challenging problems that are driving the horrible overdose lethality that we’re observing in our country. And that requires new strategies, new models that evaluate what would be more effective.
Also, there are advances in novel effective interventions and technologies that are being deployed and understanding how they compare with other more traditional venues is crucial in their long-term implementation.
And very fundamentally I think that as we look into the country, how are we going to be providing these services, we have to realize that healthcare is one of the most important infrastructures that we have for screening and treatment of substance use disorders. And so how do we develop new models of care and delivery has been one of our priorities, but we need to evaluate the economic outcomes and their sustainability.
At the end of the day, it is health economic research that informs on what should be the reimbursement level, what should be required for proper coverage, how should those benefits be designed so that the largest number of people can benefit from it.
And I think that this is what this forum, this partnership between NIMH and NIDA are aiming to achieve. Unfortunately, I will not be able to stay for the meeting because I have another scheduled that I have to attend, but I really look forward to hearing your ideas, and I see this as an opportunity to create a dialogue that is in both directions, and to strengthen the partnerships between NIMH and NIDA to advance health economic research. So thanks very much, and I am delighted to welcome you to the meeting.
JENNIFER HUMENSKY: Thank you. Quickly, to respond to a question that came up over the Q&A, these recordings will be available on the YouTube page. It takes about six to eight weeks for the recording to be prepared for public release, but then it will be made available.
So I’m Jennifer Humensky, I’m the Chief of the Financing and Managed Care Research Program, or the Health Economics Research Program, in the Division of Services and Intervention Research at the National institute of Mental Health.
I’m also currently the Acting Chief of the Research Methods Program, which is also an area of interest to many attendees today. This morning I will give you an overview of a health economics research interest in the Division of Services and Intervention Research. Our work focuses primarily on issues pertinent to the US Domestic Mental Healthcare.
In a few minutes my colleague Leo Cubillos will describe the research interest of NIMH’s Center for Global Mental Health Research, and then Sarah Duffy will describe where to find NIDA’s health economics research interests and the importance of economics to the NIMH mission.
NIMH’s research interests are laid out in our strategic plan, which is released very five years, and it outlines NIMH’s overarching goals. NIMH’s goals follow the spectrum from basic science to treatment development to testing treatments on patients at bedside, to disseminating effective treatments to the larger population. Much of our supported health economics research would fall under strategic objective four.
The five-year strategic plan is updated annually. The most recent update was released in June of 2022, highlighting some particular areas of ongoing challenges and opportunities for innovation. These include COVID 19, suicide prevention, early intervention psychosis, mental health equity, digital health technology, among others.
The goal of our Health Economics Research Program within the Division of Services and Intervention Research, is to expand the understanding of the role of economic factors in the delivery and use of mental health services and mental health related behaviors and outcomes.
We think of this both from the supply side, that is, the economic factors affecting incentives to providers and payers, for example through reimbursement and financing of mental healthcare services, and from the demand side, that is, how economic factors impact clients and families. This would include social determinants of health such as employment, housing, and other socioeconomical factors that can influence an individual’s experience with mental health services and mental health outcomes.
Additionally, we would like to note that all economic research must be consistent with the NIMH notice of priorities for health economics research, and you can google the notice here. NOTOD 1605. It states that economic research supported by NIH must have as its goal to improve a health-related outcome, such as access to treatment, services, or clinical or functional outcomes.
Analysis that seeks to reduce costs for example without improving quality of care would generally not be consistent with this notice. Lea and Sarah will be speaking in more detail about this notice in a few minutes.
But first I wanted to highlight a few funding opportunity announcements that are listed on the NIMH website that may be of particular interest to this audience. We’re participating in a few dedicated funding announcements, including this RO1 opportunity to examine the role of employment in addressing health disparities.
We also have some notices of special interest, or what we call NOSIs, these are published notices of NIH public interest on a particular topic. And generally, an applicant would apply to a more general funding announcement, and would cite, there’s a space on the application to cite this particular NOSI that the application is responding to.
We have a few NOSIs that may be of particular interest to this audience, including strategies to enhance mental health interventions and services within an employment and job training settings. COVID19 pandemic era mental health research, which is a more general notice. And even a more specific one on the social, behavioral, and economic impact of COVID19 in underserved and vulnerable populations. And each of these notices can be googled or are readily available on the NIMH website.
But we would also like to point out that now is a great time to approach innovative work that aligns with the research interests outlined in the NIMH strategic plan. Up on the slide right now are announcements for our Innovative Mental Health Services Research Pilot R34 and R01 announcements. And additionally, investigators are always welcome to apply to the NIMH parent announcements.
Finally, I would like to highlight a concept that we have recently cleared through NIMH council. This represents a joint collaboration between our Division of Services and Intervention Research and NIMH’s Center for Global Mental Health Research with my colleague Leo Cubillos. And highlights our joint interest in developing quality measures to enhance mental health access and outcomes.
So the concept clearance is available on the NIMH website. Certainly, one could reference that available on the website. And please stay tuned for more to come in this area. And with that I will turn it over to Leo.
LEONARDO CUBILLOS: Thank you Jennifer. Good morning, good afternoon, and good evening, everyone. Thank you for joining this Health Economics meeting at NIMH and NIDA. My name is Leonardo Cubillos, and I direct the Center for Global Mental Health Research. And I am going to share my screen right now to tell you a little bit about the Center for Global Mental Health Research and its interest in health economics.
So the Center for Global Mental Health Research’s main goal is to generate knowledge to improve the lives of people that are living with or are at risk for developing mental illnesses, particularly in low and middle income countries. Our core work is to expand the science which facilitates or informs decisions around scaling up and adoption of evidence-based interventions, in LMICs mostly, and these interventions can be across the spectrum from preventative interventions to very heavily gated interventions.
Our core is to close that treatment gap which is described in the literature by closing not only the access gap but also one of the Right to Health Frameworks, the accessibility and availability, applicability, acceptability and quality of care maps. In funding research, we are also very interested in continuing to strengthen our research capacity in low- and middle-income countries.
And also in the potential for multi-direction knowledge exchange. That term is sometimes called reciprocal learning, sometimes called reverse innovation, sometimes called south/north learning, which is essentially what can high income countries or middle-income countries learn from low-income countries respectively.
So it is a shift in a way from the traditional paradigm. We know that for example interventions that have been developed and tested in South Asia or in Sub-Saharan Africa, are now being adopted, and sometimes even delivered in the domestic setting here in the US. So we are interested in also capturing those multidirectional exchanges, and to foster them as well.
In the spectrum of research, all the way from discovery research to services research, the Center for Global Mental Health Research is on the left side of this slide, on the services/research perspective, with a heavy emphasis on implementation science.
And at the same time other divisions within NIMH are working in other areas of domestic mental health research with some portfolios in the global setting either of translation of discovered research or in the case of additional past research in the intersection of AIDS and mental health.
The Center for Global Mental Health Research is one of the funding offices of the National Institute of Mental Health, and in green is the objective that I mentioned earlier, we have four key activities, of course the primary one is supporting research, but we are also heavily investing in the capacity to link in partnerships and in the area of communications.
The six areas that we think are of primary importance for us as far as research is concerned is suicide prevention, integration of mental healthcare, social determinants of health, human mobility in mental health, mental health systems, and training and career development. And if you see some of these priority areas are sometimes also a point mentioned here in the US as important areas for services research, mentioned by my colleague Jennifer earlier.
Suicide prevention in mental health systems are specifically mentioned in targets 3.4 and 3.8 of the Sustainable Development Goals, and we keep very close support of the achievement of those Sustainable Development Goals. Training and career development is very important for us as well. And in supporting this research we encourage our investigators we support to utilize, to advance the use of implementation science or attracting new social sciences to the field of implementation science, hence this meeting.
The use of data science, we believe that experimental trials are very important in the field, and that there is also an immense wealth of secondary data out there that comes from different sources that could be used for the benefit of global mental health, and of course using new technologies like (indiscernible) technologies and the likes of them.
Jennifer mentioned the Notice of Interest that clarifies NIMH priorities for health economics research. I copied one section of that notice, which is the highest priority areas of NIMH, and I put in bold and highlighted the second bullet, which reflects in many ways one of the areas that is of highest priority for the Center for Global Mental Health Research.
I’m just going to read it. To understand behavioral, financial, and other factors that influence the communication, adherence, dissemination, and adoption of medical discoveries into health care. So using the slide that Jennifer presented earlier, it could also be construed as understanding more what happens in the supply side of the elevated health systems in low- and middle-income countries.
This slide has in red those funding opportunities that have expired, but I think it is important that we’re thinking around these areas. And in black, the funding opportunities that are current. I encourage you to take a look at them, to study them. We mentioned the concept note and quality measures that Jennifer mentioned.
And this last slide presents one of our active calls for applications, which is the social drivers of mental illness in low- and middle-income countries. Again, this is in the path of social determinants of health that I mentioned earlier.
The objective of these calls for applications is to solicit research that we hope to identify and explain the mechanisms by which one, interventions targeting social drivers of mental illness effect the mental health functioning, and also to understand interventions that prevent, detect, and treat mental illness that impact the functioning of children and adolescents living in low and middle income countries, adolescents understood as younger than 24 in this case.
The date for reception is February 17th. So thank you or your attention, and thank you for joining us today. We are glad to invite my colleague Sarah Duffy from the National Institute of Drug Abuse to join us now.
SARAH DUFFY: Thank you so much, Leo. So, like Nora, I want to thank NIMH for thinking of and inviting us to collaborate in this meeting, and also all of you for attending.
So, why would NIMH and NIDA be interested in health economics research? I think Jennifer and Leah already mentioned this, but I like to go a bit more into why economics as a specific discipline. One reason is that the resources available to produce healthcare services, and really everything, are limited. There is not enough to produce everything everyone could possibly want. Economics is in essence the study of how to make the most of those limited resources to best meet the needs of society.
Another reason is that economic factors, prices, wages, budgets and so forth, all affect healthcare decisions, including what care is produced and what care is sought. Understanding how individuals respond to these economic factors is crucial to understanding the likely population effect of interventions, for example.
Finally, economics as a discipline has long used rigorous methods. Economists tend to derive testable hypotheses from parsimonious models of economic behavior based on economic theory that directly consider the individual’s goals and the constraints under which their choices are made.
Economists then test these hypotheses using advanced econometric methods that for example estimate causal relationships even in the context of the unobserved variables. We certainly don’t always get everything right. There is room to improve. But that level of rigor, especially when combined with the strengths and insights of other biomedical disciplines, as you’ll see today, can lead to meaningful results.
This being NIMH, also as Leo and Jennifer mentioned, the focus needs to be on health, including health outcomes and health related behaviors, such as utilization.
For example, studies that estimate the potential effects of a given intervention on healthcare utilization, or that examine the effects of financial measures on implementation discoveries, or how various factors affect health disparities, are all likely relevant. Those for which the focus is not health, but on things like directly on wages, profits, or market structure, are important areas of research, are really not considered relevant to NIH.
And again, you can take a look at NOT-OD-16-025 for more, but also please feel free to reach out if you have any questions about the relevance of your potential application.
So I think Nora did a great job of describing NIDA’s current areas of interest. But I wanted to show you some other places where you can find them. One of the main ones is NIDA’s strategic plan. This includes a direct reference to health economic studies in the context of implementation science but covers a wide variety of areas that might also be the subjects of health economics research.
The vision of epidemiology services and prevention research, of which I’m a member, also has several notices of special interest, which you can find in the NIMH guide. The three I am showing here are those from each of our branches. NOT-DA-23-012 describes how our treatment services research areas of interest.
In short, we’re interested in rigorous health services economic research related to the availability, delivery of efficient, effective drug, alcohol, and tobacco treatment from recovery support services. We give you some examples in the notice but welcome your ideas.
Likewise, our prevention research branch in the NOSI references the importance of understanding economics and prevention to the long-term dissemination of effective program practices, including research that analyzes the cost of adopting evidence-based interventions, and also the cost relative to benefits of delivering them.
Our epidemiology research branch is especially interested in applications that enhance our understanding of the nature, extent, distribution, etiology, comorbidities, and consequences of drug use, misuse and addiction, that contain a well-described path towards translation and/or public health impact. Specifically called out are studies that address variation in broad structural factors, including access to resources such as housing and healthcare, and studies that can support causal inferences.
Other places to look for NIDA’s interest are our approved concept page, which describe research ideas we’re thinking about, Nora’s blog, and papers published by NIDA leadership. As Nora mentioned, our field moves rapidly, so it is good to check these places to see what we’re thinking.
Thank you again for attending and know that health economics research is relevant to NIDA, and please feel free to contact me. With that, let me un-share my screen. And I think we can move on to the next session of this. We have four distinguished scholars, which will be the Payment and Finances Session.
We have four distinguished scholars who will address various aspects in this area. First, we are going to have Dominick Hodgkin, Ph.D, from the Heller School of Social Policy and Management, Schneider Institute on Behavioral Health at Brandeis University. He will talk about issues in paying providers of substance use disorder treatment.
Next will be Dan Chisholm, Ph.D, of the Department of Mental Health and Substance Use at the World Health Organization Headquarters, who will speak on financial protections for mental health conditions.
And then finally we will have two scholars from RAND, Alex Dopp, Ph.D, of the Department of Behavioral and Policy Sciences, and Marylou Gilbert, JD, of the Pardee RAND Grad School and the RAND Center on Housing and Homelessness. They’ll talk about a tool for developing a sustainable financing strategy for evidence-based practices. Please again feel free to enter questions in the Q&A as we go along, and we’ll turn to them at the end. Dom, thank you for being here, I think I’m sending it over to you.
PANEL #1 – PAYMENT/FINANCING
DOMINIC HODGKIN: Thank you Sarah. I have no conflict of interest to disclose. And I’m going to be talking about work that’s been funded by the NIDA Center that’s joint between Brandeis and Harvard about improving system performance of substance use disorder treatment.
So we recently got renewed, the center is now called SPIRE, and so we conduct and extend research on SUD service delivery organization and payment methods. It is led by Sharon Reif and Haiden Huskamp, and various people named on this slide gave very helpful feedback on the research being presented here.
The general context is thinking about the way that payment and delivery are intertwined, and if they’re not connecting well how that can affect the sustainability of new interventions.
So as was coming up in some of the introductory comments by Sarah and others, there is a lot of innovation going on in behavioral health delivery. Researchers and others are continually developing new interventions and delivery models, things like screening brief intervention, referral to treatment or experts.
So a lot of this research into new intervention and delivery models if funded by governments and foundations, and the funding typically takes the form of time-limited grants to develop and test the new interventions.
And the original grant funding is time limited, so eventually communities, clinicians and others must find resources to replace the grant funding. And if they can’t fund the services then the innovation won’t be adopted across the health system, and that won’t happen even if the research had shown that it would be effective and cost effective to society.
So Richard Frank and others did similar work on this 20 years ago at collaborative care models for depression, which was being found to be effective and cost effective, but was not being adopted to scale. So this is a non-problem in a lot of areas of healthcare more generally.
You can find too many quotes like this. Two frequently efficacious interventions implemented in schools and communities through grant funding failed to survive the withdrawal of that funding. So it is great if your research is getting solidified, but if that funding is evaporating then who does it help?
So you could ask the question, well if these interventions turn out to be effective and cost effective, why doesn’t the health system automatically fund them? So one piece of this is released of payment. I’m not going to claim that payment is the only thing going on, there are many other issues, including cultural ones. But since this is an economics conference it seems fair enough to focus on payment as having a big role.
And in the US, fee for service payment is still the dominant way that individual healthcare providers get paid, and especially true for behavioral healthcare. So when these novel services come out, they may lack billing codes initially, and those take a long time to get approved and implemented, and then even once their billing code is approved then each payer or state has to adopt those new codes.
And these innovative models I’m talking about may include components that fit poorly with the fee for service payment model. Sometimes new models include services that don’t involve patient contact, like physician-to-physician consultation without the patient being in the room.
The model may include services that are not healthcare. It could be peer support, carpet cleaning for asthma patients, temporary housing. And so some of these are addressing social determinants of health, but that is not part of the traditional health insurance model. And also some of these interventions are designed without considering how well they will fit with the payment system.
So to kind of recap from the provider perspective they’re delivering a bunch of usual services that they can fund out of these traditional funding sources like Medicaid and employer health insurance, but they also want to deliver these novel services, but those services come out of a different funding stream that is typically time limited, like a targeted grant.
Here is an example of the timeline issue, where 20 years ago SAMHSA awarded five-year grants to six states and one tribal council to promote the adoption and sustained implementation of screening, brief intervention, and referral to treatment or SBIRT. And the states in turn selected sites where they would fund the SBIRT implementation.
And after that grant ended about 30 percent of the sites stopped retaining the SBIRT services. That’s not a bad retention rate, but still. And then SAMHSA went on making other similar grants. But it took until 2007 for Medicaid and Medicare to add billing codes for SBI nationally, so only for part of the intervention, the SBI part.
But ten years later, on behalf of the state Medicaid programs that actually activated those codes, so it’s nice that CMS approved the codes, but for providers to get paid in the states each state Medicaid program had to then activate the code, and that only happened in half of them. In fact, even activating the code is only the beginning of things, that may not be the end of the story.
So the question posed by Alex Cowell and others in a paper was whether SBIRT could be sustained in the wild after it had been presumably bred in captivity.
I’m going to talk about a more detailed example, current office-based opioid treatment, or OBOT, and then finish up with some conclusions for sustainability more generally for novel services.
So in the case of OBOT, it is a delivery model that is intended to make medications for opioid use disorder, particularly buprenorphine, widely available in general medical clinics and offices. And there is evidence supporting the effectiveness and cost-effectiveness of this model. The uptake of the OBOT model in the US has been relatively slow.
One issue, until recently physicians had to obtain federal X waivers to prescribe buprenorphine involved applying, doing a day’s training, things like that. And that requirement was recently pretty much relaxed. But until then less than six percent of physicians were actually going and getting these waivers. And even among those who got the waivers there were low rates of prescribing buprenorphine.
But stepping over to the organization perspective, one important barrier to faster diffusion of OBOT may be the financial challenges facing the clinics that could be adopting it. And in particular it has a couple of components that raise these issues I was talking about that fit with a payment model.
So the OBOT model includes coordinated care management, where a clinic staff member like a nurse or social worker will coordinate the patient follow, the prescription refills, the drug testing, and queries to the prescription drug monitoring program. Often it is a nurse care manager model.
And similarly, the OBOT model will include technical assistance so the clinics that adopt, including both training sessions and provider to provider consultations like when primary care doctors call into an addiction specialist who can give them guidance about treating patients. But again, these novel services are not necessarily billable in fee for service systems.
So if we think about it from the point of view of the clinic that is deciding whether or not to adopt OBOT, the provider organization in a fee for service environment, it is delivering only two types of services, just to keep it simple. One type is billable, B, and the other is non-billable, N, and there is a separate fee for each type of service, then you could write the clinic’s operating margin as the revenues from the billable services minus the costs, which are driven by both billable and nonbillable services.
So if a clinic is deciding whether or not to adopt OBOT, it needs to consider the likely impact on several different factors. One is what will adoption do the volume of non-billable services. Well, it’s going to increase those, and that will increase cost, but how much? The impact on the volume of billable services, because adopting OBOT might generate new revenue on the pharmacy side for example, and then what will it do to the cost of delivering billable services, is it possible that adding OBOT would reduce how expensive it is to deliver billable services.
So another way to think about it is what are the ways you could pay for these non-billable services. And there are some different possibilities. I will go through. Which ones are available depend on what is going on in that state and what the payer rules are. I won’t read them here, but the first possibility is that the clinic could cross-subsidize those non-billable costs if it can receive higher fees for billable activity.
And an example here is in Virginia, where Medicaid lets waivered clinicians apply to become quote preferred OBOT providers. So they have to meet certain requirements. If a provider is approved, they then get enhanced fees for treatment and care coordination services around opioid medications for opioid use disorder. And you could see those higher fees as being a cross-subsidy to help the clinic cover its cost of non-billable services. So that is one approach if the state is doing that.
Another possibility would be that adopting OBOT may generate additional other services that are billable, and that results in revenue that can be used to cover the non-billable services. So in one health system in western North Carolina, Farrar et al report, when they adopted OBOT they got nearly two additional visits per patient per month among the OUD patients. And those additional visits were billable, so they could use those, if there was enough of them they could use those to support a nurse care manager position.
Another possible approach is to cover the cost of non-billable services by obtaining external grants. And for example, in Massachusetts when the state started a rollout of OBOT at community health centers, the state gave block grants to the centers to cover non-billable services. But that is unusual, most states don’t have that. Some clinics elsewhere have been able to access grants from federal agencies in the US like Health Resources and Services Administration or SAMHSA.
And then a fourth possibility is to cover the costs if the payer is using alternative payment models like capitation or bundled payment. And for example, Medicare three years ago, started a bundle payment for OBOT, they created a new billing code that covered a bunch of services combined, including traditional parts of OUD treatment and also care coordination, and underlined here, that’s something that wasn’t previously billable.
And similarly various state Medicaid programs have opioid health homes that are paid on a capitated basis, as a way to cover non-billable services like care management and care coordination. One challenge for this approach is the adequacy of the flat payment, is it enough to really cover the additional services.
Another possibility is sometimes the payers may add requirements like they want to be really sure this clinic has all the services that are in the bundle. And so if the requirements are seen as onerous that may make the use of bundle codes unappealing to the clinics.
So the clinics vary a lot in what options they really have for covering non-billable cost of OBOT. It depends on the size of the clinic, its mix of payers, and particularly on what are the policies in that state. There was a recent paper by Jonathan Fried and others that did simulations about different parameter values, what would make OBOT sustainable in different environments.
So clinics considering adopting OBOT would need to see what options apply in their own state. But just to reinforce an earlier point here, what we’re seeing is that OBOT may be cost effective for society, but it doesn’t guarantee it’s going to be cost effective from the individual clinic’s perspective. And if that’s the decision maker that’s going to be determinative in what actually happens.
