The final panel session of the workshop addressed research gaps and how they might be filled. The four panelists were Caroline Fichtenberg, Managing Director of the Social Interventions Research & Evaluation Network (SIREN) at the University of California, San Francisco (UCSF) (who presented slides developed by Laura Gottlieb, SIREN Director and Associate Professor of Family and Community Medicine at UCSF); Kelly Doran, Assistant Professor in the Departments of Emergency Medicine and Population Health at the New York University School of Medicine; Genevieve Kenney, Co-Director and Senior Fellow at the Urban Institute Health Policy Center; and Timothy Waidmann, Senior Fellow at the Urban Institute Health Policy Center. Daniel Polsky moderated an open discussion following the short presentations by the panelists.
SIREN, explained Caroline Fichtenberg, is a research center focused specifically on building the evidence about how health systems can address social risks. She described the framework SIREN has been using
to think about the different types of health sector activities related to social risks; the framework includes five categories, three of which focus on clinical care and two that look at community-level activities. These categories comprise the charge to the National Academies of Sciences, Engineering, and Medicine’s Committee on Integrating Social Needs Care into the Delivery of Health Care to Improve the Nation’s Health.
The framework starts with awareness, which includes ways the health care delivery system increases awareness around social risk factors. Next comes adjustment, which refers to the way in which providers and care teams modify care to mitigate the effect of the social risks and the barriers to health. The third category, assistance, focuses on activities that help patients connect to resources that address any social needs and any barriers to getting help with those needs. Alignment and advocacy activities refer to using the full range of health care resources to help improve social conditions at the community level.
SIREN groups evidence across these categories into effectiveness and implementation research. Effectiveness research, she said, is about what works, why, and for whom. Implementation research is about understanding how best to encourage adoption of these effective interventions.
Turning to awareness, the first group of activities, Fichtenberg said one of the big evidence gaps is understanding how to best identify needs and the patients who can benefit from the types of interventions available in the health care sector. This evidence gap exists, she added, despite the fact that there is a great deal of interest and work occurring around measuring social risks in health care settings. Researchers at Kaiser Permanente-Washington looked at the validity testing that had been done for 21 multi-domain social risk screening tools and found that the median number of quality validity tests conducted was two out of a possible eight (Henrikson et al., 2019). “That tells us there is more work that we need to do to identify valid and effective ways to screen,” said Fichtenberg. Along the same lines, Stacy Lindau and her colleagues have shown that even small changes in the widely used Hunger Vital Sign instrument for measuring food insecurity changes the percentage of patients who screen positive (Makelarski et al., 