So turning to general lessons about sustainability, a lot of resources go into developing new interventions and new delivery models. But if those new models will end up going unused, why are we making those initial investments? Well an alternative would be at the time of making initial investments we could pay more attention to what can we do to make them more sustainable. So I’ve got some lessons for different kinds of stak4eholders.
First, for the funders of research, many funders are well aware of these issues, and would not bat an eyelid at the things I’ve been saying today. But the funders should require grantees to develop sustainability plans while they’re still grant-funded. And some already do this, like SAMHSA. The funders could actually require financial sustainability planning to be a component of applicants’ proposals, even before the funding is awarded.
And that could prepare the community stakeholders to plan better for sustainability from the beginning. But this is often quite remote from what the clinics and community health centers and so on are thinking about, so they may need some technical assistance around identifying these potential funding sources.
For payers a lesson would be to use the period of time in the grant funding to be working out sustainable payment models, and these should employ stable funding streams to cover any previous non-billable services, ideally fold the novel services into existing stable funding steams. I’m talking about things like Medicaid, employer health insurance, the providers know that we’re around for a long time, so they don’t need to say if I hire somebody will the payment still be around in five years.
Payers could also consider using flexible payment models to ensure that otherwise ineligible, otherwise non-billable services will be covered.
And for researchers the lesson is when designing these new delivery models and interventions, consider what would make your intervention more sustainable beyond the end-of-grant funding, particularly financially sustainable.
Collect input from the payers early on. You could be asking what evidence might persuade payers or states to cover the intervention if it does prove effective, what performance measures could the payers use to reward payers who adopt the intervention and implement it with fidelity.
And these discussions could have benefits like educating researchers around the real-world challenges to sustainability and engaging the payers more directly in the care delivery innovation as you’re designing it. And I’ll stop there and pass it over to the next speaker, who is Dan Chisholm.
DAN CHISHOLM: Hi everybody. Good morning. So, many thanks for the invitation to this meeting today. I’m delighted to join it. And let me start by sending greetings from WHO, and also from our Director of Mental Health and Substance Use, Delora Kestel who by the way enjoyed a very excellent day at the National Institute of Mental Health, I think it was back in September, including meetings with Dr. Gordon, Leo our cohost, and colleagues on his team.
Back in March of last year I contributed to a session, also dedicated to mental health economics and financing, where I was speaking in the context of health systems research and development. And there I think I focused more on the efficiency question. As Dr. Gordon said, we have to make the best use of available resources, particularly in lower resource countries and contexts.
But today I wanted to focus on one of the other primary questions or issues in health systems financing, and that’s financial protection, or if you like the equity rather than the efficiency dimension, and why is it an important question, what is the state of our knowledge, what are the gaps in our knowledge, and what can be done to change that. That’s the intent.
So let me start just with kind of a quick overview of what I really want to say, which is that of course as we know individuals and their households living with mental health conditions, they often face life-long, pervasive impacts as a result of lost employment or income generation, as well as out of pocket spending on treatment and care.
So given that context, it is essential that those individuals and households are protected from the financial risks associated with their ill-health, just like with any other condition.
So financial coverage or protection is a critical element of the universal health coverage, as I will show in a second. There are several ways in which that can be extended or improved, but a critical mechanism is of course through inclusion of mental health conditions in health entitlement packages or national government funded health insurance schemes, among others. So I want to unwrap all of that and talk a bit more about those issues.
Just to set the scene, I think I showed this last time, and it just provides, for me anyway, a useful kind of framework to situate what I want to talk about today. And I’ve circled those elements. It’s partly around equity, but ultimately, we’re looking at financial protection as one of the critical aims or objectives of a health system is to not just provide services that are of good quality, but to also provide appropriate protection against financial risk, so that people do not experience undue financial hardship.
So if I start first with this equity aspect, I’m not really going to focus on that, but just to set the scene and the context here, I will show you here some findings from a European survey which is reporting in a Health Equity Status Report from three years ago, which shows a quite amazing socioeconomic gradient with respect to mental health.
So this is using the WHO-5, a simple five question mental wellbeing index, and it is showing the scores that fall below 50. So that is indicating poorer mental health by income quintiles, which were reported for each of the countries, or more easily for whole subregions of the WHO European region.
So there are very substantial social gradients, with people at lower income quintiles having notably lower levels of mental health and wellbeing. So this is an important equity dimension, which is seldom reported. There is seldom data to support these. I’m sure in the US you have ample data to show this, but when we try to find data more globally and in low- and middle-income countries it is really hard to find. But if it did exist, it would likely demonstrate a similar gradient.
Let me show you this. It may be familiar to many of you, but this is a famous sort of box, the Universal Health Coverage Box. And it just sort of tries to unpack what is meant by universal health coverage. You will see it is a three-dimensional box, with both the breadth, which is the proportion of the population who are covered, financially protected. The depth, which is around the services that are available to the people. And also something about the financial coverage and what are the out of pocket or direct payments individuals have to pay at the time of use.
And of course, there is often a tradeoff to be made as you move to try and expand the size of that box, to fill it up, there is a tradeoff that is needed at a policy level in terms of equity versus efficiency. So trying to get that balance between covering the population, but with a sufficiently elaborated set of, in this case let’s say mental health services and conditions.
So this is again a kind of framing question. So let me look at first the extent of the coverage. And this is taking some data from the World Mental Health Surveys from about 21 countries, and it basically just shows the incredibly low amount of effective treatment rates or service coverage for people with depression, anxiety, and substance use disorders, both in lower, middle-, and low-income countries as opposed to high income countries.
So if we think it’s low in high income countries, an average around 22 percent, you’ll see that in lower income countries it is less than five. So we have 95 percent effective treatment coverage gap for depression.
Let’s move to the financial coverage. So this is generic numbers, not mental health specific, but to give you some sense from a recent report from WHO and the World Bank looking at financial protection. And on the left you will see the red. The red is the proportion of total health spending that is out of pocket as opposed to government financing or external aid or voluntary health insurance contributions.
So if you compare for example bottom-right, high income countries, about 20 percent, versus low-income countries, which are 44 percent, or low and middle income countries, also 40 percent. So more or less double. And you will see that the government expenditures on health, whereas it is 50 percent in high income countries, is 20 percent in low- and middle-income countries. Huge disparities there in terms of how services are spent.
And if we look at the right-hand side at the implications of that, you will see that globally the number of people who are spending more than 10 percent of their household income on healthcare expenditures is actually growing. If you look at the figure in 2000 about half a billion people rising to one billion people in 2017.
Where if you look at the bottom of those three sets of figures, the population with impoverishing health spending at the relative poverty line, which is 60 percent of per capita GDP, you also see that these numbers are rising to more than one billion people on planet Earth.
The bottom piece is a more recent set looking at data, at what are the kind of underlying reasons that people give for why they’re unable to access healthcare services. And as you see here the main driver, particularly in low-income countries, is indeed financial.
Moving to the mental health space, this is data coming from WHO’s Mental Health Atlas, which is carried out periodically every two or three years. And shows again this quite steep gradient. To compare, the top one is for paying for mental health services, whereas the bottom panel is around medications.
And just showing again this distinction between the upper-middle- and high-income countries where it is 10 percent or less of the countries where people are expected to pay mostly or entirely out of pocket, whereas if you compare that to low-income countries, more than half of mental health services rising to 70 percent for medication.
Then we think, well, what’s the inverse of that? And it is driven partly by the inclusion or not of mental health conditions in national health insurance or reimbursement schemes. And this is showing the opposite side of the decline. So h ere we see that only roughly one third of low-income countries actually explicitly include the current treatment of person-specific mental health conditions in those programs, whereas it is 90 percent or more in upper-middle and high income countries.
So moving to a research project, which is actually funded by the European Union, not by NIH. And it is quite unusual in that this was a call for health system strengthening. So a bid was put in by Professor Thornicroft and others, WHO was one of the partners on this, looking at different health systems strengthening issues with respect to mental health. So looking at governance, looking at financing, information systems.
And as part of the financing and resource work package as they’re called, there was opportunity to carry out a household survey in six lower-middle income countries, including Nepal and India, as well as in Africa, Ethiopia, Uganda, Nigeria, and South Africa.
So this was an unusual opportunity to capture data about household economic cost specifically associated with mental, neurological, and substance use disorders. And one of the most valuable aspects is that this was a controlled cross-sectional survey. In other words, we had a sample which included households which are not affected by those conditions, but by other chronic health conditions, and therefore we could look at the relative impact of these conditions on household welfare.
And just some hardline examples of results here. 16 percent of households with one member having a mental disorder reported withdrawing their children from school due to financial hardship, compared to 10 percent non-affected.
Nearly a third, 31 percent of households with a member having a mental disorder reported reducing their use of health care. These are coping strategies. And again, a significant increase compared to those without these conditions.
And a final example was even around the frequency of meals. So 36 percent versus 26 percent reporting reducing the frequency with which they had meals. So I think this is the kind of powerful evidence demonstrating the impoverishing effect of mental health conditions on household welfare.
And so we can say, we talk about this topic, that households and individuals who are exposed to adversity, it’s a two-way street. If you’re exposed to unemployment, indebtedness, inadequate access to housing and health services, education, these are all known risk factors for mental ill health. And inversely, the flipside of that coin is that the experience of mental illness for individuals and families exacerbates the level of socioeconomic adversity, as shown in the previous slide.
So the implications from that, and I’m delighted to see that some of the calls and focuses on social determinants of mental health, is that of course the extent of socioeconomic development can influence the burden of mental illness by targeting those upstream determinants. And on the UHC or the Universal Health Coverage side, as said at the beginning, inclusion and protection of mental health within financial protection schemes seems to be an obvious requirement.
A few more slides before I conclude. So if we are talking about UHC, Universal Health Coverage and how to move towards it, what is it going to cost? And so several studies in the past have attempted to estimate this. This is from the Disease Control Priorities Three project, which had a whole book dedicated to mental, neurologic, and substance use disorders. And it is basically showing that it is very cheap, particularly in lower income regions and countries, to scale up a package of evidence based mental healthcare for priority conditions.
So it is affordable. It may be a huge increase over what is currently being spent, but in absolute terms, and in comparable terms to other major public health challenges and threats, it is not.
And one thing I wanted to quickly show here was that there is this kind of extended cost effectiveness analysis approach, which beyond just looking at the health benefits, attempts to also quantify the financial protection effects of policies, including public financing.
So this is an example of the application of that to mental health care in India, also concluded in that DCP3 volume. And basically, it is helping to add an argument, which is that by changing the financing structure away from households towards publicly financed services, not only do you of course avoid private spending by affected households, but you also can see that the monetized value of that insurance is very substantial, particularly for lower income quintiles.
So let me finish with this one, which Leo requested we come up with some research questions stemming in this area of financial protection. Brainstorming is always done best in a group, but I had a quick think, and I came up with a few here which you might want to add to.
Some which came to my mind was that in the equity space what is the relationship between socioeconomic status and mental health status, a very basic question, and what is the relationship between SES and mental health service access, do we have enough evidence about that. I don’t think so.
Health determinants. So we know about programs like cash transfer programs. There is a small amount of available evidence already there. But can we have a better sense of what are the mental health impacts or consequences of poverty alleviation programs more generally, and other actions to address the social determinants of mental health?
The economic burden as I was showing an example of a study, so what is the actual economic welfare lost to households, and what are the consequences, not just in out-of-pocket spending, but other aspects of their consumption?
And finally, financial protection. So to what extent are affected houses actually protected, financially and socially, and what eminent mental health and substance use conditions and services are explicitly covered in publicly financed health protection schemes? So I leave it there. And thank you for your attention. And I’ll now turn it over to the next speaker, which I understand is a colleague from Rand.
ALEX DOPP: Thanks everyone. I’m here with Marylou Gilbert from the Rand corporation. My name is Alex Dopp, and we will be talking with you today about the fiscal mapping process. This is a strategy that we’ve been developing to improve capacity to financially sustain evidence-based treatments in youth mental health service agencies, and a lot of the issues that we were hearing about from Dr. Hodgkin and Dr. Chisholm already this morning related to the complexity of payment and the equity issues that come up there were really big motivations behind why we did this project in the first place.
And this work involves an amazing interdisciplinary team, you can see their names here. We are funded by NIMH and have no competing interests to disclose.
So just a brief outline of what we’ll talk about with you today. We will go over some background info. I will give you an overview of the fiscal mapping process in terms of the steps and conceptual foundations and tell you a bit about the pilot-testing project that we have been working on for the past couple of years.
And then I will turn it over to Mary-Lou, and she will talk about some of our findings from surveys and focus groups and our experience as trainers and coaches, which both Mary-Lou and I did, we were trainers and coaches for the fiscal mapping model. And we will wrap it up with some next steps.
So, the aims for this, this is an R21 pilot project. The first aim was to develop the fiscal mapping process, and the second aim is to evaluate its preliminary impact. This is a two-year grant, and we’re talking about something that we really wanted to have an influence on sustainment. SO that is why we used the word preliminary there, because it is challenging to look at sustainment in a short time period.
But really, even though these are two separate aims, they are interconnected, because we were building the plane as we were flying it, we were developing and with our partners at the same time evaluating, learning what was working and what wasn’t working, and making changes as we went along.
But really what you need to know here is that the fiscal mapping tool is a multi-step structured Excel-based tool that guides youth mental health services agencies and their partners in selecting the optimal combination of financing strategies for their sustainment efforts. And we focused on two well-established evidence-based treatments, parent-child interaction therapy for disruptive behavior problems, and trauma focused cognitive behavioral therapy for traumatic stress.
So, to say more about the actual steps involved, you can see a graphic here that is sort of orienting you to the five steps that are embedded in the fiscal mapping tool. And these are the steps an agency would follow in their strategic planning efforts.
So, to start, we work with agencies to help them identify what resources they have needed. So you’re wanting to sustain TFCBT or PCIT, what is actually involved in implementing and sustaining this model. And then the second step is to say, once you kind of have this list of all of the resources that you need, what are your objectives. This is where you really start to get strategic with the tool, what are the biggest priorities that you need to focus on to make sustainment a reality.
Once they’ve defined their objectives then what we do is we work with a menu of financing strategies to go through with the service agency, what are all your options for meeting these objections, what are all the different ways that you can bring money in the door to support these objectives. And the first presentation today really gave a nice example of all of the different financing strategies that an agency might be considering.
Step four is the actual fiscal map piece. Because the funding from different sources may cover some or all of your different objectives, so the map is the way to put all of the previous steps together and see can we actually cover all of our objectives with a mix of the financing strategies that are currently available or new ones that we can pull in, and have we reached the point where we can support this evidence based treatment sustainably.
And then step five, monitor and sustain is there because we know that sustainability is a dynamic process, needs, and the funding landscape can all shift over time. And so agencies need to be regularly reviewing, even if they have a sustainable plan, to figure out what are next steps we need to do to keep things in place, or to move with the needs as they shift, so that ultimately you can have a sustainable plan that grows with the agency and with the broader context that it is operating in.
And we did a consensus process with the stakeholder participants, and pretty quickly reached agreement that these five steps made sense as the critical steps to underlie the tool.
Another piece that underlies the tool is the Public Health Sustainability Framework. So this is a conceptual grounding that we used, and we found it really valuable. This outlines important capacities that agencies need to maintain evidence-based programs. And you’ll see that strategic planning is in the center here. That’s because strategic planning really links all of the other capacities together, you need to prioritize how are we going to invest in these different areas of capacity, and how are we going to build them up.
So we thought strategic planning is going to be really important, but also funding stability. And so that’s probably the one that strikes you as most obviously relevant to fiscal mapping.
So of course, if we’re looking at financial sustainability, this concept of having long-term plans based on a stable funding environment is really important to financial sustainability but linking it back to the strategic planning is really important.
Because of course funding stability matters in terms of money, but it also matters in terms of your staff and your partnerships and everything else that you have going on for why that evidence-based treatment is important as a priority in the first place. So we don’t want to talk about the funding piece in a vacuum, and having strategic planning at the heart of fiscal mapping was really a useful way for us to do that.
So here is just an overview on the next slide of the design of this pilot study. And we had 48 participants from different stakeholder groups. So this included representatives from the mental health service agencies that pilot tested fiscal mapping, but also intermediaries who serve as trainers for TF-CBT and or PCIT, as well as funding agency representatives that the service agencies nominated as people they had worked with to secure funding for those treatment models.
And I should say that there were ten service agencies that participated in pilot testing. And you can see that they enrolled in the project, and then they had an initial training, followed by up to 12 months of monthly coaching that was either with me or Mary Lou to support their use of the tool.
And they also participated in a series of data collection activity. So pretty much once a quarter they were doing something. And I’ll tell you more about those data collection activities on the next slide. And then I’ll turn it over to Mary-Lou after that.
So the first thing I will talk about is surveys. So they did these in the first and third quarters of pilot testing, and this was an opportunity for them to give us structured feedback on the fiscal mapping process and training and coaching activities, and also like I mentioned in step three we have a menu of financing strategies. There are 23 financing strategies listed there. And so we also use the surveys to collect information about how the agencies were perceiving and using the information from that list, and we incorporated that.
For focus groups, these were opportunities to discuss their experiences, and they also provided structured ratings of sustainment capacity and intentions. And then two other pieces that I’ll talk about quickly because I see I’m going a little too slow. We won’t get into these much today, but we did incorporate document review to get records like actual copies of the fiscal map for things like grant proposals. We are going to use those for case study analyses later on.
And we did a process mapping exercise to help us sort of solidify what we thought our next steps would be with the tool. Again, not something that we really have time to get into today. So let me turn it over to Mary-Lou to talk about our experiences with the surveys, focus groups, and coaching and training.
MARYLOU GILBERT: Thanks Alex. So, I am going to just provide a brief review of the feedback that we received for each step from the qualitative survey and focus groups that we conducted, as Alex mentioned. So just a little deeper dive here. We analyzed these data using rapid analysis methods to identify major things that could immediately inform improvement of the fiscal mapping process and then guide the tool in its actual use as it was being practiced.
So each step’s summary will identify a key facilitator and barrier of the fiscal mapping process, and then an example of an update that was made based on the feedback that we received. So in step one you can see that participants noted it was comprehensive and inclusive of different perspectives throughout the service agency. And particularly levels of familiarity. However, it might be missing some important information. So as a result of that we provided more resource categories, and a checklist that included things such as overhead expenses, community outreach, evaluation.
For step two, while documentation is important, agencies aren’t always fully engaged with this step. So what we did to update it is included criteria for smart objectives, and just brief descriptions also of PCI and TF-CBT to help engage the various stakeholders, folks within agencies who are not necessarily familiar with EBTs.
Step three identifies gaps for internal and external stakeholders. But users commented that they need more guidance on the applicability of the strategy. So what we decided to do is enhance the financing strategies resource tab, and provided some summaries to help inform the strategy selection.
Step four is seen as actually the most useful for sharing with others, either externally or even within the agencies. As Alex mentioned that is the actual tool itself. The amount of data was seen as overwhelming. And also perhaps pointed out deficit. So what we did was we reframed the information to enhance what the agency is doing well in terms of assets.
And then for step five, although tracking in general is considered a good reminder step, in order to make this more relevant to agencies we added space to document action items to be completed between updates, and we shortened the recommended time for periods of updates based on the feedback as well.
It is important to note that we also asked participants as Alex mentioned whether there were other critical steps that were missing, or in lieu of these five steps for EBT financial sustainment planning. But none were identified, and in fact the qualitative data overall established a consensus for these five steps.
The rapid analysis of open-ended survey items and the focus group notes also led to many more updates, more than I’ve mentioned here. So more guidance and structure around financing strategy selection. We included additional resources, just in order to engage a broader group of stakeholders. And then more space to document action items.
We also received important information or feedback on key considerations before even starting the selection process, such as building the correct team and getting clear information on who to involve and what their roles would be.
This slide is related to organizational strengths and constraints, so the big picture applicability of the fiscal mapping process. And focus group analysis revealed also how stakeholders viewed the fiscal mapping process as improving certain sustainment capacities, such as strategic planning and financial stability. And as is evidenced by a couple of quotes we wanted to highlight.
It was interesting that folks mentioned it brought about intentionality, that it facilitated conversations and discussions amongst staff, that it made the effort productive, impactful, and that there was more accountability.
Nevertheless, there were constraints as well by other capacities. More importantly I think is the organizational capacity for the EBP and environmental support. It is important to consider and weigh organizational hierarchy, size, scope, buy-ins, and champions, both internal and external. And that cannot be understated, the importance of the organizational capacity before even beginning the process.
Interestingly, some participants even felt that there was applicability beyond the current program that was part of the study, and that overall, they felt the fiscal mapping process could be adaptive to some of these agency constraints, as long as they were identified early on in the planning stage.
So in addition to analyzing the survey and focus group feedback, we also analyzed notes from the training as well as the first coaching session for each agency. And the training feedback largely captured agencies’ representatives’ views on the fiscal mapping tool, rather than feedback on the training itself.
In general folks appreciated the visual layout and the features of the tool, and that it captured important information that could easily be shared. They appreciated particularly that it was interactive. However, as I mentioned before, it felt overwhelming for some. And users felt they could perhaps use more guidance in using it.
So, for example, some of the suggestions were clarification of the steps, and just helping make them more relevant and more relatable to the agency by tailoring the coaching sessions, or inviting others to attend as needed, and sending reminders to folks to stay on track with the tasks and the actionable next steps, even so far as suggesting contouring internal agency meetings to work on the fiscal mapping process.
Overall, the training was seen as a critical component of the fiscal mapping process, as this quote suggests. There was even some suggestion for additional training in different formats, such as web-based or instructional videos.
Coaching was also considered a necessary component, and some suggested continued coaching in the form of a consultation process, as noted in this quote. There were also challenges that the coaches faced. As Alex mentioned it was a 12-month study, about 50 percent of our agencies remained with us to the end.
And of course, not to underestimate the importance of organizational capacity. Meetings were often moved or rescheduled, and in fact sometimes we only ended up doing email check-ins. So things like staff turnover and understaffing are really important considerations.
So, what did we learn? The process of collecting user feedback through multiple venues, along with pilot testing, led to numerous beneficial updates to the fiscal mapping process, and helped us identify the best use cases where dedicating time to the tool would be beneficial for a service agency.
So we didn’t really have time to discuss fully the fiscal mapping process tool in detail today, but one key takeaway was that there was a broad agreement that integrating the fiscal mapping tool into existing EBT training initiatives is preferable, than introducing it separately or later, as we did in this project.
We always anticipated that there would be a need to adapt and fine-tune the fiscal mapping process, and even our third version has new features that will require additional testing.
A couple of other items. We created a completion checklist or a fidelity checklist, which we introduced in coaching calls, that was widely accepted and felt very useful to the service agency reps. So that will be incorporated into the final version. And we added a budgeting tool to assist with calculations, also based on feedback that was requesting that.
Now that we’ve wrapped up data collection and our coaching, we are combining all of our data sources. They are pretty rich within this comparative case study approach that we’re using. And looking towards dissemination of the tool and the findings. Of course, we will want to extend this pilot study with a larger evaluation of fiscal mapping process as well.
And I hope I didn’t go too quickly. But as often happens in these presentations there are a lot of details that we weren’t able to cover today, and in fact if you are interested in getting more familiar with the tool itself, please reach out to us, I’m happy to answer questions as well. And you can contact us either via email, and Alex has a Twitter handle that you can also reach him at. And so thank you for your time, I appreciate it. And Sarah, I believe, it is back to you.
SARAH DUFFY: It sure is. I am going to invite everybody from the panel to come back on, because we have a few questions from our attendees. And I’m just going to go in order. We have just a few minutes left, so I want to go a little fast. So the first one was, the move towards mapping social determinants in this recent call is very welcome.
However, and I imagine they’re talking about a call from NIMH, but I wanted to gather your thoughts on health economics approach within the paradigms and methods used in development economics and political economy. And I think Leo answered the direct question in the chat, but in terms of NIH, whether it is in alignment with NIH. And I would just say that we really need to think about the project itself. But Dan, can you elaborate on that? Do you have any thoughts on that?
DAN CHRISHOLM: I don’t think it’s a question for me so much as for NIH.
SARAH DUFFY: I just wanted to know if you had any thoughts about integrating health economics in the paradigms and methods of developmental development, economics or political economy. But in any event – Leo, I don’t know if you wanted to chime in.
LEONARDO CUBILLOS: I agree with you, it depends on the question. And the Center for Global Mental Health Research encourages the use of both the sciences that are needed to answer complex questions related to the areas that we presented as of our interest, whether health economics or not, whether specifically a subdiscipline within health economics. Again, it really depends on the question.
SARAH DUFFY: Now I am going to go to a question for Dan. Dr. Chisholm, did the survey include questions about disability insurance and disability policies, given that severe mental health and drug use disorders have long-term disability impact? Which actually leads me to a question I had. But go ahead and answer that, and then I will ask my question.
DAN CHISHOLM: Thank you. The questionnaire that was used for this household survey, it asked about, not specifically about disability allowanced or payments, but for any form of financial or inclined income or sources, transfers in if you like, either from government, from community resources, or indeed from family members. So it was a kind of income expenditure, family-based survey.
I’ll take the opportunity to say what we used was based on the SAGE Study on Global Aging and Adult Health. So this is a study that was funded by NIMH through the National Institute of Aging, so there we go. So the NIH supported survey made its way into this Emerald study and survey.
But I think Agnes is raising a point here about disability payments, and how that can actually be a kind of form of discrimination, particularly if someone is diagnosed with a chronic condition at a young age, and then is kind of pigeonholed as disabled for the rest of their life, and therefore cutting off opportunities for employment, marriage maybe depending on the society, and so on. So I think that it is an important point.
And the holistic response of course is that there should be appropriate financial and social protection available for people at the point of need, and sadly that the situation is nothing like that in most low- and middle-income countries.
SARAH DUFFY: Thank you for answering that. And I think it is important to understand that slide, because I think a lot of people would be surprised to think that the United States is not a high-income country, because a lot of people would also say that the government really doesn’t fund 100 percent of any of this stuff. Important to qualify that that’s not really what the survey was after.
And so I think at this point, we do have a few more questions, but I think we want to try to keep on time. So I just want to really thank all of the presenters, and again feel free to reach out with Q&A, but I am going to turn it over to Jennifer for our first plenary. Thank you.
JENNIFER HUMENSKY: Thanks Sarah. I am pleased to introduce Dr. Darrel Gaskin. He is the William C and Nancy F Richardson Professor at Johns Hopkins University, and he will be discussing the economics of health equity. So I will turn it over to Dr. Gaskin.
Agenda Item: Plenary A: The Economics of Health Equity
DARRELL GASKIN: Thank you very much for the introduction. So I’m going to talk about the economics of health equity. So, I have no conflict of interest to report. And so I want to talk about the difference between health equity and health disparities. About how economics can inform health inequities. Highlight some conceptual frameworks. And then talk about some examples of research in health equity.
So these are two sort of well-known definitions of health equity and health disparities. Health disparities work is in a real sense what we’re doing there is we’re trying to look at differences between advantaged population versus a disadvantaged population, and those differences that are associated with socioeconomic or environmental disadvantage that is creating the difference.
Whereas health equity is more this notion that people should have the right and ability to reach their highest possible health standard, and that again these factors such as poverty, discrimination, environmental disadvantage, those factors should not hinder them in reaching that particular standard.
So how does an economist perhaps think about this? Well, health economics is about studying sort of the complexity of the health production function, demand and supply of healthcare, and the allocation of how those markets work, the organization of the healthcare marketplace.
We know that the market for healthcare is not like any other market, in that it has a lot of idiosyncrasies that create market failures, or at least complexity in the market. And then we also know that the demand for health in a real sense is a demand that is not only influenced by the consumption of healthcare, but also by the consumption of other goods.
And so we typically like to assume people act in their best interest based on the information that they have and the choice architecture, the menu of which they are able to choose from. And so I think the fundamental question from an economist point of view is how these sort of underlying factors influence health outcomes, and potentially create health inequity because of the decisions of patients, providers, organizations, governments, and all the other actors are playing in. And what we hope to do is inform those decisions so that we can improve and actually promote health equity as opposed to health inequity.
I was sort of thinking about this several weeks ago. I said, my goodness, the premise of my entire center, I direct the Center for Health Disparities Solutions at Hopkins, and I like to say that we are the place-based disparity center, meaning we are interested in how contextual factors influence health outcomes and healthcare utilization.
And then I sort of thought about this movie that featured Eddie Murphy and Dan Akroyd, of particular interest to me because Eddy Murphy grew up in my neighborhood, and we’re about the same age. And this movie Trading Places, the thesis of this movie is sort of the thesis of our work, in the sense that there are two wealthy brothers, the Duke brothers, who bet whether it’s nurture or nature that determines one’s outcomes, one’s life outcomes.
And they bet each other a dollar, and they have two persons that they’re going to run this experiment on, this Louis Winthorpe III who is running their firm, their investment firm, and this Billy Ray Valentine, who is a beggar in the street.
And what they do is they in fact do some things to make sure that they trade places, and that they bring Billy Ray into the firm, and then they plant drugs on Louis Winthorpe III and sic the police on him and ruin his reputation and his life, fire him from his job, disinvest him of his home, and they then watch what happens, and Billy Ray becomes a successful investor, and Louis Winthorpe III becomes a criminal.
And they then conclude that wow, it wasn’t nature, it was in fact nurture. And then they also then decide well but we really can’t let Billy Ray run the firm because he’s black, and we don’t want that, and so they then planned to somehow get rid of Billy Ray.
And it’s sort of like this whole thesis of this movie that I guess is probably in the early ‘80s, seems to be an underlying thesis about how the context in which people live in, and then there are people who can in some places have power to control that context, and how that influences life outcomes.
This right here is the NIMHD research framework. And oftentimes we, when we focus on differences by race and ethnicity, we are sometimes fixated on biology and behavior. Biology and behavior. But I describe that it’s really these other boxes, the physical built environment, that sociocultural environment, and the healthcare system, that really influences peoples’ outcomes.
And I often say that it is easy to tell doctors to do better and to tell patients to do better. It is harder to tell the mayor to do better, the business community to do better, the governor to do better, the Congress to do better, the people who essentially set up or are responsible for setting up the contextual environment in which people try to live their lives, maintain their health, seek healthcare.
Oftentimes Medicaid gets branded as a bad program, but in fact Medicaid is in some ways designed to do poorly in the marketplace. It is really astonishing that we haven’t figured out that if you put poor people in the marketplace and you give them half the purchasing power of other people in the marketplace, that they’re not going to be able to buy services.
It would be just like if we took the food stamp program and we decided that we were only going to pay 50 cents on the dollar for a loaf of bread, and then asked people to find food. It is designed in a real sense to fail. And that’s not the fault of the people who use the programs, that’s the fault of the people who designed the program.
And so this is a chart from unequal treatment that tries to explain healthcare disparities. And it divides it up into these three components, in which clinical appropriateness, patient needs, discrimination, biases on the part of providers. But in the middle there is that operation of the healthcare system, the regulatory environment, the climate. It is those factors which I really think sort of drives some of the health inequities that we see.
So this is a chart that is designed to sort of show the difference between health inequity and health disparity. And so health equity is about reaching a target, an idealized target. And here the target for congested heart, death rates for congestive heart disease, in the Healthy Peoples 2030, is 71.1 deaths per 100,000. And so that’s the target rate. If we were trying to reach health equity what we would be trying to do is trying to reduce the rates of blacks, the black population, the American Indian and Alaska Native population down to target rate.
If we’re trying to address health disparities, what we would be trying to do is to essentially lower the black rate to the white rate. Which unfortunately also is above the target rate. So I have come to the conclusion that a health equity goal is a much more superior goal than a health disparity goal, because the health disparity goal in some cases would essentially encourage us to address a health condition but not address it sufficiently to really allow that population to achieve health equity and wellness.
Now, one of the things that I think is a challenge in our data is that oftentimes we observe disparities and utilization disparities in health outcomes. And disparities meaning something between an advantaged population and a disadvantaged population.
So in this chart right here, the gold here is the advantaged population, and the blue is the disadvantaged population. And so we typically try to think that the disparities, what we observe in the data is what we see in that first cluster, that the advantaged population has achieved optimal health, and the disadvantaged population hasn’t, and therefore we need to try to address the disparity.
But maybe it’s not that. And maybe it’s actually the second cluster, which is that both populations, if we’re sort of thinking about utilization of services, maybe both populations are over-using something, and the advantaged population is using it more.
Or maybe, if we don’t see a difference, they’re both under the target. Or perhaps we do see a difference, but in one case one group is over the target, and one group is under the target.
And then finally sometimes we see where the disadvantaged group actually is performing better than our so-called advantaged group. And then that causes us to question our data and the validity of our data, because in fact sometimes we come into these studies thinking that the advantaged group is the advantaged group.
So I think trying to develop these sort of absolute targets looking at these societal goals are better than looking at just differences between populations, and declaring one population advantaged, because in one it causes us to think about what is the real societal goal.
So if you think about vaccinations, what we want to do is we want to get vaccination rates up such that we get herd immunity, not necessarily so that we can match populations. (Audio drops) minority pathologies concept, and that somehow there are some broken people out there that need fixing, as opposed to broken systems that need fixing.
And then it also allows us to appreciate resiliency in the so-called disadvantaged population, by looking at the heterogeneity within the population. And then finally it gets us away from this zero-sum view of the world, in which there are winners and losers, and that somehow if we take care of the disadvantaged population, we’re taking something away from the advantaged population, because the goal is not to achieve the advantaged population, the goal is to achieve the health equity goal.
I just want to, in just a few more minutes, just talk about some things which I think are important in doing this kind of research. I think it is certainly important to understand from whence we came, in the sense that these reports, the Heckler Report and unequal Treatment are sort of must-read for anyone that is going to embark on doing health equity work and health disparities work in the sense of trying to understand how we think about these things.
But I also would encourage you to read a little history and some sociology, because if context matters you’ve got to understand how that context has been created. And so there are several books, I will just highlight the Ira Katznelson book, it talks about when affirmative action was white.
In a real sense it talks about why our safety net programs are not floors but nets and have essentially holes in them which people fall through routinely. And those holes are by design, and they are designed in a real sense to either leave some people out or to advantage other people. And some people would say as we were thinking about enacting the Affordable Care Act, some people deliberately lobbied for leaving out immigrants in that landmark program. And so it becomes a place in which people are disadvantaged structurally by these things.
This is a reader, I probably should encourage my colleague Tom LaVeist to update this reader, which has a number of seminal pieces in it regarding health disparities and health equity among a number of factors.
The work that I do primarily uses I think a David Williams and Chiquita Collins piece on the fundamental cause of racial disparities in health, which lays it at the feet of residential segregation, and Kamara Jones’s work on racism as sort of guiding principles.
Because in a real sense where you live has a lot of impact on your educational opportunities, your employment opportunities, the quality of your neighborhood housing, and which then can influence your health behavior, your healthcare utilization, and also how safe public safety, the nature of your community. And that these things, residential segregation is uniquely related to race of, and particular institutional racism.
I usually like to show these maps, having to just show that we don’t live in the same places. This is a map of Atlanta. The blue areas are places where there are 20 black persons to one white person. The orange areas are places where there are 20 white persons to one black person. The yellow areas are places where there are equal numbers of blacks and whites.
And what you can see is that there is a lot of blue in the center, there is a lot of orange and brown and not a lot of yellow. And so therefore once you sort of distribute people in this way then it is easy for policy makers who do not have the best interest of certain interests in mind to then do things that would harm them in a sense of depriving them of amenities that would improve their health, as well as expose them to harms that would harm their health.
This is Hispanics and whites in Chicago. Again, we see a lot of blue, a lot of orange, not a lot of yellow. And this is Asians and whites in Los Angeles. Again, a lot of blue, a lot of orange, not necessarily a lot of yellow. And so then place-based policies then have racial and ethnic impacts on people.
At my center one of our signature projects is this exploring health disparities in integrated communities. We have this paper that’s called Place, Not Race, we found these communities that were predominately integrated and economically balanced between both blacks and whites.
And one of the things we found out in that is if you compare this community to communities nationally, or disparities nationally, you find that the disparities are really narrow, especially the health disparities, and that the whites, this community is a poor community, so the whites in this community are doing as poor as the blacks in the community. And so the argument here is that it is really the context, not the biology if you will, but it is the behavioral mechanism that is really the context.
This paper on hospitals’ quality, this paper just basically shows what is reaffirmed in this paper by Asch and Werner is that where you go to the hospital in a real sense determines a lot of your healthcare outcomes. So throughout the COVID crisis we were telling people it’s about these health conditions. If you underline health conditions that are creating these disparities.
And then Asch and Werner do the study, and they control for these underlying conditions in patients, and then they put it as hospital fixed effects, and all of a sudden the disparity goes away.
And unfortunately, we don’t learn the lessons from this, in the sense that we continue to then go back and say patients do better, physicians do better, but we don’t address the system. And so who is making sure that hospitals are appropriately equipped? Why do we allow ourselves, allow our healthcare system to be such that there is such a disparity in the quality of facilities, and just allow it to persist?
So I am going to stop here in the interest of time, and turn it back over to Jennifer for some questions.
JENNIFER HUMENSKY: Thank you so much Darrell. Really thank you for coming over and coming to join this webinar. We wanted to invite you as a plenary speaker because I think that the information that you have given, and I’ve been able to hear your talk previously, it is so informative to and inspiring to mental health researchers and substance use disorder researchers and raises some questions that people can go back and think about.
I especially appreciated your comment about you can tell patients and providers to do better, but how do you address the overarching system, and how do you address the policy makers. And so thinking about how to address those social determinants of health is something that is so incredibly important.
To turn it over to some questions, the first question was whether some of the resources, the list of resources that you showed could be made available. We can put up during the break we can maybe put together a list of the resources to make available. Of course the presentation will be available on the YouTube page.
So turning it over to our next question, it seems that the research community is still lagging in meaningfully engaging consumers and communities in the research process. Can you give an overview of how or whether this is shifting, and are there any bright spots?
DARRELL GASKIN: Well, I think so. Because one of the things in which we have learned is that if you really want to try to address the social determinants of health, you’ve got to talk to people on the ground.
And so our efforts in our research to sort of do community based participatory research design projects that bring community voices into the process of asking, developing what the questions should be and how to address the question, I think is a good step in the right direction.
And the other thing is that oftentimes the solutions to the problems, the people know what the solutions to the problems are in some cases. They know how to, because they’ve been dealing with them, they’ve been living with them. And so it is important that if we’re going to try to address and help people reach some of that health equity target, to achieve their best health, that we engage them and involve them in developing what interventions that work.
And if we do that then that also builds on your sustainability, because then those are the persons who are going to insist to the mayor, insist to the governor, insist to their state legislature and insist to their business community that these things remain in place.
JENNIFER HUMENSKY: We had two questions that came in that are somewhat similar, so I am going to try to combine them and hope I do both of them justice. One of the questions is, we know low and middle-income countries, which are the focus of our Center for Global Mental Health Research, also have substantial income inequality. And are there any lessons that come to mind from your work or other work in the US that might be relevant to low- and middle-income countries?
DARRELL GASKIN: Well, I would first say that I always like to flip that around, in the sense that there are lots of low- and middle-income countries who are doing better with their disadvantaged populations than we are, meaning that they have fewer resources and they’re doing better.
And so I was in a meeting with some Caribbean healthcare finance ministers, and started to talk about our safety net, and they said well the problem is that you have a safety net, we have a floor, in the sense that we don’t go through this business of trying to figure out who should be in and who should be out.
But I think there are some lessons in terms of trying to make sure that there is no solution for us without us, in the sense that we don’t parachute into a community and then bring a solution, you really have to work with people on the ground and understand their context, their history, in order to improve their health outcomes, so that whatever it is that you’re trying to design actually fits in that community.
But I usually like to turn it around, because I think there are a lot more lessons to learn how to do things there in scarcity. Scarcity sometimes seems to be the mother of invention, I think somebody said that.
JENNIFER HUMENSKY: One more question. The question is is your center undertaking specific work around health inequity for First Nations groups, indigenous peoples particularly in terms of place.
DARRELL GASKIN: We have a researcher in my center, her name is Theresa Brockie. She is actually based in the School of Nursing. And she does work around suicide prevention in Native communities and has interventions that are designed to improve their health outcomes and their outlook on life. So we do have some work in that area.
JENNIFER HUMENSKY: Thank you so much. We will now break for a short break, reconvening at 1:30. Thank you all, we look forward to seeing you at 1:30 for our next panel on behavioral economics.
LEONARDO CUBILLOS: Good afternoon everyone. Thank you for joining back after this 30-minute break. It is my pleasure to introduce this second panel of the Workshop on Health Economics. This panel will be on behavioral economics.
And we are delighted to invite Marisa Domino, working at Arizona State University, Kate Orkin working at University of Oxford, and Kevin Volpp working at the University of Pennsylvania. Without further ado I would like to invite Marisa Domino. Take it away Marisa. And thanks for being with us so late.
PANEL #2 – BEHAVIORAL ECONOMICS
MARISA DOMINO: Thank you very much. I am delighted to talk about a study that we have been working on that was sponsored by NIDA. The talk is entitled Can Providers be Nudged into Expanding Their Skill Set? And I am going to talk about a state-wide two-stage experiment that we did in North Carolina. But first I want to acknowledge my collaborators on the project, Sean Sylvie and Sherri Green, both at the University of North Carolina.
So I again want to acknowledge funding from NIDA, from AHRQ, and North Carolina’s Department of Health and Human Services for this study. And as I will talk about in a minute, I want to also thank our fairly large group of ECHO investigators and our project manager.
So you probably don’t need this information in a NIDA sponsored panel, but opioid use disorders currently claim, we’re getting close to 200 American lives per day. And this Epidemic is not just alarming because of the impact it has on so many individuals and their families, but it has also shown that we have some disturbing gaps in treatment areas in a lot of areas.
And we do have evidence-based treatment for opioid use disorder, and they often include a trio of medications referred to as Medications for Opioid Use Disorder, MOUD. The prior name had been Medication Assisted Treatment or MAT. And so you will see both of those used in these slides.
But basically, the three medications are methadone, which is only available through licensed facilities, Dominick touched on this a little bit earlier today. Naltrexone and buprenorphine which Dominick also talked about, and buprenorphine until very recently required a waiver of prescribing authority which included considerable additional time by providers to undertake the required hours of training and to fill out the paperwork to become a waiver provider, and that’s currently in flux. And so these kinds of requirements create severe shortages of providers that are able to prescribe treatments for opioid use disorder.
And our study focuses in North Carolina, which is one of the states with a high opioid overdose rate, and also an increasing overdose rate that was greater than the national average. And similar to I think the results that Dominic highlighted earlier, fewer than seven percent of North Carolina’s primary care providers actually had one of those waivers to be able to provide OBOT treatment or Buprenorphine treatment in their office-based practice.
And so we know that primary care providers in particular have a potential to play a significant role in helping combat the opioid overdose epidemic. And research does show that primary care providers can safely deliver buprenorphine in their practices, and in fact a study by one of my former PhD students, Alex Gertner, indicated that the quality of care provided by primary care providers is equivalent to that provided by behavioral health specialists.
So in order to help address this opioid provider shortage in North Carolina, my colleague Sherri Green established a web-based ECHO clinic, which is a supportive sort of learning collaborative mechanism that is often used to provide additional resources and additional feedback and support to providers learning new areas such as prescribing MOUD.
And so participation, when we first started this ECHO clinic, was initially low. Providers reported in interviews that we had with them that the unfunded time that they spent in these ECHO clinics was a barrier to participation. We also learned about considerable stigma that exists in terms of prioritizing this patient population for their practices.
So we thought about ways that we could jump in and contribute to increasing participation in this learning collaborative. And as economists we thought about monetary incentives, but also in the theme of behavioral economics also wanted to think about pro-social nudges. And so we know that sometimes monetary incentives for something like this can be ineffective or even counterproductive, depending on how those incentives are structured, who is incentivized, and what the purpose is of the incentive.
And for pro-social tasks like increasing access to providers available to provide OBOT kind of care, it can actually be counterproductive. So it can crowd out pro-social motivations that providers may have to participate in this training to become a licensed OBOT.
And so we thought about nudges as well as a way to present choices in a way that appeals to social needs, and that may be a more effective way. How pro-social nudges and monetary incentives interact together was unclear. There has not been a lot of research in this area. Some indicates that they could crowd each other out, but there are other reasons to believe they might be complementary, because they could motivate for example different kinds of providers to participate.
So to date few studies have applied concepts of behavioral economics to encourage greater provision of care, and we knew of no other studies that were in the space of opioid use disorders. And so NIDA funded the ECHO nudges study at the University of North Carolina, and it was our way of sort of testing these different approaches to inviting providers to participate in this learning collaborate.
So what we did for the study is we identified and randomized all, almost 16,000 primary care providers in North Carolina, clustered within their own practices, to one of four recruitment arms. So two of the arms used pro-social nudges, and I will explain what that means in particular for this study, and two of the arms used financial reimbursement messaging to try to increase participation. So that is we created this sort of two-by-two table where some providers receive both types of messaging, prosocial messaging and financial incentives, some received one or the other, and some received none.
So what did that look like? So this is our sort of basic letter of invitation to providers in North Carolina. For providers that were randomized to the prosocial arm we included a message at the top that talked about their specific county. In North Carolina counties are relatively small, saying this now in Arizona where counties can be relatively large. And so these are sort of thought to represent the provider’s local community.
And so we had customized messages for their specific county, and we had some facts and figures that were specific to their county. For providers who received messages relating to reimbursement it just indicates at the bottom that reimbursement may be available for their time, and I’ll explain why we weren’t very committal about that initially.
So this is what the letter looks like, and again fro providers who got both prosocial and financial messaging they got this whole letter, without the circles but with the circled components. If they only got one or the other then the relevant messages appeared or were not part of the letter.
So we sent those out. Among the providers who then responded that they were interested in participating, we randomized them to one of three reimbursement amounts, either nothing, so they received nothing for participating in the sessions, but were certainly welcome to participate in as many sessions as they could. They had an amount that was based on 30 percent of North Carolina Medicaid’s rate for one hour of treatment, and an amount based on 100 percent. So we kind of wanted to test the nonlinearities in treatment-response there.
And providers were blinded to the idea of randomized payments, or they didn’t realize that they were in the second phase study where they had different amounts from one another.
So the 16,000 providers were from just under 8000 practices. And we compared each of the three active recruitment arms to a recruitment as usual arm, so the messaging without all of those extra boxes on there. We use both linear and Logit models, we use county level fixed effects, clustered standard errors by practice. Here I’m going to report the linearity probability models for the most part.
And we also stratified the models by provider type. And so I am also going to show you some geographic analyses that we did where we also ranked counties by the level of the overdose death rate in that county in order to determine whether in fact we saw a differential response depending on the county needs or county overdose death rates.
Some of what I’m going to show you, the first phase was recently published in HSR, and here is the citation for that.
In the second part of the study, we then looked to see how much the financial reimbursement mattered in terms of the number of ECHO clinic sessions, learning collaborative sessions that they participated in, both in terms of counts and hours of participation, because sometimes people didn’t stay for the whole session. And then we looked to see if there were interaction effects between the phase one randomization by the messaging and the phase two by the financial reimbursement amount.
So what did we find? So this is a table of results from our study. First, I will draw your attention to the bottom row here. These are the 16,000 providers, pretty evenly split by the four messaging arms, which are in the columns here. And you can tell very evenly split by the practices as the randomization was setup that way. But a few differences by the number of providers. About half of our sample were physicians, about a third were nurse practitioners, and the remaining 15 percent or so were physician assistants.
In terms of results, we found that the baseline response rate in the people that received the most generic form of the messaging was about 8.67 per thousand. And we found a considerable increase in all of the other messaging arms. So any of the three active treatment arms.
And in particular you can see that the combination of prosocial and compensation yielded the highest response rate for this group. In the right side you can see that there was heterogeneity by provider type. And in the interest of time, I won’t go much further there.
This is a graph that shows what the response rate looks like by the county ranking. And you can just see that the prosocial and compensation arms, which are the blue and light green in the upper part of the graph, did separate from the others, and in fact we see larger differences in counties that had higher overdose death rates.
And then finally just very quickly, this is the effect on hours so in the second phase of the study. And in fact, we do find that higher reimbursement rate it is no surprise to the economists in the room did increase participation in the clinics, but in a nonlinear way. We found only a modest increase from people who got the 100 percent reimbursement over those who got the 30 percent reimbursement. And this graph in the bottom right corner just shows the interaction by the first phase of treatment.
So of course there are limitations. This is one state, and we are limited to provider for whom we had active email addresses. And generally, response rates were low, although this was unsolicited email, so we were delighted that we got as many responses as we did.
So in conclusion we found that both nudges, prosocial nudges, and incentives have strong effects. And that this is a technique that could and should be used to invite providers to participate, because the costs really weren’t that high to generate these messages and customize them per county basis. And we do find complementarity between prosocial and financial incentives. So let me in the interest of time stop there and turn it over to Kate.
KATE ORKIN: So thank you so much. My name is Kate Orkin, I am an economist, and working at the University of Oxford. This is a study that was done with big interdisciplinary teams, so combining economists with researchers who work in global mental health and in psychiatry.
And we set out to answer the question of what effect does treating mental health conditions have on economic outcomes, particularly labor market outcomes, in low- and middle-income countries. So we did a big meta-study, and I’m going to talk a bit about the results for that, and then also the directions that it threw out for future research.
So this was the question that we set out to answer. If you have a mental health disorder and you’re treated for it, how does that affect your economic outcomes if you’re living in a low- and middle-income country? By economic outcomes we mean people’s economic choices in the sort of daily economic lives.
So whether they go to work for example, how long they work, and then the downstream impacts that that has on things like their asset wealth or their income. We also look at things like the extent to which they invest in their children’s education.
So we did a systematic search for all studies in low- and middle-income countries that screened for mental health condition, tested a mental health intervention, and measured an economic outcome. And we look at these until August 2022. We screened a large number of papers, and we get a fairly substantial sample of effects from 39 interventions studying 39 randomized trials.
So I’m going to report on a meta-analysis across these findings. So I will discuss the paper, but then I will also talk a bit about what this taught us about the state of the field in this area, and some directions for future research.
So, why did we do the study in the first place? There was a lot of reason to think that treating mental health disorders could be a highly effective way of changing people’s economic position.
So we know that clinical improvements in mental health disorders are associated with people having improved functioning in their daily lives. Concentration, sleep, management of emotions. And that might plausibly affect their work and investment.
We know that in high income countries treating these disorders reduces sick leave and absenteeism and helps people return to work. And we know that if you take treatments from high income settings into low-income settings, they work well to reduce the symptoms of mental health conditions. But we don’t know about what treating a mental health disorder does to people’s economic choices and outcomes.
So on the one hand if you think about the choice to return to work, perhaps we might have a bigger impact if we treated someone in a low income setting because they would be in informal work, they wouldn’t be getting as much support from their workplace to begin with, and so the mental health treatment may make more difference. But on the other hand it may be really hard to return to work if you are in a situation where there is less of a social safety net, you have less aces to welfare or to employment support. And so that question was what motivated the study.
We look as I said, so we’re in low- and middle-income countries, the population of all screened positive for a mental illness, and they’re over 14. We are going to look at a range of different interventions, those that were psychosocial, used a drug, or combined those two approaches. We don’t include studies that combine any mental health treatment with an economic intervention. So we look only at randomized trials. And then we looked at a pretty broad list of economic outcomes.
We have a quite comprehensive search strategy. So we searched published studies but also trial registries for results and progress, repositories of working papers to get unpublished studies, and then we looked at the references of all the studies that we found. So we did a fairly comprehensive search and we end up with these 39 interventions that we study in the paper.
So these are spread across various different country income levels. So this just gives you the number of interventions of the 39 in these different categories. We also have a pretty good mix across different geographic areas, although more studies are in Asia than in other places.
So today I’m going to focus on interventions. We basically are going to conduct three separate meta-analyses that look at interventions of a particular type that target a particular disorder. So that’s to minimize heterogeneity in the type of treatment and the underlying population. So the main ones that I will look at today are treatments for common mental disorders, depression and anxiety, that are psychosocial, we have ten interventions of that kind.
Treatments for CMD that were combination, so drugs and psychotherapy. Or treatment for severe mental disorders that were a combination. And that again had 10 studies. I won’t talk as much about treatments for PTSD or substance use disorders, we had fewer studies there, so there is more heterogeneity, but we do talk about those in the paper.
So the bottom line we got from the study was that mental health treatments have material and important economic impact. So in the lefthand column we have the effect across all the different kinds of interventions, and then on the righthand column we have the effect looking by these sort of subtypes of for example the orange dots are psychosocial interventions that target common mental disorders.
So we see this is the overall effect across all types of economic outcomes, and that is coming in fairly strongly at about 0.15 standard deviations. We see that those effects occur across a range of different outcomes.
So there are some effects on peoples’ time in work. We positively code the next two outcomes so that an increase is an improvement. So we see an improvement in whether people are able to work. There is no effect on the number of days which they are unable to work. And then effects on the extent to which people are able to function at work on self-reported scales of that measure.
We also see on average some effects on people’s investment in their children’s education. And on asset wealth, although fewer effects on other measures of income and expenditure or of subjective poverty.
And then if we look at disaggregating into these intervention/disorder combinations, we see that psychosocial interventions have in general less strong effects on the work-related outcomes, and they do have some effects on some non-work related outcomes, so particularly education and subjective poverty, that’s all the orange dots there.
But in general, the psychosocial interventions targeting depression and anxiety are having fewer economic effects. The stronger economic effects are coming from combination interventions, so a drug and a therapy of some kind.
And so whether the combination is targeting common mental disorders or more severe disorders like schizophrenia, those are having much more positive effects on a range of work related outcomes. We don’t measure as many non-work-related outcomes for those combination groups, but we see pretty strong effects, around the 0.15, 0.2, even larger, of these common interventions.
So this isn’t driven by the sample of interventions, the psychosocial interventions not being effective in treating mental health conditions. So here we also coded up all of the different measures of mental health symptoms.
I won’t go through these in detail, but the broad message is that the sample of interventions in our study are effective treatments. They are having, as we know for these treatments in other contexts, they are having benefits on mental health symptoms, and that’s true for the psychosocial common mental disorder treatments, as well as the other treatments.
So all of the treatments improve mental health and functioning, but it seems to be that the people who are receiving the combination treatments, which we know are potentially people who face more severe disorders, those are the people whoa re seeing the real economic benefits.
Then we also do an analysis which examines the effect sizes on the sort of mechanism. So effects on functioning and effects on mental health. And so we see here that if an intervention has a bigger effect on these mechanisms it also has a larger effect here, we look at the work-related outcomes, but the pattern is also true for non-work related outcomes.
And so the fact that these effects on the sort of mental health mechanisms and economic outcomes are correlated suggests potentially that the effects of work-related outcomes are occurring through the effects on these underlying mental health mechanisms. This is all in the paper, I won’t go through it, but we do a lot of different methods of analysis, and we find these findings are pretty robust, and the study sample doesn’t suggest that there is much publication bias.
And the thing that is also quite compelling here is that the benefit/cost ratios for these mental health interventions are pretty favorable relative to common economic interventions that try to achieve these same outcomes. So improving employment, reducing poverty.
So we collect costs for 20 interventions in our sample, which had available cost data. You will note here the average is 363 US Dollars, that’s much bigger than what we were seeing when these interventions were delivered at scale, that’s the cost in a trial So the comparison gets even more favorable if you’re looking at treatments at scale.
But we basically look at the benefits in terms of economic outcomes, or the benefits in standard deviations per 100 US Dollars spent. The mental health treatments in achieving changes in these outcomes are an order of magnitude more effective than some other common economic programs. The one that gets the closest is the unconditional cash transfer programs that Dan earlier spoke about.
And so I wanted to talk a little bit about where this leaves us, thinking about what we know about the evidence base, but also the study threw up a lot of places where we thought it was important to build this evidence base further. So we know that if there are effects on the economic outcomes of mental health treatments that can really enhance the case for funding mental health treatment.
But there were a number of places where we still need a strong evidence base than what we were able to present in the study. So I’m going to talk a bit about samples, measurements, thinking about active ingredients, and then tailoring these treatments for poverty reduction.
So the first was on samples, and I think it was an important insight from doing this meta study, is that if we want to study effects of economic outcomes we need more power in our studies of mental health treatments. So most of the individual studies in our sample were intended to study effects on mental health outcomes, so they are powered to study effects on mental health outcomes.
But effects on labor market outcomes or on poverty tend to be smaller than effects on mental health. And so just to give you a back of the envelope calculation, the average of the studies in our sample had 600 people in them, and that would detect an effect of about 0.2 standard deviations, but most effects of employment or poverty interventions are smaller than that.
Most studies in economics in this field would be 1,500 to 2000 people. And so a lot of why we don’t see these benefits in individual studies is an issue of power, and that is something really to think about in funding and constructing these studies.
The second point was thinking about measurement. So we can’t say anything about income, hours worked, physical health, because few studies measure them, and so we can’t do a metanalysis on them. But studies of mental health treatments could easily include a short battery of economic outcomes measured at multiple timepoints, And also potentially more mechanism measures.
And this would really enrich our study of this research question. There are really good standard short economic measures developed by for example the World Bank that account for particular economic conditions in low- and middle-income countries.
For example, many people are in the subsistence economy, and they don’t earn a wage, but we can still measure their production. And so if we could integrate a short module into every mental health trial then this would change the evidence base overnight, and I think is a really high value thing to consider doing.
Also thinking, of course, every academic would say this, but looking at multiple timepoints, there were a few studies that came up in this meta study where we find incredibly persistent effects. Mental health treatments have continued to have effects on people’s economic wellbeing seven or ten years after the intervention has occurred. That’s the most cost-effective thing that exists in development economics. So if that is true across the board, that would be a really important thing to know. But we don’t follow enough studies up over time.
Also thinking about mediation to examine the sort of full causal chain and looking at more mechanism measurement to think about how these effects are occurring. For example, is it through people’s changes in cognitive styles or their affective styles, which weren’t measured in the studies that we looked at.
The last two things, the one was thinking about we don’t really know a lot about which components of treatments are having the economic effects. And so I think unbundling common psychological treatments would have enormous value, linking to the agenda trying to uncover active ingredients in the study of mental health.
So if we think about CBT for job seekers, or maybe it’s the behavioral activation component, and they’re submitting more applications. Maybe it’s the cognitive restructuring component, they’re better at responding to negative feedback. We aren’t sure when we just give a bundled CBT intervention. But knowing this is really important for studying cost effectiveness, and for thinking about whether these treatments could be used for populations without mental health conditions.
And then the final thing, just to wrap up, was that we could think more about how we tailor mental health treatments for poverty reduction as well as alleviating mental health conditions. So Dan and many others have referred to this literature suggesting that poverty can cause poor mental health, and that can lead to poverty traps.
So people are poorer, they face psychological challenges, and then they struggle to benefit fully from economic interventions. But we could also get synergies from combining poverty alleviation and mental health treatments. And we actually find some suggestive evidence consistent with that. So we think this is a really promising avenue for future research.
So in conclusion we did the systematic search of studies that were in poor countries, screened for mental health condition, and tested mental health interventions. And we find that many of these treatments have positive effects on people’s economic outcomes that occur largely through their effects on mental health.
And this suggests a new smart research agenda where we could continue to study these economic effects and make the case for more funding for mental health treatment as a poverty alleviation tool, as well as to improve the quality of life of people suffering with these conditions. And then I think I need to turn over to Kevin.
KEVIN VOLPP: Thank you. I am going to talk to you a bit about some of the work our group has done on behavioral economics and improving health more broadly, really starting from the premise that behavior is the final common pathway when we look at a wide range of public health challenges, whether it’s the 45 percent of patients who after a heart attack take their cardiovascular medicines in a year following a heart attack, to challenges with COVID vaccine, to changing clinician behavior, or substance abuse, mental health medication adherence, lots and lots of opportunities for us to try to improve.
And the starting point for how I have often thought about this is really recognizing that rationality only partly describes behavior change. When we think about the models that used to be used of these stereotypical econs who would take information, calculate the probabilities of all the things that could happen to them, and have a clear sense of the utility or disutility of different outcomes, that would just imply we would educate people on what to do, we could simply use prices to adjust what people would do, using incentives, and we could just layer on increasingly complex interventions without worrying about cognitive bandwidth limitations and other phenomena that we’re now very familiar with.
And instead, you can really think about the mind as being like a high resistance pathway where we’ve developed a series of behavioral reflexes which happen almost automatically, and which imply a very different type of intervention strategy. Instead of using information, we can think about using choice architecture, social norms. Instead of straightforward financial incentives, we can think about ways to make those incentives more behavioral, whether they’re financial or non-financial. And really with everything I think it’s so important to focus on simplicity, given the limited bandwidth that people have.
So I am going to briefly talk about this in three different realms: choice architecture and defaults, financial incentives, and simplifying complicated processes. So probably the biggest policy success of behavioral economics has been around financial savings behavior.
And this is an example of one of those studies, there were very many, which basically show if you switch enrollment in a voluntary contribution retirement plan from an opt-in to an opt-out type mechanism, comparing green versus orange, you see very big effect sizes, and those tend to be very stable over time.
Now, when we think about translating that to a health context, that is actually really challenging, because we can’t just default people into a smoking cessation program and assume that if they do nothing we will magically get lots of people that quit smoking. Obviously, those programs require ongoing engagement of a kind that you don’t need if now your retirement contributions are electronically deposited from your paycheck if you do nothing.
So some of our work has been around trying to increase program enrollment and engagement. And here is an example of how we’ve tested this in the context of people with poorly controlled diabetes, simply by pivoting and shifting the on-ramp from the standard opt-in type approach of asking people if they want to enroll, and if they do telling them where to go to do that, to reframing the invitation and really highlighting that based on clinical criteria we think you’d benefit from being in this program, unless you opt out we’re going to plan to enroll you or put you on a path to enroll.
Now, of course, since we don’t want that to be a pyrrhic victory, we did as part of enrollment here people had to come in, meet with a nurse, learn how to use the remote glucometers, and actually go home and start using them. So given that, we were very pleased to see we were able to triple enrollment rates, hemoglobin A1C improved to a similar degree in both groups.
And we have since replicated this in a variety of different contexts, ranging from colorectal cancer screening to patients post-MI, and getting them enrolled in programs, and it seems pretty robust, that you can roughly triple the enrollment rates by reframing the invitations.
But I do want to highlight that nothing is quite as simple as just changing people’s defaults. And this is an interesting example of some work led by my colleague Kit Delgado who looked at a shift from no default to a default, in terms of opioid prescribing in our emergency room at Penn Medicine.
And what you see here is when you have no default, you have a fair number of people prescribing 11 to 19 pills, these are all for fairly minor conditions, 20 pills. But when you impose a default of 10, there is a shift to the left of this distribution, and so fewer physicians are prescribing 11 to 19 or 20, and you see the number that prescribe 10 shoots way up. The difference between before and after with greater than 20 is not significant, so I want to highlight that is probably a no difference.
But one thing we do see, which I think is important to highlight here, is we see fewer clinicians prescribe less than ten. And that’s why I wanted to share this slide with you. Lots of evidence you can change behavior through defaults, you can change it quite strikingly, but you always want to think about what these unintended consequences might be.
You might ask, well why didn’t you set the default at five? And of course, that might have worked, but if the default is really low then people might think that’s no longer applicable to the clinical circumstances of this patient, so I am going to override it.
Here is another example of how we have tried to think about testing different types of nudges. This was a cluster randomized RCT that was conducted as a pragmatic trial with rigor of consent, both clinicians and patients, roughly 4000 patients, 150 clinicians, 28 primary care practices.
And here what we’re trying to do is increase statin prescribing among patients who clearly meet clinical criteria as being very elevated risk of atherosclerotic cardiovascular disease. And here we tested a clinician nudge, a patient nudge, the two combined, or usual care. And basically, what you see here is that the patient nudge itself was actually not very effective at all. The clinician nudge was much more effective. The two together, there is some added benefit.
So this is another illustration that it is also important to think about do you want to try to influence population health by changing clinician behavior, patient behavior, or both. Great to talk more about that.
And then I also wanted to share this example with you of some work we did with CVS where they were interested in increasing enrollment in an automatic refill program, that seems like an obvious way to increase medication adherence. But when we talk to them about doing this as an opt-out and shifting it from an opt-in, they said we can’t do that because we had to charge everyone’s credit cards, and people would be upset if we haven’t explicitly authorized that.
So we came up with an approach we called enhanced active choice, in essence press one if you prefer to refill your prescriptions by yourself each time, or press two if you’d prefer for us to do it for you automatically. And it’s really just a simple way of highlighting the convenience of this type of program, and you can see that it really doubled the rate at which people enrolled in the program.
So briefly I want to talk a little bit about financial incentives and health behaviors. Lots of evidence of effectiveness in different contexts. Some of the work that we have done is about smoking cessation. This was a program that had a fairly straightforward financial incentive, where we randomized people at either usual care, or usual care plus $750, roughly tripled long-term smoking cessation rates, and led to a benefit design change among all GE employees in the US.
We then followed that up with a study where we were trying to test whether we could capitalize on loss aversion by creating these pre-commitment or deposit contracts where people could put their own money at risk, and we match it roughly four to one. But what you see in this chart here is the reward conditions where you just get the $800 if you meet the milestones for quitting actually was more effective than the deposit contracts, because the deposit contracts, the challenge is that roughly about 14 percent of people opted into actually deposit their own money. Fifty-two percent of those who opted in, actually quit. So that number was very intriguing to CVS and based on that they rolled out a program called 700 Good Reasons To Quit, which had $50 deposits and 14 to one matching. Unfortunately, their data didn’t allow for us to really evaluate this, so I can’t tell you whether the high quit rates among those who participated were followed up here at scale.
Another study I briefly wanted to share was a study we did using time limited incentives to increase antidepressant adherence. And this was a study we did in 120 patients prescribed antidepressants in five of our primary care practices. We randomized them to either usual care and escalating incentive where you started at $2 per day for being adherent, increasing by a dollar a week, to $7 per day by the end. Or deescalating incentives, where you started at $7 and then decreased to $2.
And there are a couple things that I want to point out here. One is that we do see pretty significant increases in terms of adherence. But what is particularly interesting is if we look at depression symptom remission we see 35 percent in the escalating group, 26 percent in the deescalating group, and 8.6 in the control group.
So these are preliminary, this was a pilot study. But it does suggest that in conditions like depression where the disease itself conspires to make it more likely for people to be non-adherent, and where a time limited intervention might really produce dividends because the patient then might feel better after taking the medication for six weeks, it might actually be really important to investigate more fully.
And then finally just a few words on simplicity. A lot of patients really struggle with refilling medications, and if you’re on 10, 12, 15 medications, the process of being organized, if each has a different refill date, is actually a pretty complex logistical task. In one of the studies that we did, we looked at synchronizing prescription refills with Humana, and found for those with low baseline adherence rates we were actually able to increase adherence by about 13 percentage points, which was a pretty big effect.
In general, I think one of the challenges that we all face is that whatever we do, whatever we test is embedded within health plan designs that tend to be quite confusing. So this was an example of a plan that an employer asked us to comment on for them, and there were literally about 70 pages of small print like this, meaning that few people probably read through it, few people really understood what were all the incentives. There was a veritable alphabet soup of different incentives that were being used.
And if you think about something as simple as trying to encourage urgent care versus ER visits, the problem is that if you have a health plan that is using co-insurance, it is actually impossible to calculate ex ante what the cost of that ER visit will be.
And so we worked with Humana to design a new health plan, it’s called Humana Simplicity, that’s copayments only. In studying this it seemed like that was the cost sharing mechanism people really understood best. There are only seven price categories, and the whole description fit on two pages.
So for example, if you are deliberating going to the ER or getting urgent care, you can see here it is $375 versus $100, and your ER copayment is actually waived if you are admitted.
So a few key takeaways here in general that I think are worth highlighting. One is that providing information alone is less likely to change behavior than changing choice environments. And I think it is important to think about this because we often underappreciate that, and even in the rollout of studies we can use choice environment manipulations to increase the likelihood of engagement.
I also do want to highlight though, that changing defaults, while effective, it is also important to think both about intended and unintended consequences, as I illustrated with our opioid prescribing example. Financial incentives can influence behavior, they can be enhanced by various behavioral elements.
I don’t have time to go into that in detail, but a lot of work has gone into trying to figure out how to enhance financial incentive designs, including for example the work done by Nancy Petry on contingency management, which has really mapped out quite fastidiously ways of trying to development incentive designs to make them more effective.
But in all of this it is important to come back to the notion that simplicity really helps make incentives more clear, and without simplicity there is also a real risk that people won’t understand what they are being incented to do. So with that I want to thank our NIA Roybal Center for all their work on these and other studies, and I will end there. Thank you.
LEONARDO CUBILLOS: Thank you. We have a few questions. I’m going to start Kevin with you, given that this asks for clarification. The question is from Emory. He asks for the Harper et al 2015 paper, did you collect some biomarkers of exposure to confirm cessation? I guess smoking cessation. Or did you seem to rely on self-reported data of cessation?
KEVIN VOLPP: Yes. In that study, as in all of our other smoking cessation studies, for anyone who self-reported cessation we would then do cotinine test to confirm whether they in fact had quit smoking. And generally, these was saliva cotinine tests, but in cases of people who were using nicotine replacement therapy we used urine tests, because there’s an assay that can only be run on urine to differentiate the metabolites of nicotine replacement versus from cigarettes.
LEONARDO CUBILLOS: We have two questions for you from Manus Sycamore(ph.) who actually asked one question in the earlier panel as well, so thank you Manus for that. The first one is can you comment on how much these RCTs, especially those done in LMICs, knock on to their routine public health services, and can we generalize these treatments to draw these conclusions, that is the first question.
And then the second question is did you say that combination of treatment were mainly used for severe mental disorders, and those demonstrated greater benefits?
KATE ORKIN: I think the first thing was just to think about how these map onto routine mental health services. So I didn’t show that disaggregation. We can look at who the provider of the treatments is. So we do find that there is a slightly bigger effect for providers who are universities or sort of research institutions than when these are provided as part of usual care. I think that’s quite usual for mental health treatment.
So the efficacy of treatments does go down somewhat but the effects are still positive and we can’t usually reject, we can’t conclude that there is a significant difference between the two types of providers.
So there are some studies in our sample that are at scale as they would be delivered in the normal course of treatment. But I think it is an important point that we see some difference in effects, that the research studies are likely to be an upper bound on the effects that we see.
And thank you so much, your clarification point is really helpful. So we actually find stronger effects for combination treatments both for severe disorders and for common mental disorders. My understanding is that even for common mental disorders, if you get a combination treatment, that can mean that it is people who have a slightly more severe common mental disorder.
But I agree with you that it could be for other reasons that these combination treatments are more effective. But I think that the important thing was that the treatment modality and potentially treating these people with more severe conditions was having really huge effects on economic outcomes, much moreso than any other kinds of intervention. So we think that was really important.
LEONARDO CUBILLOS: Marisa, a question to you. Do you know if other states or other entities are interested in implementing similar initiatives?
MARISA DOMINO: In terms of the ECHO collaborative initiative, there are a large number actually of states and supporting mechanisms that supported implementing an ECHO for MAT. In terms of doing sort of the kinds of studies that we did to try to increase participation in that, I am not aware of any other states doing that kind of approach.
LEONARDO CUBILLOS: Kevin, I have a question for you, if I may. You had a slide that you presented the impact or the outcome of not just to patients who could receive statins, but then to prescribers who provide statins. And you communicate that the combination of both is better than either one of them.
And I got to thinking of, in the context of low- and middle-income countries and decision making in different countries, too. Do you know any studies that study not just on the supply side, not on prescribers, but upstream in the decision-making process of health systems, either health system managers or hospital directors, or any other individual in the decision-making chain?
KEVIN VOLPP: It is a great question. But I actually don’t know of studies that have been done that don’t in some way work through the clinicians. There are a lot of, I think in statins in particular a lot of people believe the benefit to risk ratio is so favorable it would make sense for them to be prescribed to a lot more people, and you can imagine that being done in a variety of ways.
One way in which we are currently testing is to do the prescribing through pharmacists, and the pharmacists have use agreements with clinicians that then allow them to prescribe on their behalf. But even there it was interesting, the dialogue with clinicians, because we had thought about instead having nurse practitioners in that role, and this particular group of practices they were just more comfortable with pharmacists doing it because they had worked with pharmacists in this way for some period of time, and just there was more part of what they’re accustomed to.
But more broadly I think it’s an important question, particularly if we think about for example, cancer screening, or other modalities where you really don’t need a doctor or even a nurse to do the ordering on if the system could have ways to do that more systematically, that would certainly make sense.
LEONARDO CUBILLOS: Kate, one question for you. At some point during your talk you mentioned something along the lines of saying, I am going to paraphrase this, so apologies in advance. Potentially the clinical RCTs could also include measures of economic outcome. Could you expand on that point, could you provide a couple of potential outcomes that could be of interest to capturing those?
KATE ORKIN: I think the few basic concepts that are really important for people’s economic wellbeing would be in poor countries we would tend to measure what’s people’s wealth in terms of the assets that they hold. It’s much harder to measure income, but that’s usually a good measure of their sort of overall poverty level. I think how much people are working, so how many hours did you work in the last week, and then what were your earnings for that.
And then one of the things we really worry about with mental health conditions is the extent to which people make economic investments for the future. So there are a few measures like have you been investing in buying crops, or buying stock for your business, or sending your kids to school, or in preventative health treatment. So those sorts of investments, I think those can also be quite easily measured.
But I’d be very happy to, we actually did as part of the outputs put together a list of those sort of common economic indicators. So I would be happy to share that. We gave some example questionnaires as well. But there are some quite short modules that can be used.
So I think the idea would be a kind of standard measurement in the same way as we’re starting to standardize depression or anxiety scales. I think you could think of a short module like that that could be done across a range of different studies, and that would be very powerful.
LEONARDO CUBILLOS: Very interesting. Thank you Kate. Kevin, to this specific question: Are there entities, payers, state Medicaid programs, and others, that have been interested in implementing economic incentives like the ones you described?
KEVIN VOLPP: Well we found the biggest receptivity is among health plans and employers. And I think a lot of that becomes because in essence any private health insurance plan, any employer in essence has an incentive plan that is already in place, and it is just a question of incentive plan by dint of having a health insurance plan, and it’s just a question of are they intentional about what is being incented or disincented or what’s not.
And historically most of it has related to health service utilization, but increasingly we are seeing a lot of insurers and a lot of employers layer on incentives related to various health behaviors, markers of wellness. A lot of that isn’t particularly evidence based, so I think there is a really good opportunity for the field to really contribute there.
And some of it relates to lowering barriers. So for example barriers to mental health services, where a lot of employers are moving to eliminate cost sharing, or make it minimal. So lots of potential there. But I would say it’s loosely based on evidence.
LEONARDO CUBILLOS: Marisa, there are two questions for you. The first one is have you found resistance to these incentive programs from providers or others. And so how have any concerns been addressed, that is the first question. And then the second one is what do you see as your future directions of this line of research for you and the team?
MARISA DOMINO: Thank you for those questions. In terms of resistance, the biggest resistance obviously of relatively low response rates, and so we were able to round up a number of providers who were interested in participating and receptive to the kinds of invitations that we were sending, but obviously that’s in the minority.
In terms of what we did after the study, we revealed the fact that providers were randomized to incentive amounts, and then we had a set of interviews with them, and they didn’t seem to mind at all having been in that study, and completely understood why that was happening. And so we really didn’t see any resistance to that.
I can’t think of any other forms of resistance that are worth mentioning. I think that providers were just fantastic to work with, and the ones that did respond really were very engaged in the topic.
In terms of next steps for the research, this was obviously one study in one particular setting. We would love to be able to do more of this, possibly in other states. Certainly as was pointed out, the way that providers are able to provide office-based treatments of buprenorphine in particular are rapidly changing, and so a lot of what the echo clinic did was talk about the process of prescribing medications, which hasn’t changed, but the process of getting a DEA waiver to be able to prescribe this particular treatment area, and that has not changed considerably in the recent past.
So I could see still a need for this kind of approach. But we would also like to do similar work in other treatments areas, so maybe either way to nudge providers into providing more treatments for depression, because I think as Kate pointed out that is still the minority of people who are receiving adequate treatment in that area.
LEONARDO CUBILLOS: Marisa, we are receiving another question for you. What do you think is going to be the effect on prescribing buprenorphine, now that the X waiver has been rescinded by the authorities? Particularly in primary healthcare providers.
MARISA DOMINO: I think maybe one doesn’t need too much of a crystal-clear ball to predict that this will increase treatments because a barrier to providing treatments has been reduced.
But I think as also was highlighted in Dominick’s opening statements, we know that lots of providers still choose not to provide treatments in this area, even when in the prior regime they had a waiver to do so. I think there are still considerable supports that providers and practices need to engage in treatment. So I think this doesn’t resolve a lot of those barriers, but I do think that we will see a greater provision of treatments.
LEONARDO CUBILLOS: Thank you so much. On behalf of both the audience and the organizers, thank you so much Kevin, Kate, and Marisa for your comments and presentations, it was wonderful. Personally, I took some notes.
MARISA DOMINO: Thank you.
LEONARDO CUBILLOS: It is now time for our second plenary of the day. And I am delighted to invite Varun Gauri. Varun is a Lecturer of Public and International Affairs at Princeton University, and the cofounder and principal advisor.
Varun was co-director of the Royal Bank’s World Development Report 2015 on Mind, Society, and Behavior, that placed behavioral economics at the center of the conversation of international development back in 2015. So thank you Varun for joining us today.
PLENARY B: Biases and Mental Models in Health Care
VARUN GAURI: Thank you so much Leo. Leo and I had the pleasure of working together on health and human rights at the World Bank, which was a joy. What I want to talk about today is the biases and mental models in healthcare delivery. The point of departure for my remarks are the World Bank’s World Development Report 2015, which as Leo mentioned was focused on mind, society, and behavior.
Our core arguments were that development economics and public policy more generally are ready for a redesigned based in a more realistic understanding of how people think and behave. In the standard approach, economists typically believe that individuals carefully weigh their choices, make decisions individually by and for themselves, and consider all readily available information as it’s presented to them, without any filters coming inside.
What we wanted to show, and there are lots of ways to get at behavioral economics, Kevin talked about resistance pathways. The way we presented it was that people think automatically. That is, they use mental shortcuts.
People think socially, in other words we cooperate with each other, we rely on social networks and social norms, and that we think with mental models. In other words, the information coming in is filtered by the world views and mental models that we have, many of which are drawn from society and that shared sense of history.
One way we think about this is we all have two systems of thinking. There is an elephant, which is always on, sniffing around, unaware of its own actions. It is very strong and powerful but error prone. And then there is a rider, which is who we think we are, which is smart thinking. But this kind of thinking is energy intensive and allocates attention selectively. We often think that we are merely the rider, but in fact the elephant is guiding much of our behavior.
Now, this angle tells us a lot about why some problems in healthcare arise. Why do patients not adhere to medication protocols, not show up for appointments, not follow through on vaccinations, not engage in healthy behaviors, not sign up for health insurance, not get tested for diseases? It’s because the elephant is in a significant way in charge of their lives instead of the rider per se.
And as Kevin, Marisa, Kate mentioned, there are a variety of nudges, but in other behavioral interventions, sometimes going beyond nudges, that can speak to the elephant. These involve soap operas, social norms interventions, reframing, encouragement, a variety of automated reminders. This is what a lot of behavioral economics is about when applied to healthcare, we argue in the report.
But what I want to talk about today is the fact that it’s not just patients who are following the elephant, but all of us are, including elites and clinicians, are also in fact roaming around using what behavioral scientists often call system one, this sort of large, slow, efficient sometimes but error prone system. Studies suggest that not just people out there, but experts, in fact share a lot of these biases. And that is really what I want to address today.
The starting point is a study by Redelmeier and Shafir published in JAMA 1995, which asked family physicians to think about a situation in which a patient who is a candidate for hip surgery had tried all the medications, they hadn’t worked but for one.
The question was would you send this patient directly to the surgery or would you try ibuprofen first. 52 percent said that they would just send the patient directly to surgery. In the second scenario there was one additional medication added, piroxicam in this scenario, 73 percent chose to send the patient directly to surgery.
It’s kind of odd that adding a choice would increase the number of providers choosing surgery. The interpretation was that this is an example of a choice overload. In fact, when you just have one choice you think, eh, maybe let’s try it. But when you have two choices people say it’s too complicated. Clinicians themselves were thinking this way.
This is an example that you might have heard of in the context of say investment decisions, a lot of times when people are trying to make an investment, if there are too many choices they opt to keep money in cash rather than to invest. This applies to experts, to clinicians as well. That’s something we need to keep in mind.
Now, it is also true that data, too much of it, is exhausting to everyone. If there is too much information coming at us, if things are complicated, we just decide to let the elephant lumber along.
My former boss at the World Bank, Paul Romer, the Chief Economist called out our own department for manipulating data for political reasons. It led to a controversy. Paul was now encouraged to step down. This is an example of how the complications involving the interpretation of data can lead to charges of confirmation bias and other kinds of problems on the part of experts.
To investigate this, we conducted a study of World Bank staff and also UK civil servants. On the UK side, Kate’s colleague Stephan Durkin was running the work. In this we looked at a variety of biases that experts have. We started with confirmation bias, which as you probably know is the tendency to seek out and interpret evidence consistent with prior views.
We presented a pair of fake, randomized control trials, and asked experts at the World Bank and UK Government to interpret them. In one setup, the data involved the minimum wage, which is a controversial issue among economists. Another setup, the data involved skin cream, which is usually not all that controversial.
So this is what it looked like. Respondents were asked to evaluate this study. In a certain number of cases people use skin cream, in a certain number of cases they did not. In a certain number of cases the rash got worse, in a certain number of cases the rash got better. So are the data most consistent with the fact that the skin cream works or it doesn’t? And you can see that there is a right answer, and if you set it up as the table below shows you can figure out what it is.
Then we flip the columns, and basically as you can see now the answer is different, and it looked like not using the skin cream actually made outcomes better than using the skin cream.
We then used the same set of data, but talked about the minimum wage. Has the minimum wage raised poverty or decreased poverty? In this case some municipalities implemented minimum wage, in another scenario they didn’t, and again there was a right answer, the same as before. We flipped the columns and saw again there was a right answer just as before.
The main finding, this is the vignettes respondents saw, the main finding was this, that both in the UK Government officials and at the World Bank people were more likely to get the right answer when the data involves skin cream than when it involved the minimum wage.
And you might think wow, this is crazy, a lot of these are economists, they are experts, and they couldn’t quite read the data. And the answer is that what’s going on here is that the skin cream framing was easier to answer precisely because people were economists. They came into it with a certain kind of view of how the world works.
Errors were in fact correlated with ideology. People were asked a question from the World Value Survey about whether they in fact support wage equality because it’s good or support a limited wage inequality because it’s a motivation to effort. And notably people were more likely to get the right answer when the framing was consistent with their prior views.
We also, in this series of studies about developing experts, asked about responses in the context of the Ebola crisis, asked a series of questions about risk taking. As you will recall, the crisis had a significant impact on West African countries, dramatic declines in foreign investment. Some of the respondents were world bank country staff, country economists who were providing information about the risk of going into these countries at any given time. They wanted to see how they were thinking about risk.
And to do that we used a classic experiment from Daniel Kahneman and Tversky, which they called Asian Disease, because of something like this, you may have heard of this, your country is preparing for new disease expected to infect 12,000 people, there are treatments one and two.
In treatment one 4000 people will be saved, in treatment two there is one third probability that 12,000 people would be saved, and two thirds probability that no-one will be saved. The expected outcomes are the same, but one is a safe policy, one is a risky policy. The finding was that this framing, 22 percent of respondents chose the risky policy.
Then when you switch it and say rather than 4000 people will be saved, 8000 people will die, and talk about deaths in the second treatment too, you find that the number of people choosing the risky policy increases dramatically in the loss framing. These are almost exactly the same results that Kahneman-Tversky found in the original study.
So, and this is true both at the World Bank and DFID in the UK Government. So what we’re finding then is that respondents were much more likely to choose a risky option in the loss frame, and this may be consistent with the choices of investors and the advice they’re getting not to go into countries in West Africa at the time.
We also asked respondents to predict the views of poor people. When it came to self-efficacy 70 percent of people made the statement that what happens in future mostly depends on me, but only 20 percent of poor people in Jakarta, Nairobi, and Lima would think this way.
In fact, it was even higher in those countries. The same is true with helplessness. Respondents were also extremely pessimistic about attitudes towards vaccines in Jakarta, Nairobi, and Lima. In fact, people were not as worried about the risks, particularly the risk because of sterilization as people thought.
So people, what’s happening here is that World Bank staff have a mental model of themselves, a different mental model about how poor people think, and they weren’t right in fact about how those poor respondents might respond. They have lots of data on how poverty works, but not enough data on the mental models associated with poverty.
In deliberation exercise, Stephan brought 100 different economists together in an annual retreat and found that deliberation did reduce confirmation bias, but really had no effect on framing, the reason being we think is that in a confirmation bias there actually is a right answer, and deliberation can reduce the likelihood of getting it wrong.
In the Asian Disease Risk Preference Study, there wasn’t necessarily a right answer. The former White House scientist (Name) called Aha Problems. The latter is not.
So these are some examples on the desk. What do we do about these kinds of things in the field? One great study from Nav Ashraf and colleagues involves the sale of female condoms in Zambia, in which hairdressers were incentivized to provide, to sell female condoms in a couple different ways, a 50 percent commission, a 90 percent commission, or social recognition, merely stars. And what they found was that the social recognition intervention was far more effective than a financial reward of any size.
We, inspired by this paper, some colleagues and I in the World Bank’s Behavioral Science Unit, the MIND, eMBeD, Mind Behavioral Development Unit, did some work in Nigeria. We were interested in this question about social incentives for expenditure tracking.
As you know, there are concerns about efficiency, cost effectiveness, accountability or leakage. And one of the responses is to track expenditures in clinics in developing countries. But the problem is it is often the case that people simply don’t fill out their spreadsheets, so the whole tool doesn’t really work.
We did some work with governments in two states, of Nigeria, Niger State and Ekiti and what we did basically was to send in observers to look at these spreadsheets, and see they were actually tracking expenditures, and then convert these assessments into STARs, like in the paper in Zambia. We posted these up in different locations, and found that, in at least one of the states, the STAR treatment, the Social Recognition Treatment, improved tracking of expenditures, at least over a short period.
I also want to share with you, let me just say that the general idea here is that when you use standard incentives for healthcare providers, it’s often something like if you do more we will increase your pay by $X per month.
There are a variety of other ways to do incentives, and Kevin of course was talking about this. We can say we’re increasing your pay by $X per month unconditionally. This is simply gift exchange, and there is a lot of work showing that unconditional gift exchange really works. It’s less effective, because you don’t have to monitor what happens, and people feel kind of this feeling of reciprocity to improve their work.
You can also do social recognition, which is something like we are measuring your effort on this activity, and we want you to do more because it’s important. The principle behind the STAR treatment and the answers from Nigeria and Zambia that I just showed.
And then finally you can do something like if you do not do more we will decrease your pay by $X, and this of course is loss aversion. It’s very hard to do, but there is some work done, John List did some work on schoolteachers in Chicago that shows that this really powerfully can make a difference.
One of our coauthors on the health section of the World Development Report was Ken Leonard, University of Maryland. And I wanted to share a little bit about some of his work. He did a fascinating study involving health workers providing outpatient care in peri-urban Arusha in Tanzania.
The idea was that the patients leaving certain clinics were interviewed about how well the clinician they just saw followed standard protocols for assessing situations in which patients came in with fever, cough, and diarrhea. These interviews were conducted with patients after they left the consultation. The interviews were unannounced, although they may have heard a little bit about, clinicians may have learned something about this.
So what they found was that at baseline about 74 percent of clinicians outside Arusha were complying with standard protocols for treating patients for these kinds of conditions.
Then they sent in a peer to watch, to see how they were responding. And this had the effect of increasing adherence to these standard protocols. People felt observed. This would be standard incentives. This is a principal agent model in which you’re watching me and I worry that I might lose something if I don’t do my job well. However, post scrutiny, a few weeks later, those same clinicians were monitored, and their overall adherence went back down again.
Then, rather than scrutiny, they tried something that they called an encouragement visit. In this scenario a well-known doctor visited each health worker, and with a script said hey, why don’t you do this? Why don’t you think about that? In some cases, key components of that protocol were primed, like what did you think about this drug, did you not think about that drug, did you take that to conduct that test, did you check for temperature, blood pressure, that kind of thing.
So there was priming and non-priming in these encouragement visits. And what they found was that two weeks later the adherence to the protocol went up. And then four to six weeks later, after the study, it went up even further. Priming had a bigger impact than not priming.
The really striking thing about this study to me is that then they went back two years later, and you can see all the way on the right that in the long-term follow-up it was significantly above baseline, even two years, just from a single encouragement visit and awareness of the intentions behind that encouragement visit that the clinicians experienced.
So my takeaways from this study is that in some situations at least encouragement may be more powerful than scrutiny. We tend to think of incentives as along the principal agent model of payments for good behavior and loss for bad behavior. But sometimes encouragement can be more powerful.
Now, once people switch to a new model, like hey, we’re all in this together, something like that, one may then need to induce social incentives and monitoring incentives to maintain that new kind of approach to what is going on, the new approach to behavior and adherence.
So I will just leave you, I think to save a little bit of time for conversation, I will just leave you with the thought that there are a variety of ways to motivate behavior and overcome blind spots in change efforts that are effective for clinicians and providers as well, not just for patients.
And that we ourselves should be active, to be thinking about our blind spots, to realize that we are not just the riders that we think we are, but in fact oftentimes when we behave are just simply following our elephant instincts. Thanks so much.
LEONARDO CUBILLOS: Thank you so much Varun for your presentations, and the thoughts you shared. I would like to invite our audience and our panelists to use the chat box to submit questions that you may have for Varun.
We have one question from, we’ll probably mispronounce her name so I apologize in advance, Yuyu Zhou, it says Varun, have you stratified these interventions by populations from different age, sex, ethnicity groups? Will these interventions be effective the same, or will they be fair and thus improve health equity?
VARUN GAURI: That is a good question. We did not do that, as the sample sizes were too small. And I think, if I could take a minute to talk a little bit about that Nigeria paper, one of the interesting things about that paper, that work, is that we found that the social STARs interventions worked in one of the states, Ekiti, but not in Niger, as I mentioned. And we thought wow, what is this, what’s going on, how can we explain this heterogeneous impact?
And fortunately, there turned out to be a quantitative service delivery module in those two states, in Nigeria, just before we did our survey. And we thought fabulous, let’s put in everything we can find, to see if we can explain why this intervention worked in one of the states but not the others.
And so we put everything from like population density to types of physicians, physician experience, every kind of thing you can imagine, and none of those things explained the difference, could really explain why it worked in one state and not the other.
So our key takeaway is we are out there measuring in our work a lot of demographic data to understand what works and what may not when we look at heterogeneity, but we don’t measure behavioral data and mindsets. We don’t collect systematically data on whether people are groupie or not, or how risk averse or loss averse they are. To really understand how these interventions are going to work in human populations we need to be collecting more of that behavioral data.
LEONARDO CUBILLOS: Another question. Why do you think that social recognition was so important as a motivator to hairdressers, even about large financial incentives?
VARUN GAURI: Hairdressers made most of their money through hairdressing. The condom sales were sort of a side gig. And so their actual impact on total monthly income through these commissions wasn’t huge, whereas the social recognition was new and interesting and it signaled importance.
So the takeaway from that paper really isn’t well let’s get rid of salaries and live on STARs, that’s not the point. The point is that at the margins you can change behavior through social recognition. Many organizations know this. But the point really is that the public sector can do this a little bit more effectively.
LEONARDO CUBILLOS: There is a comment that I would like to turn into a question if I may. The comment reads, very nice takeaways, in LMICs we are struggling with the limits to how much lay health workers can be burdened with complex interventions.
So yes, lay health workers deliver not only interventions to address mental health, but they are delivering antenatal care, and maybe delivering interventions to manage diabetes or other non-communicable diseases.
You spoke about the role of priming in raising fidelity and quality. Do you have any thoughts about how that can be used in LMICs in the setting of complex interventions, where the interventions are complex and the timing cannot be necessarily linear?
VARUN GAURI: So, I guess I would point to a couple of things. One of my former colleagues, Julian Jameson, and also Berk Ozler were doing some work in Cameroon about the question of the uptake of contraception. Obviously it is a complicated topic for young women, there are a lot of social norms about whether women should be doing that, should be having sex at certain ages, a lot of social pressures. And that might not be the right choice for some women in some contexts.
And so they were developing basically a decision tree which could be automated to help rural healthcare providers without necessarily high amounts of training to think through that decision. And it was kind of an interactive protocol, not just do this, do this, do this, but sort of elicit information from the patient, from the girl, and then see what’s effective.
Similarly, we were doing some work in India about the transition from breastfeeding to complementary feeding in BHAR, and we were beginning to think about what a decision tree and an automated kind of cognitive support could be for rural health providers, Anganwadi workers in India. So I think that is something that is definitely worth pursuing. It gets at the idea that we have limited cognitive bandwidth, cognitive resources. And one way to do that is not just to sort of simplify, which is good, as Kevin was saying, but also to add decision supports.
LEONARDO CUBILLOS: Thank you. A question from Agnes Rupp. Josh Gordon mentioned Agnes, she was a pioneer in health economics at NIH, and she is now retired. But she asks Varun how can implementation science at NIH can use behavioral economics techniques to speed up the use of new technologies in real life settings, particularly NIH as it relates to mental health and substance use?
VARUN GAURI: Well Agnes, you and Leo might know more about that than I do, because I haven’t done a lot of work on mental health interventions. But I think the principles really are broadly applicable. Social norms in particular, I think would be appropriate in the context of high levels of stigma. What I think about is that work that Abhijit Banerjee and others did on the MTV soap opera Shuga, which tried to increase take-up of HIV testing across Africa, another situation in which the disease is highly stigmatized.
And when it was embedded into a soap opera, kind of normalizing it, trying to make it kind of fun, that made a difference. We know that soap operas work in other contexts, I’ve seen work on financial literacy in South Africa, there is a lot of work out there some of you may be well aware of. So I think community norm types things and soap operas can be one way to get around a situation where you have something that is really stigmatized.
LEONARDO CUBILLOS: Let me ask a follow up question to your answer to the last question. At the beginning of your talk you presented data that showed there was international professionals from the World Bank in (indiscernible) for example were also subject to (indiscernible) Do you know of research on health researchers and health science administrators, how we could also do policy work you do not see in these countries?
VARUN GAURI: I think the Shafir Redelmeier paper speaks to that, and Redelmeier has a couple of follow-up papers on that question. But I don’t know recent work that’s systematic in large scale on that, and I think that would be extremely interesting to pursue.
LEONARDO CUBILLOS: Yuyu Zhou asks another question about if these interventions were disaggregated by subpopulations, asks a follow-up question. He says thanks for the insights, follow up of the same heterogeneity questions. What are the examples of mindset questions we should ask?
VARUN GAURI: One that comes to mind, Rachel Kranton has done some work identifying when people are what she calls groupie. Some of us are very susceptible to peer pressure, we care about what other people think. Others of us are just kind of loan wolves, we don’t really care, screw it, I don’t care, that kind of attitude.
If people are more groupie, they may be more susceptible to social norms interventions, perhaps to soap operas in a different way, that’s one example. And she has some questions in her papers to see if you can identify whether this person is groupie or not.
Obviously there are modules for risk aversion that have been out there for a long time. There’s a fair amount of controversy about whether risk is the same across domains. Some would hesitate about using a module for risk taken from a profile about say buying insurance for financial outcomes to the health domain. But one could develop, risk in the health context is certainly one kind of question that one would like to ask in these areas.
LEONARDO CUBILLOS: Well, the time is up. On behalf of the organizers and the audience, thank you so much for joining us here today.
VARUN GAURI: Thank you so much, it was a pleasure.
LEONARDO CUBILLOS: Now we are going on a 15-minute break. We will resume at 3:15 Eastern time. Sara will lead us on the panel on social determinants of health. So 15-minute break, we’ll be back at 3:50.
JENNIFER HUMENSKY: Welcome. Thank you for joining us following our break. We have one additional panel on Social Determinants of Health. As we heard in Dr. Gaskin’s plenary earlier today, it is so important to understand the role of social determinants of health in understanding the impact on mental health and substance use disorders.
We have three panelists here, Victoria Baranov from the University of Melbourne, Bohdan Nosyk from Simon Fraser University, and Marjorie Baldwin from Arizona State University. I will turn it over to Victoria.
PANEL #3 – SOCIAL DETERMINANTS OF HEALTH
VICTORIA BARANOV: Thanks for inviting me today. I will be talking about basically two papers that I have been working on, on the impacts of treating maternal depression in particular, but broadly speaking, this fits into the research agenda on mental health and economics.
Mental health is particularly important for economics, and the way that economists think about mental health is in particular not just an outcome that we of course care about but also because of its productive capacity and the potential for what we call a psychological poverty track, which is the idea of sort of a vicious cycle where poverty causes stress and other conditions that increase the risk of mood and anxiety disorders which subsequently potentially affect economic decision-making in ways that reinforce poverty.
So, a specific way of thinking about it might be that poverty increases stress which increases the risk of depression; depression affects, for example, beliefs, creates pessimistic beliefs, pessimistic viewpoints about the world around an individual and their abilities. And an individual might not invest themselves and might not invest in the future, might not work as hard as they could, so therefore they might not earn as much as their potential, which reinforces their state of poverty.
What we know from the literature is that there is actually a fair bit of causal evidence for the link from poverty to mood and anxiety disorders. There has been a recent meta-analysis published in Nature and Human Behavior with many studies because we have a lot of randomized controlled trials that provide relief in cash transfers to individuals in poverty, and through that mechanism we can see that there is alleviation of mood and anxiety disorders as well when cash is provided.
What we know much less about is how mood and anxiety disorders or mental health in general impact economic decision-making from many different angles. One of the ways I will be focusing on is specifically economic decision-making by women during the time of childbirth and the early years of child development in low and middle-income settings. This is primarily what women do in this context; they are homemakers, and they care for children, so that is, in essence, their work even if it not remunerated in the market. But we will look at women’s empowerment and child development as well as the sort of spillover effects and knock-on effects of treating mental ill health.
The first paper that I will discuss was published in 2020, which looked at the impacts of — it’s a long-term follow-up of a big RCT that treated perinatal depression in Pakistan. In low and middle-income countries we know that basically it is essentially untreated, so women who have perinatal depression are not treated at all despite the fact that we know that there are effective treatments, one of which is cognitive behavioral therapy.
So we looked at the Thinking Health Program, which was one of these very large RCTs that was very successful in the short run in reducing depression. It was also delivered by local health workers. They trained the local community health workers, so it was not specialist care; it was meant to be low cost and scalable.
Initially we know that it was very effective in treating depression. What we did is we followed up essentially seven years after the intervention concluded and we gathered a rich set of measurements around what happened to the mothers, their financial empowerment, their investment in children and other outcomes, especially child development outcomes but other sort of outcomes that we think might be important in elucidating mechanisms, things like fertility for example that might have been affected by treating depression and might also have effects on how parents invest in their children and so on.
One of the things we added to the literature — you might have seen Kate Orkin’s talk where they looked at many different trials and did a meta-analysis of many of these types of trials. This one really looks much further out in the long term. Also, we were very careful in measuring a broader set of outcomes that we think are important for looking at economic decision-making in this context. We also measured these outcomes in ways that were not just self-reported, so we had enumerators go out and visit the schools that the parents were sending the kids to in order to measure school quality.
We had visual inspection of learning materials and educational materials. We had the enumerators trained in assessing how the mothers interacted with the children to look at the type of parenting that was occurring.
These are sort of expanding the set of outcomes and looking in ways that might be not subject to sort of Hawthorne effects or experimenter demand effects because of their enumerator-measured or not subjective measurements but objective measurements, which I think is also important in a setting where if you’re treating depression you might have high levels of experimenter demand.
In short, the findings we have build on the initial findings which were published in the Lancet, not by me but a related team. Basically, they found really large effects in the first year of treating maternal depression, but what happened was when we conducted our seven-year follow-up we actually found statistically significant — the error bars look like it is not a persistent effect in reductions in depression, about 7 percentage points in depression rates, and about .2 standard deviation improvement in mental health.
The interesting thing here is we should really think about this as a really long-term follow-up. This was like baseline, one year later and then seven years later, but if we extended it out — and this is from another paper showing the different treatment effect sizes and then the time after enrollment from different studies. So this is from a paper by Bhat, et al. They have this really nice figure which actually shows the current study that I’m talking about where it’s kind of an outlier in how long they follow these women.
Unfortunately, we didn’t have any data on the intervening periods, but we do have a very, very long follow-up with quite a large sample of nearly 600 women and their children that we included. So these are fairly novel results to show how treating depression at one point in time and then having no other contact with these women actually had these really persistent benefits for their mental health.
Then when we look at the downstream effects on economic decision-making what we see is quite large and persistent effects on women’s financial empowerment, as well as impacts on parenting. We call it parental investment because we can’t just say it’s the mother’s investment. We don’t know if it’s specifically hers. But what we see is large and persistent increases — Seven years later we still see these effects of monetary and time-intensive investments. We see less effects in terms of parenting style, but, as I said, these were objectively measured and not just self-reported, in particular the monetary investments, which is an important thing to note.
Our effects seem to be driven by mothers who had girl children rather than mothers who had boy children, so the girls had bigger effects both in terms of the effects on depression as well as the effects we see on decision-making and investment in children.
Interestingly, when the children were about seven and one-half years old, we did a battery of tests to look at child development with socio-emotional development being mother-reported. However, cognitive development was assessed using a battery of tests that were objectively measured.
What we find is limited effect on child development. I think we worry about the fact that socio-emotional development was mother-reported because an intervention that treats mothers might make them more attuned to their children’s socio-emotional behavior, and so we might see that mothers who experienced a psychosocial intervention might actually be more likely to report emotional problems in their child, whereas mothers who are disengaged might say their children are fine.
And so we think that getting more objective measures of socio-emotional development — which is I think one of the main hypotheses that would link maternal depression to child psychological, socio-emotional development — we think that is probably the biggest link and biggest area where we think the effects show why we need better measurement and is something that we’re building on in future trials that I’ll talk about.
Just to recap on this paper, what we have is an intervention that we see improves mothers’ financial autonomy and parental investment. In terms of mechanisms, we find that treated women sought social support and had better relationships, which is part of how the therapy works. It sort of taught them to do that.
We found that there were no effects on fertility, mother’s physical health or husband’s income, and so to some extent these are dimensions that are not driving or part of the mechanisms that explain the effects on parental investment that we do see. However, we see these limited detectable effects on child development.
We ran another intervention related to this one called the Thinking Healthy Peer that was more scalable, but we also wanted to know, as we’re measuring new things, whether we can improve measurement of early child development, especially in that first year when children can’t really report things very well on their own, and whether we can identify better markers of child development through biomarkers and also looking at does treating mother’s depression have any physiological effects — again, if we can measure this with biomarkers.
In the second paper, as I mentioned, it is a similar intervention with a new set of mothers. We followed them over a long period of time. The study is assessing the impacts of intervention on biomarkers measured of HPA axis hormones from hair at one year postpartum from both the mothers and the infants in that age group, and what we are going to try to see is whether we can detect any of the effects of the intervention at the physiological level both for the mothers and infants. We focused on HPA axis cortisol and DHEA in particular because HPA dysfunction is thought to be involved in depression and also plays a key role in brain development. So it is directly linked to depression. As we’re treating the mother’s depression but also looking for the spillover effects onto the children, it might be able to give us a glimpse into what’s happening for the infants.
The results that we find — the cortisol, cortisone and DHEA here, I’m just showing the effects if I just compare the depressed controls to a healthy comparison group, and what you see is the depressed women have higher levels of cortisol and cortisone, so higher levels of stress hormones, but lower levels of DHEA. And when we treat these depressed women, it lowers their levels of cortisol and cortisone and increases their levels of DHEA, bringing the depressed women closer to the healthy comparison group. So these are the physiological effects we see for the mothers.
For the children, we actually see no pattern and no relationships with depression in terms of cortisol and cortisone, whether the mother was depressed or not. There are no treatment effects there. But we see these meaningful effects on DHEA where children who were treated have higher levels of DHEA, and we see that the infant’s DHEA at one year is predictive also of future cognitive development that we are able to objectively measure via Bayley scales of language receptivity.
This is a very small sample. We have an N of 100, so it’s a subsample of the bigger trial at one year that we have measured these biomarkers, but it could be a fruitful measurement to try to capture some of the mechanisms that are underlying these depression interventions.
I will stop here and turn it over to Bohdan Nosyk who is going to be taking over with the next session.
BOHDAN NOSYK: Thanks very much, Victoria. My name is Bohdan Nosyk, and I am very pleased to be invited to present at this great meeting. It has been a really interesting morning already. The title of my presentation is The Role of Race and Changing Demographics on the US Efforts to End the HIV Epidemic.
Race and ethnicity were not a primary focal point for us at the outset of this project; this is really designed to identify combinations of evidence-based strategies for different cities in the effort to support their HIV/AIDS responses and to really make the point that the cities are very different and they required tailored strategies quickly.
As we progressed with the work it became evident that you cannot talk about HIV in America right now without talking about race, so I will thank NIDA for providing funding to support this work. We are now in a renewal stage and extending the work that I will be presenting to you, and I am also grateful to the Fulbright Fellowship program.
I will acknowledge that this work was primarily conducted on the territories of the Musqueam – Coast Salish peoples including territories of Musqueam, Squamish, and Tsleil-Waututh, nations who are in beautiful Vancouver, British Columbia, Canada. I also spent a fair bit of time at Emory University, which is on the territories of the Muscogee Creek people.
I wanted to mention that this presentation is on race in America and race in the HIV epidemic. And the Fulbright program is great; it is meant to be a cultural exchange and facilitated a lot of the work that I will be presenting, but I feel the need to point out that Senator Fulbright himself was a segregationist. Those sorts of divisions are still prevalent today. I hope some of the work that we heard earlier today — Dr. Gaskin’s presentation was fantastic and was a great send-up for what I’m going to be presenting today. Clearly, there is a ways to go here.
Just to give you a background, this presentation is on HIV and focuses primarily on ending the HIV/AIDS epidemic initiative which was announced in 2019. The goal is a very ambitious one, to reduce HIV incidence by 90 percent in 2030, so, in a ten-year span. The premise is that we have the tools that we need to end the epidemic in terms of diagnosis, treatment, prevention, and our ability to respond to growing epidemics. The tools are there. At the moment, they are not implemented as effectively and as fairly as they need to be, so the questions our research is asking broadly are what will it take to meet these targets and how much will it cost.
The “Ending the HIV Epidemic Plan” is really predicated on this notion that the epidemic is concentrated now in a small number of large urban centers primarily, so 48 counties bear a disproportionate burden of the HIV epidemic. Though this project began in 2016, several years before EHE was announced, we were lucky enough that we managed to cover six cities and 12 of these 48 counties which eventually were included in the EHE strategy as focal counties — just driving home this point that geography is becoming more and more of a factor.
I think inequities in access to care are growing between the South and Southeast and the rest of America, and a more and more larger share of new infections is being concentrated there. Network-level factors are really driving these things, and we will talk about that in just a moment, but geography is one element, race is another.
Another point I want to make before really jumping into this is we have seen a disconnect between the randomized trials that demonstrated the preventive benefits of antiretroviral treatment in serum-discordant pairs. We have seen a disconnect between what we’re seeing in tightly controlled randomized trials and population-level implementation studies.
Stephan posited a theory as to why we are seeing that. It’s really core group theory, which is where prevention and treatment gaps among the relatively few who are most at risk of acquisition and transmission can really sustain an epidemic, and I think that is what we’re seeing now, and unless we can act in a way that reduces inequities in access to care, primarily among young black and Hispanic men who have sex with men, we are going to continue to see this disparity grow between the South and Southeast and the rest of the nation.
Our objective for this project was to consider combinations of 16 evidence-based interventions to diagnose, treat and prevent HIV infections, and we aimed to identify the highest value combination implementation strategies to reduce the public health burden of HIV in six US cities. We have judged value on the basis of quality-adjusted life years.
I’m assuming, since this is a health economics meeting, a meeting of health economists, you will be familiar with that concept. The idea is QALYs simply capture more time spent in good health, and in context, in particular it’s important for capturing benefits from reduced morbidity, mortality and transmission from HIV, and there is an implicit focus on equity and maximizing population health.
These are the six cities that we initially focused on. They are very diverse. We have Atlanta and Miami in the Southeast. They are non-ACA-adopters; they are fast-growing cities and with very different racial and ethnic makeup. On the other end of the spectrum, we have Seattle and New York who are very advanced in their response, and so that breadth really allowed us to test this notion of each city requiring a tailored response.
A lot of our background work is already published in the public domain. Again, this is year seven of this project so we are in the process of updating and expanding all of this, but our scientific case that we made for this, the evidence synthesis under the primary data analysis, the validation of our model, is all in the public domain and I have listed the papers here.
We have used a dynamic compartmental model for HIV transmission and focused on the population of people age 15 to 64, stratified in a number of different ways: race/ethnicity, sexual risk behavior, HIV risk behavior. These are all strata that were important to our model, and the evidence synthesis that we conducted, again, is published. You can see the lengths that we went to to get city-specific information on these factors wherever possible.
We first tested these interventions individually to determine their impact on incidence, which is on the right panel, and their cost-effectiveness. This is for the city of Atlanta, Georgia. You can see our 16 interventions in the left margin here. They were prevention interventions including syringe exchange, MOUD as well as PREP, a range of different options for testing, and ART initiation, retention (indiscernible). These were all chosen from the CDC compendium. We used real-world evidence on scalability and the best evidence that we could find on effectiveness to populate these. We didn’t want these to just be hypothetical large increases in access that had never happened before. All our scale-up assumptions were based on the best available evidence out there.
The first story is that there is no magic bullet here; there is no single intervention that will reduce incidence by more than 7 percent in Atlanta, so this really does need to be a combination strategy. There are a lot of different things that have to happen simultaneously to get anywhere close to the 90 percent target.
I also want to point out that we didn’t make any assumptions that inequities in access to care would magically disappear. When we scaled up access to these interventions we assumed that the distribution of black and white and Hispanic people accessing these interventions would stay the same, so this is proportional to baseline service levels. That was our a priori assumption for our initial bit of work.
What we have here are health production functions; on the Y-axis we have incremental QALY gains from implementing all reasonable combinations of the 16 strategies, 16 evidence-based interventions, that we considered; and on the X-axis we have incremental costs over a 20-year time period. So the top line, the health production function, really shows you the best-valued strategy that provides the highest level of health benefits for a given investment level.
We had an idea that these would look different from city to city. I think even we were surprised at just how different these were. It was an interesting exercise, the sort of thing you see in textbooks, but we don’t have that many opportunities to do this in actual applied work so we were excited to see it.
Of course, each city’s health-maximizing combination of implementation strategies was unique. Every city had a different mix of interventions, and there were between nine and 13 individual EBIs in these. Of course, the health impacts were very different, and we saw the greatest value in the cities with greatest need.
Miami was an interesting case study. Miami has had very little scale-up in PREP access in particular, and Miami is a city with a large concentration of Hispanic people. On the left, the left cluster is all strategies that included pre-exposure prophylaxis uptake, and on the right, these did not include PREP, so PREP was really a big driver. You can see over a 20-year time horizon we were projecting a cost savings of nearly $500 million from scaling up access to these combinations of interventions.
Another point I want to make is that if you look at the furthest right, the costliest strategy — so there was limited testing available and no scale-up to syringe exchange or PREP — it was estimated to cost almost $1 billion more than the optimal strategy and produced only 30 percent of the QALY gain. So the strategies that you invest in matter greatly, obviously.
Again, all of this was done under the auspices of no changes in the distribution and no attempt to address inequities to access. You can see they are large. Here we plotted the incidence ratios between black and white in the left panel and Hispanic versus white in our six cities starting with the 2020 estimates and then our status quo and ideal implementation scenarios at 2030. You can see, even if we scale up these strategies to ideal levels, so 90 percent of the target population, we still would not be eliminating racial and ethnic disparities. So it is a persistent problem and it needs targeted intervention to address.
And this really brings us to the heart of the matter. We knew this was a factor. I think personally I underestimated just how much of an issue this is in HIV. What it boils down to is mutually reinforcing manifestations of racism at the heart of the HIV epidemic in the US. On the left we have predicted our dropout rates, which we estimated from the HIV RN database. This gives us information on people accessing antiretroviral treatment in clinics, and so we broke this down by region and by race and ethnicity and you see dropout probabilities are highest in the South and highest among black individuals.
On the right we have our assortative mixing patterns between low-risk and high-risk black, white and Hispanic individuals. This is something we have seen over and over again. I think this reinforces what Greg Millett’s systematic review on this topic showed, which is that there is very little mixing, particularly in black populations, among the other races, and really you see these stratified racially segregated sexual networks. The darkest segments signify choosing a sexual partner within your stratum at the highest level.
We have these inequities in access to care and strong barriers to sustained access to care, and then we have these racially segregated sexual networks. What that means is, if you have a prevention intervention like PREP that is scaled up in white populations, we just don’t see the benefits of PREP filtering into black and Hispanic populations quite as much because of this racial segregation. So it’s a combination of both these factors.
I will point out that only one of these is mutable; we only have control over access to care. We are not going to necessarily influence sexual behavior among people. It’s partly preference but it is also influenced by socioeconomic status, socioeconomic strata. So this is sort of at the heart of what is driving HIV incidence today.
More to that, Atlanta has obviously a large black population but it is growing quickly, and its Hispanic population is actually projected to double in size during our study period between 2020 and 2040. Similarly, Miami is growing quickly and its Hispanic population is growing rapidly as well. Disparities in access between Hispanic and white are not as large as they are between black and white, but there is still a differential there. What this means is that there is more fuel going into this fire and this is going to get harder and harder to control unless we address this now and unless we make some significant structural changes to make it easier to reach these populations.
We decided we really needed to articulate this point in a very tangible way so we conducted a distributional cost-effectiveness analysis. A typical cost-effectiveness analysis is focused on maximizing population health regardless of who receives it. Distributional cost-effectiveness analysis thinks about how those benefits are distributed between racial groups or different economic strata.
Here we decided to focus on race, so we considered two different scale-up approaches. Again, our initial was proportional access, proportional scale-up to these different interventions. The alternative was an equity-oriented approach, whereas scale-up is done across racial and ethnic groups proportional to their new diagnoses in 2019. In other words, proportional to where the new cases are coming from and where the need is coming from.
To demonstrate, our status quo — If the initial distribution of people accessing PREP is 78.7 percent white, 5.3 percent Hispanic and 16 percent black, in doing the proportional services approach, which is what I’ve shown you up until now, we are keeping those portions constant. But in the equity-oriented approach you’re scaling up according to where new diagnoses are happening, so you end up with a very different distribution afterwards.
JENNIFER HUMENSKY: Could we ask you to you wrap things up in the next minute or two because we want to –
BOHDAN NOSYK: Sure. This is the main point that I’m trying to make. Obviously, an equity-oriented approach is hugely beneficial, it’s going to make a huge difference. It actually is impossible to get anywhere near the targets in Atlanta without an equity-oriented approach. You go from a 34 percent decline in incidence by scaling up all these strategies up to a 68 percent decrease in incidence at much lower cost.
There are plenty of reasons why these inequities in access persist. There are strategies to do this. I think it’s going to take significant structural changes and implementation strategies that are tailored to reaching populations of black and Hispanic men who have sex with men in particular. That is what we need to focus on in the coming years.
I will skip through the last bit. I just wanted to show this map, following up on Dr. Gaskin’s maps. Segregation and the structural built environment is really such a persistent factor in Atlanta. You can see the difference between the white and the black populations in the North and South and where income is concentrated and where services are concentrated. The South of Atlanta is very much a care desert, so there are things that we need to address there.
Thank you for your time, sorry for going over. I will acknowledge all my collaborators and thank you. Next up I believe is Dr. Marjorie Baldwin.
MARJORIE BALDWIN: Thank you, Bohdan, and thank you, Jennifer, for inviting me to talk in this very interesting and provocative meeting that you’ve been holding today. My topic is Job Accommodations for Workers with Serious Mental Illness in Regular Jobs. This is joint work with Rebecca White from ASU and Stephen Marcus from the University of Pennsylvania, and this research was supported by an R01 grant from NIMH.
Let me begin by defining a few of the terms in the title. By serious mental illness we are referring to diagnoses of schizophrenia, bipolar disorder or major depressive disorder. Job accommodations mean any adjustments to a job or workspace that make it possible for a worker with functional limitations to do their job. The functional limitations most frequently associated with serious mental illness are in the cognitive, social and emotional domains.
And finally, by a regular job we mean one that pays at least the minimum wage, is not set aside for persons with disabilities and was not obtained with the help of mental health services.
So how does this topic relate to the overall topic for this session, which is social determinants of health? In fact, for persons with mental disorders, work actually promotes health. It adds purpose and structure to their lives; it’s a source of identity, dignity and self-esteem; it provides them with a means of social inclusion in a work community, and it is their pathway to financial security and independence. In fact, for many persons with SMI, a paid job is their number one recovery goal.
One of the workers in our study told us that working in a paid job makes them feel normal and being normal validates you.
Many persons with even the most serious mental illnesses are capable of regular mainstream employment, contrary to the negative stereotypes of people with mental illness as incompetent, unreliable, undependable. The Americans with Disabilities Act and its amendments mandate that employers provide reasonable accommodations for workers with disabilities, but we know almost nothing about the accommodations that are provided to workers with SMI in regular mainstream employment.
The objectives of this study were to describe the nature and frequency of job accommodations for workers with serious mental illness in regular jobs, and then to identify factors associated with the probability of job accommodations for these workers. We are going to look at two types of job accommodations. The first are requests for employer-provided accommodations, those that would be mandated under the ADA. The second are accommodations initiated by workers themselves, and this is the first study that is looking at this phenomenon of self-initiated accommodations by workers with disabilities.
Our data come from a survey of 821 workers with serious mental illness that we conducted between 2017 and 2021. We recruited the workers through the Pulse survey, which is a large national survey sponsored by IBM. They identified potentially eligible subjects for our survey. The eligibility criteria included that they were employed in a regular job for at least six months post-onset of SMI. We put in the six-month criterion because we wanted to identify a relatively high-functioning cohort of persons with SMI who were capable of regular mainstream employment.
They had to have a self-reported diagnosis of SMI, be of working age — that is, age 18 to 65 — and the survey collected data on workers’ demographics and work outcomes, their human capital and characteristics of their illness, characteristics of their job and workplace, whether or not they had disclosed their mental illness to their employer, and any accommodations they requested.
Accommodations is the focus of this talk so I want to be explicit about the questions we asked. First, we said, did you request that your employer make any change or accommodation in your job or workplace to meet the needs of your mental illness. Then we asked, did you make any of the following changes in your job or workplace on your own to better meet the needs of your mental illness. In each case respondents could choose from a list or they could identify some other accommodation that was not on the list, and we organized these into meaningful categories based on groups in the literature.
These are the gamut of workplace accommodations that we looked at.
First was scheduling accommodations such as the freedom to take on scheduled breaks. Second were workspace accommodations such as the freedom to work from home, moving to a different workspace or having access to water and a refrigerator near your workspace. Another category was job modification such as job restructuring or changing off tasks with a coworker. And finally, there was a set of accommodations that involve the worker’s supervisor. These included having reminders of deadlines or being given extra time for tasks.
This is the first slide showing some results, and these are simply the frequency of accommodations that we observed in our sample. Column 2 shows the percent of workers requesting each type of accommodation, and Column 3 shows the percent of workers self-initiating that accommodation. The first thing you notice in the first row is that only 25 percent of the workers in our sample requested any accommodation from their employer but 84 percent of them said that they initiated accommodations on their own, so this is a huge finding given that the literature virtually ignores self-initiated accommodations when talking about workers with disabilities.
The most frequent kinds of accommodations were scheduling accommodations and then changes to the workspace. Slightly over one-third of our sample said that they self-initiated changes to the workspace, and we asked them what those changes were. It may be interesting for you to see what they said.
One workers said, “I rearranged the setting of my cubicle to put my viewpoint as a wall where I would have no distractions and could focus on the tasks that I have to do.” Looks like he has difficulty concentrating. “I made the lighting so it was easier for me to concentrate.” “I added a white noise machine.” Another said, “I changed my location in the office to a less busy location and less interaction.” Social interactions are one of the areas with which many people with SMI have difficulty.
Another said, “Mostly declutter. When everything is put away it gives me a calmer workspace.” This was a very common response to what they changed in the workplace. Many workers said that they reorganized, they cleaned so that they could be better able to function.
It’s the end of the day so I am not going to go through any Greek; I’m just going to end. These are pretty simple logistic regression models that we estimated. In the first, the binary dependent variable equals one if a worker requested employer-provided job accommodations; in the second, the binary dependent variable equals one if a worker-initiated job accommodations on their own; and the independent variables in the model represent factors that affect either the need or demand for accommodations, or the feasibility or the costs of initiating accommodations. We divided these into three categories. I am going to skip this slide and go right to the independent variables.
The first category is demographic variables. We included age and education categories, race, ethnicity and gender. The second category of health-related variables, we controlled for the worker’s diagnosis of mental illness, and we include binary variables that equal one if the worker self-reports moderate to severe cognitive, social or emotional limitations. We also include binaries identifying workers with comorbid physical or substance use disorders.
Then we have as set of workplace-related variables. Job autonomy equals one if a worker reports that they have some control over how they work, with whom they work, where they work, when they work. Job intensity equals one if a worker says that they have to work to strict guidelines or there’s a rapid pace of work. We include job tenure, measures of workplace culture, supportive firms, supervisor and coworkers and occupational categories. And I should say that many of these barriers — for example, the functional limitations variable and the workplace culture variables — were derived from questions on validated scales and measures.
Here are the descriptive statistics for our sample, and in the interest of time I am just going to focus on a couple points. Only 35 percent of this sample is male, but this reflects the US population with serious mental illness. Eight percent of our sample have some college credit or a graduate or post-graduate degree. This is higher than the US population and leads us to suspect that we were able to identify our target population of highly functioning workers with serious mental illness. Still, when you look at the limitations, more than 75 percent of the sample report moderate to severe cognitive, social or emotional limitations.
Approximately 50 percent of workers say they were in a supportive firm or had a supportive supervisor, but only 30 percent say supportive coworkers. Also, 46 percent of our workers have job tenure greater than three years, so this again distinguishes them from a population receiving mental health services where the average tenure of a job is more like six months.
This slide shows the results for the logistic regression model of the probability of requesting accommodations with odds ratios in the right-hand column. I reported variables that are significant — a bachelor’s degree as a positive correlation with requesting accommodations. These limitations are what we would expect: workers with more limitations are more likely to request, and more severe limitations are more likely to request accommodations. Workers with comorbid physical conditions are more likely to request accommodations but we didn’t get a result for substance use disorders.
Workers with longer job tenure are more likely to have requested accommodations. This causality could go in both directions. It may be that workers wait a while to disclose their mental illness because of the stigma associated with mental disorders, or workers who have been at work longer and have established themselves are more likely to disclose and request accommodations.
On the other hand, it could be that those who have requested accommodations and received them are more successful and able to continue to work for a longer period, and, not surprisingly, having a supportive firm or a supportive supervisor are also positively correlated with the probability of requesting accommodations.
When we looked at the model of correlates of self-initiated accommodations there are fewer variables that are significant, but those that are are highly significant with strong effects. Relative to the omitted group of workers aged 50 to 65, younger workers are much more likely to have self-initiated accommodations.
Workers with social limitations are more likely to self-initiate accommodations but we get no effects for cognitive or emotional limitations. And as we expected, job autonomy has a large and significant effect on the ability to initiate self-accommodations.
In summary, what do we think are the main contributions of this study with the focus on workers with SMI in regular jobs? This is one of the first studies to focus on workers with serious mental illness employed in regular mainstream jobs not set aside for persons with disabilities. And we think it goes a long way to combatting the stereotype that workers with serious mental illness are not capable of these jobs. We talked to hundreds of workers employed in jobs ranging from lawyers and accountants to office managers and welders.
The second thing is the prevalence of self-initiated accommodations which has almost entirely been ignored in the literature. I don’t know if this is a phenomenon that is also prevalent among workers with physical disabilities. It may be more prevalent among workers with mental illness because of the stigma of the disorder and they prefer to not disclose and initiate accommodations on their own.
Another important finding is the frequency of workspace accommodations that the workers detailed. This is contrary to the literature which says that workspace accommodations are not that common for workers with mental disorders.
There are some limitations to this study, of course. This is a national sample but it is not necessarily representative of all workers with SMI in regular jobs so we have to be cautious in generalizing results. Also, small sample sizes for some variables may have made it difficult for us to detect correlates. This is especially true of workers with substance use disorders; only 5 percent of our sample reported these, and also workers in some racial and ethnic categories.
Finally, let me end with what I think our study says to employers of workers with mental illness. First, when you look at the nature of job accommodations for these workers they are very different from the accommodations that we think of for workers with physical disorders. Employers may not be aware of the types of accommodations that are needed by workers with mental illnesses. They can refer to the job accommodations network. JAN can help. Their website provides detailed information on accommodations, and they have specialists that will work with employers at no charge. Their website is on this slide.
Secondly, job autonomy is a highly significant correlate of self-initiated accommodations, so it’s a good job match for workers with SMI, and everything that employers can do to give workers with mental illness, mental disorders, more control over their work will help them accommodate their functional limitations and be successful.
Then we saw that workers appear to have ways to accommodate social limitations on their own, like we saw with the worker who moved his workspace so he would have less interaction with coworkers. But accommodations for cognitive limitations appear to require assistance from an employer, such as reminders of deadlines and written checklists of tasks.
And finally, our results suggest that a supportive firm culture makes it comfortable for workers with SMI to disclose their mental illness to their employer and to request the accommodations that they may need.
With that, I will stop and turn it back to Jennifer.
JENNIFER HUMENSKY: Thank you, and I welcome back our other panelists, Victoria and Bohdan. Thank you so much for these presentations which were fascinating. I especially want to give a shout-out to Victoria because if you are, in fact, in Australia it’s bright and early in the morning for you. We have quite the time zone spread here.
We have time for a few questions. Victoria, you are not going to be able to go into a lot of detail on this. The intervention that you mentioned that showed seven years of impact on women’s empowerment, can you say anything more about how you think the implementation of that intervention might have led to such longstanding effects? In addition, you actually found a stronger effect among girls than boys, and do you think there was anything about that implementation that might have led to that?
VICTORIA BARANOV: Yes. The implementation specifically I think was delivered by health workers, which I think is very helpful because they already were working in that space and they were very invested in helping the women recover.
But I think part of the success of these interventions is that they are treating maternal depression, and so systematizing CBT for this group who typically have a very — not typical, but very similar types of stressors and triggers for depression. I think the treatment is easier to make in a scalable public health kind of way than like a broad, let’s treat anyone with depression who has maybe much different triggers for it or stressors or risk factors.
The other thing that made it generally persist is — and I think that this is just an aspect of CBT — that it helps the individual — There is a learning aspect where the individual then becomes aware of, okay, these are the emotions that I’m feeling and these are the conditions that I have, and seeing those when they have a recurrence of depression or they’re starting to brew they the recognize the symptoms and are able to use the same strategies again.
So I think that CBT is not just a one-off treatment like pharmacotherapy would be where once you stop there are no persistent benefits. I think we see some evidence that this type of therapy does have persistent effects because it treats individuals in a way where they learn how to cope and take action to help treat themselves. I think that is one of the reasons it persists.
We’re not sure exactly why the women who had girl children benefited most. We know that in this setting women who have girls are technically disempowered over time because of the patriarchal society and the son preference. So I think one of the reasons perhaps the women who had girls particularly benefited might have also been because the community health workers were empowered women themselves and so there might have been a little bit of that interaction.
We don’t see this really strong engendered interaction when we use the peer-delivered intervention, so this is purely a hypothesis. We actually see that it benefited the boys, mothers who had boys, more than mothers who had girls. So we think the modality, who’s delivering the intervention and what their norms are, might actually influence who benefits more.
We have a very limited amount that we can say on that I think at this stage, but that is a very good question. Thank you.
JENNIFER HUMENSKY: For Dr. Baldwin, your finding that the self-initiated accommodations are so much more common than employer-requested accommodations, how do you think that this could be informative to, say, supportive employment programs or other workplace — you state workforce programs that are aimed at improvement in employment outcomes.
MARJORIE BALDWIN: I think that the correlates of self-initiated accommodations — job autonomy, in particular — enable a worker to initiate their own accommodations.
There has been a lot written about supportive employment and the accommodations available there. The nature of those accommodations is really quite different. The most common accommodation is the assistance of a job coach, so the structure of that job and how much independence the worker has would affect their ability to self-initiate accommodations. But this could have impact for job coaches in helping their workers to identify ways that they can accommodate their functional limitations themselves.
Was there another group you wanted me to –
JENNIFER HUMENSKY: I think that largely covers it. We are over time. Bohdan, there is one question I wonder if you could answer in one minute, which is that access to PREP has shown the disparities in access widened between black and white people. How realistic is the assumption that disparities will remain the same?
BOHDAN NOSYK: You are absolutely right, Heather, PREP access has been far greater in white populations and has driven a wedge and will continue to drive a wedge so long as we have these structural barriers in place. Our intent was just to set up a comparison between maintaining access at current levels and trying an equity-oriented approach and putting a dollar value on how much more benefit we would get in terms of incident cases of burden and how much more we can spend to make it so worthwhile.
There are reasons to believe that the disparities are going to get bigger with Gilead’s Advancing Access Program going away, so we will look at that in future iterations. We’ll consider strategies where disparities keep getting worse. At the time, we felt it was a conservative assumption. Maybe we were wrong.
JENNIFER HUMENSKY: Thank you so much to this entire panel, really enjoyed all the presentations. I will turn it over to Mark and Mike.
Agenda Item: Plenary C: Using Value-Based Insurance Design to Increase Use of High Value Care, Enhance Equity, and Eliminate Low Value Services
MICHAEL FREED: Good afternoon from the East Coast. I am Mike Freed, Chief of the Services Research and Clinical Epidemiology Branch here at NIMH, and I am really thrilled to introduce our final speaker, the high-valued Dr. Mark Fendrick. Among his many titles he is Professor of Internal Medicine in the School of Medicine and Professor of Health Management and Policy in the School of Public Health at the University of Michigan. He is also Director of the Center for Value-Based Insurance Design and is the Co-Editor in Chief of the American Journal of Managed Care.
Dr. Fendrick, take it away.
A. MARK FENDRICK: Thank you Dr. Freed, Jennifer and all of you who have slogged through the day. This has been spectacular. I have been here from the beginning. It’s our first snow day in Ann Arbor; we’ve had about five inches of snow accumulate throughout the day.
I have been asked by the team to go through a 25-year journey in about 20 minutes, so if any of you are interested in more details, I think Amy on the tech team is going to put a link in the chat to give you more resources on some of the areas I am just going to touch on, given that this agenda today has been so broad and deep that some of you may only be interested in one or two of the items that I’m working on today.
I am a practicing general internist. I have long been interested in spending our trillions of dollars in the US — I know there are a lot of non-US people on the call — more wisely. And as a clinician, I like to start out my presentation with there has never been a better time to practice medicine. It is really incredible and given many of the technologies that we have talked about today, it’s just remarkable that we are able to intervene to prevent diseases and prevent the complications or progression of certain diseases compared to when I started practicing over 30 years ago.
But if you think about this meeting and think about every other medical meeting, I think Dr. Freed would agree that, instead of talking about health, in most clinical meetings now we talk about money. And if Twitter is still around, I like to say my second most Tweetable soundbite is that I wish we could restore health to the healthcare cost debate. And many of the previous speakers have kind of inspired me to think that there really is attention to why I went to medical school, because I didn’t go to medical school to save people money; I in fact went to medical school to promote individual and population health.
As we have talked many times throughout the day, if you were to identify the most valuable services in either the prevention, diagnosis or treatment of mental health disorders, almost all of them are substantially underutilized for many reasons discussed throughout the day. And I am really glad I am not in the position of many of you, particularly those of you at NIMH, because I like to say the tension between quality improvement and cost containment is real, it is very stressful and it’s only going to get worse, largely for the reasons discussed through the day.
Our inability to deliver these services — just as Bhodan said, things like PREP — is remarkable, as our inability to deliver high-quality care lags behind the rapid pace of innovation. I like to call that Star Wars science and Flintstone delivery, and I think Dr. Freed is old enough to know who Fred and Barney are and he can explain this to the younger folks or the folks who are not in the US.
Moving from the Flintstones’ stone age to the space age, Star Wars, we need to change this conversation from how much we spend to how well we spend. Well, certainly not in some areas that we talked about — low-income areas — but in the United States almost everyone agrees that $4 trillion, or one of every $6.00 in our economy, is more than enough. It’s just that we are spending it on the wrong people, on the wrong services and the wrong places and at the wrong time. And for someone who has been doing this a long time, the alliterative phrase of moving from volume to value has largely been talking the talk.
We are starting to walk the walk a little bit, and most of that walking has been on what I would call the supply side or provider-facing initiatives, and there are a lot of economists who talked earlier in the day who want to make sure that how we pay clinicians like myself is critical in moving from volume to value. But I think it is also important and aligns with many of the previous presentations that we also have to think about the demand side, or think about the patient and how we motivate him, her or they about how they might improve their health.
In the United States — and this was touched on in some of the other presentations — how we interact with our patients who have insurance — which is a great majority of folks in the US although there is still an uninsurance problem — is consumer cost-sharing, meaning the amount of money that our patients have to pay out of pocket to see me in the office, to get a diagnostic blood test like a CD4 count for HIV, or to fill a prescription whether it be for a major depressive disorder or schizophrenia or other types of things.
The problem in the US for the most part is that this cost-sharing is what I call a blunt instrument. People are being asked to pay more for all services regardless of whether it’s something I beg them to do because I know how important it will be to them or whether they don’t need it all. They still have to pay pretty much the same amount.
A lot of these terms were discussed earlier in the day — cost-sharing, coinsurance. What keeps me up at night is something called deductibles, and for those of you who are not familiar with insurance, that’s the amount of money you have to pay out of pocket before your insurance kicks in. I have been told that 50 percent of millennials don’t know what a deductible is. All four of my millennial kids have crashed at least one of my cars, and because they have to pay it they damn well know what a deductible is.
But when I started in this space caring more about copayments and coinsurance, as you see, in the early 2000s very few people with insurance in the US actually had a deductible at all and if they did it was small. And now you see over three-quarters of Americans have a deductible, and even in 2019, pre-pandemic, individuals had to pay over $1,000 before their insurance kicked in, and families sometimes $3,000, $4,000, $5,000. This is really important in the fact that in the United States, the Federal Reserve has shown that 40 percent of Americans don’t have $400.00 in the bank. So I thank Jennifer and her team who have done such an amazing job with this event, holding it in January.
While most people are very happy to have the calendar year turn over and have New Year’s Eve parties, people with chronic diseases, particularly those in the populations we focused on today, really lament January 1 because that is when their insurance deductible fits in, and if you want to contact me I can let you know several people who inform me every January that they can’t afford in some situations life-saving therapies because of their lack of coverage, such as insulin or other really essential medications or diagnostic tests and procedures. So we have focused primarily on this insurance of blunt cost-sharing now going on over 20 years.
I just put this in here even though we have been talking about this for a quarter of a century. The New York Times this summer was starting to catch wind, and we hope NIMH and all of you on the call will understand, that a deductible is ridiculous, particularly for the populations that we have talked about today, those with substance use disorder or mental health problems. To put another barrier in front of these people to get the care they need that would improve their health or wellbeing makes absolutely no sense to me, and certainly makes my blood pressure and heart rate go up.
This is my most Tweetable soundbite, and for those of you who are following issues of cost and cost containment in the United States, so much attention is tied to prices. And I understand the price is stupid, but for me prices are way, way above my pay grade. I have focused almost all my work on what my patients have to pay to get the care they need, or out-of-pocket costs, and my most Tweetable soundbite, if anyone is Tweeting today, is Americans don’t care about healthcare costs. They care about what it costs them.
What I mean by that, for those of you in the US who have had a Covid vaccine, when people ask me what I do, I ask them what they paid for their Covid shot. And just about everyone, unless you were somewhat hoodwinked, says they paid nothing. And that is what I do, try to figure out ways to lower out-of-pocket costs for what the evidence would suggest is essential clinical services. While the Covid vaccine is a shining light and some of you might have heard about a Medicare copay cap for insulin of $35.00 a month, we still have a very long way to go, especially in the field of mental health.
I can’t give any presentation without showing the motivation for this work. This is my mom, she’s turning 89 in March. She said to me in the year 2000, “I can’t believe you had to spend $1 million to show that if you make people pay more for something they buy less of it.” And the great news is there have been several investigators who presented today showing that my mother was right. Our mothers are always right.
But as it ties to the focus of today’s sessions on underserved populations and communities of color, I wonder what my mother said when I told her that I had to publish a study showing that poor people and people of color are impacted by higher prices more than rich people. But this is understanding that the basic empirical need to show that Americans don’t make good decisions with their own money regarding healthcare has really been the crux of our work moving forward.
And on this slide, I made a couple additions specifically to the literature that I was able to quickly find on cost-sharing and disparities in mental health a really good paper. These are all fairly old, showing that depression worsened in older people eligible for Medicare who restricted their medication due to cost. Again, my mom would say that’s a duh. And then the well-established disparities in access to medications between beneficiaries with and without depression did not improve with Medicare Part D. Why I bring that up is that, in some areas, removing financial barriers is all you need to do to make some disparities go away. In the area of severe mental health conditions, we may need more than just insurance design changes or more generous coverage.
It is easy to say that blunt cost-sharing is a bad idea, we were asked by several people to come up with an idea that might be better. And value-based insurance design, for those of you who haven’t heard of it before, is hopefully an intuitive idea that instead of setting consumer cost-sharing on the price of the service where expensive things are expensive and low-cost things are cheap, what if we set consumer cost-sharing on the clinical benefit? In other words, low out-of-pocket or no out-of-pocket on the high-value care, and make people pay out of pocket — because I’m a libertarian — on the things they don’t need but they really want.
The good news is the intuitiveness, the messaging and the problem of cost-related non-adherence have led to a substantial number of public and private payers implementing some form of V-BID and they have focused primarily on chronic disease medications for which mental health disorders, primarily depression, have been part of this. And the good news is, if you target those medication classes where there are substantial offsets in other spending like we heard about earlier, we can see that improved medication adherence leads to better health without a large increase in total healthcare spending. On the right is a review of V-BID implementations that we published a few years ago in Health Affairs.
This slide is an important one tying to the last few sessions. The V-BID programs from the very beginning, when Dr. Michael Chernew and I wrote about the framework, we always felt that removing financial barriers would be equity-enhancing. And it should come as no surprise that the converse of that study my mother looked at is that, when we remove barriers, like this study of heart disease medications on the right, eliminating medication copayments reduced disparities. We found the same where we’re reducing copayments and cost-sharing for preventive health services and other chronic disease services, and I am very proud that the current administration in the White House has made equity enhancement a priority, and value-based insurance design is part of the strategies to make that happen.
For those of you who have never heard of V-BID before, I just want to tell you very briefly as we go through the speed-dating part of this presentation some of our policy accomplishments. I like to say, Dr. Freed, I published, and I still perished, but we took the V-BID idea and brought it to Washington in 2006 and had it incorporated into one of the most popular parts of that law, what many know as the preventive care mandate. That is a regulation that states that any service receiving an A or B rating from the US Preventive Services Task Force, or recommendation from the Advisory Committee for Immunization or certain preventive care and screenings from the Health Resources and Services Administration, or HRSA, these services must be covered 100 percent before the deductible by all federally qualified plans.
And this includes a couple of mental health issues that I added to this slide. For instance, screening for depression in the adult population must be covered 100 percent with no copay or coinsurance deductible. The same holds true for a recent recommendation of screening for major depressive disorder in adolescents between the ages of 12 and 18 years. And we’re hoping to see more attention paid by the US Preventive Services Task Force to the mental health issues brought up today and that as they get A and B ratings they will get covered.
And the way that Dr. Hodgkin said in the very first session, why are these cost-effective services not readily covered? Because cost-effective does not equal cost-saving. And if you buy more cost-effective things, you see Dr. Chisholm’s box grow in terms of the amount of money we have to spend. So that is really problematic, and my very short answer to why high-value services are not readily covered in the US is because they make total costs go up as opposed to total costs going down.
Some of you might know, and particularly, Bohdan, I’ll say to you and others who articulated the issue of HICV and AIDS, that PREP now has an A or B rating. It must be covered 100 percent, free deductible in all federally qualified plans. That doesn’t cover a lot of the ancillary services. You know, we can’t let the perfect get in the way of the good. But some of you might know that there is currently a case being seen right now that may overturn the preventive care mandate, and that worries me a great deal.
The good news is if it’s overturned, we have already published it and 80 percent of organizations may not rescind the coverage, but it makes me worry that the 20 percent that would rescind would impact the populations that we are worried the most about.
This slide pertains to a really interesting pilot going on in CMMI in the Medicare program that implements value-based insurance design in the Medicare Advantage Program, which now 5 million MA beneficiaries are enrolled, which includes the non-medical interventions like transportation, nutrition and telehealth. Very excited to implement telehealth before the Covid pandemic.
Hopefully, some of you know about the V-BID element in the Inflation Reduction Act. January 1st was good for the fact that the $35.00 insulin copay cap per month was implemented, as was the removal of cost-sharing for Medicare Part D vaccinations. So, for your patients over the age of 65 who were worried about paying out of pocket for the shingles vaccine, they now must get it for free. So, more and more examples.
What keeps me up at night are these high deductible health plans. We can’t cover certain chronic disease services because of IRS regulations. I thought this regulation — for 10 years, leaving the Trump Administration to put out a regulation to allow these plans to cover 14 services on a pre-deductible basis. You see we got one mental health service, SSRIs for depression, on that list. The great news is that a great majority of employers have made changes to their benefit design to make these 14 services more accessible. And there is congressional legislation to expand the pre-deductible coverage to include many more mental health disorders.
So where do we go from here? How do we pay for this? One is we increase premiums. That’s what I like, but that is politically not feasible. Second is we continue to do what we’re doing, tax the sick by raising deductibles and copayments. But I think what’s important that I just want to raise with this group and consider at the NIMH — and we can talk in the Q&A, Mike, about low-value care and mental health. Maybe we can rob low-value care Peter to pay more for high-value care Paul. I don’t know of any research underway right now looking at low-value care in mental health like we are in these other areas. I hope that happens soon and I am happy to be involved.
I am just going to take you through one last thing which I think is very important. This is all the low-value care thing which I am very inspired about but don’t have time to go into. It is all on the website.
I want to finish with this because I think this is really important. The way to move forward and, you know, make Dr. Hodgkin and Dr. Chisholm and everyone else happy is this idea of robbing low-value care Peter to pay high-value care Paul. You spend less on low-value care in red and more on high-value care in blue. That sounds easy but it is really hard from an actuarial and coding perspective, and the Trump Administration asked us to create a template for a V-BID plan for the exchanges; thus, V-BIDX.
We did this with about 40 payers and a bunch of actuaries, and we got this template that we published in the Federal Register that included a long list of high-value services that should be covered with zero cost-sharing, and on that list was PREP and also bupropion-naloxone, so we made a little bit of headway in terms of some of the services we talked about today. And to pay for them is some low-value services that we should buy less of, and are there any low-value mental health services that you think should be added so we can move the money around.
Some state exchanges, including DC, have put V-BID in place including pediatric and mental and behavioral health starting next year.
I will wrap up with this really important takeaway, which is how do we move forward. One is to continue to put pressure on payers to make these high-value services more accessible. When they ask you how are we going to pay for this, tell them that you are going to identify, measure and reduce services that don’t make Americans or people around the globe any healthier to pay for these incremental expenditures.
And the last thing, which ties to some of the earlier presentations, is if you are going to pay clinicians to do things more often, make sure it’s easy, not hard, for those patients to do it. And if you are going to benchmark and criticize clinicians for doing things less often, make it harder for your patients to do that. I am more than happy to team up with any of the mental health experts remaining on this call to create a mini-V-BIDX for mental health to be able to show folks this can be done.
Michael, sorry to go a few minutes over, I’ll turn it back to you. The last slide shows our website, and hopefully you all can get a link to the deck and other resources to learn more about what we do.
MICHAEL FREED: Thanks, Mark. I just want to comment. During my tenure here at NIMH and previously working for the DOD, I and others fought really hard to study, recommend and implement evidence-based and high-value services for people with mental illness and really discourage the delivery of those treatments and services that are untested or shown to be ineffective. Those of us who were fighting this battle know that some days we win and other days we lose.
So I really appreciate learning about value-based insurance design, and that is really the primary driver of any service that really helps incentivize the delivery of high-value care and stop enabling the delivery of low-value care.
We have a couple questions. First, the implementation of BVP models have been slow in behavioral health and the question is why, and how do we address barriers to payment reform specific to behavioral health.
A. MARK FENDRICK: These are very broad questions that I would need a whole day to explain. What we’re moving away from, thankfully, is classic fee-for-service, paying for whatever you do no matter how well it works, to incremental changes in payment to get more of the money flowing to the services that we should be doing more often. That is as far as I’ll go to that question.
We are seeing some of these alternative payment models that you know as well as I do include some mental health benchmarks. But I think what’s very frustrating from the provider side, is those examples where, for instance — I won’t speak to mental health, but let’s say something as a no-brainer like an eye exam for an individual with diabetes. I have been benchmarked on my eye exam for my diabetic patients for 20 years.
It turns out, Mike, that the coverage for a diabetic eye exam is worse in the United States in 2023 than it was when I started looking at diabetic eye exams in the year 2000, and that is largely because of deductibles. If you have a deductible that’s $1,000 and you have $500 in the bank and you have to spend a fifth of that on insulin, you can imagine how hard it might be for patients to get these ancillary diagnostic tests done.
I think the same holds true with the additional barrier that was presented on the non-medical issues for people with significant mental health disorders, which is why not only do we need to pay clinicians better and have benefit designs improve around quality, but we also have to leverage many of the interventions that the previous speakers talked about so that we really could create a model that actually puts the patient first.
MICHAEL FREED: Thanks, Mark. One of your slides is about the A and B recommendations of the US Preventive Services Task Force. I’m curious if you think removing cost-sharing is enough to help implement those A and B recommendations, or if more active strategies of like actually incentivizing strategies of screening and counseling –
A. MARK FENDRICK: I am going to try and speak quickly. This is a fantastic question. If there is anyone out there who still remains a skin-in-the game fan, which means people actually do better if you make them pay more, I think our mothers have proven that false. The question, of course, is how low should you go.
I will say that even though not everyone does things when they’re free — we had some brilliant behavioral economist earlier in the day who said if things are free they’ll do them more often. If they’re $50.00 they’ll do them more often than if it’s $500.00. In fact, we heard Kevin Volpp, who is such a leader in this field, say that sometimes we have to go beyond free, as we do in some prenatal care in Medicaid and some employers do, as you heard, with smoking cessation.
You know full well, Dr. Freed, that we made Covid vaccines free, and that is not enough. We have made 90 preventive services, including four cancer screenings and a bunch of mental screenings, free, and that is not enough. But our review, which I believe is on the page that we included in the chat, shows that if you remove barriers people use them more often. It’s often not as much as you and I hope.
But the last thing I’ll say which is really important is that the benefits of removing financial barriers are disproportionately incurred in a positive way by the populations we are all most worried about — low-income, multiple chronic diseases, and communities of color. And that is why I think you will see continued incremental implementations of value-based insurance design like a copay cap and like the Part D vaccines and other things moving forward because people understand that consumer cost-sharing is different than the core problem of prices, but it’s something that is really intuitive and just about every stakeholder can get behind as opposed to the controversies and battles faced by some of the stakeholders around total prices.
MICHAEL FREED: One other question. What advice would you give to mental health researchers who are seeking to identify or measure the value of mental health services?
A. MARK FENDRICK: Jennifer is lucky that she kept me on mute the whole day because I had comments on just about every single presentation. Here is the point I want to leave researchers with.
Value is a fraction, as you know as a PhD, and that gives great latitude for me to say something has value and you not to because there’s no pure definition. The issue that did not come that I would like to say is that, from the payer perspective — and this goes back to the very first presentations — they care much more about budget impact than they do about value.
If you have a valuable mental health intervention that involves a very few number of people regardless of its cost, I think you wouldn’t have those issues that we heard about in the early part of the day. But if you have a comparable mental health intervention to what’s coming down the tsunami for me around what to do about these effective weight-loss drugs, this is the problem, whether they are expensive or not, because one-third of Americans may need or want these drugs. That is why payers are worried.
Remember as a researcher the difference between value and — and I’m more than happy to have Jennifer connect me with people who want to learn a little bit more about that — but budget impact. The more high value even low-cost services that you buy, which may be the problem with expanding mental health high-value services, it could be more difficult to convey to a payer than just showing, oh, this is nearly cost-neutral. But if it doesn’t save money it’s going to be very hard in this environment of cost constraints to get anyone to pay for anything.
MICHAEL FREED: One final question. Shouldn’t budget impact also be a subject of economic research?
A. MARK FENDRICK: Absolutely. Neither an economist nor accountant — Tudget impact is much, much, much easier to measure than value. Again, I wish I had more mental health examples, and we talked about that this week, Michael. But I was one of the first people to say that the first hep-C drug at $78,000 was probably priced too low because if you do value calculations it falls well within the cost-effectiveness threshold. But because states like California, who I worked with, said, if we treated everyone with hep-C with these drugs at that price it would cost more than the entire public health education budget. So, automatically the conversation switched from value to budget impact even though people weren’t listening to me.
It’s very intriguing, and value is what we need to continue to work with. The frameworks are being discussed and the methods are moving ahead very quickly. It’s just when we see the frustration of the early presenters about how even when you show it is valuable it’s not covered, that is largely because of budget impact and not that you didn’t do a good job showing its value.
MICHAEL FREED: Mark, thank you so much. We will end here and I turn the microphone over to our conference organizers for a wrap up.
Agenda Item: Wrap Up
JENNIFER HUMENSKY: Thank you. To wrap up this conference, we want to thank everyone for coming to our first ever NIMH-NIDA meeting on health economics. We especially want to thank the presenters who shared their groundbreaking research and demonstrated how important it is to understand the role of economic concepts in order to improve outcomes for people with mental health and substance use disorders in the US and around the world.
We want to take a minute to thank some of the key people who helped put this meeting together including our technical experts through our contractor Bizzell, especially TaRaena Yates and Jonelle Duke and Amy Oskowski, our NIMH meeting team including Andrew Nawrot and our NIMH promotion team including Setareh Kamali.
We want to again thank our NIMH NIDA leadership without whose support this meeting would not have occurred, including at NIMH, Josh Gordon and Joel Sherrill and Mike Freed; and at NIDA, Nora Volkov, Carlos Blanco and Tisha Wiley for supporting this meeting and Leo and Sarah’s work in this area.
As we mentioned, the presentations from today’s meeting will be available on the NIMH You Tube page. It usually takes about six to eight weeks for the video to be processed and put up online. Please also be on the lookout for a special issue of the Journal of Mental Health Policy and Economics which will highlight economic-focused articles from this meeting as well as from our mental health services research meeting which took place last August. NIMH staff and NIDA staff will be providing commentaries on these articles through the special issue and will publicize that issue when it’s released.
Leo, Sarah and I are always happy to discuss new research ideas. Please feel free to reach out at any time to Sarah for ideas around substance use disorder research, to Leo for work that would be applicable to the NIMH Center for Global Mental Health Research, or to me for US domestic-focused mental health economics or methodological research.
Additionally, as this NIMH-NIDA Health Economics Collaboration is in its infancy we would love to hear from you. Did you find this meeting helpful? Would you like to see more of this type of meeting? Are there other types of meetings or other resources that would be beneficial as we help to support the health economics community? Please reach out to Sarah, Leo or I with your feedback.
With that, we can wrap up today’s meeting. Thank you for joining us and enjoy the rest of your day.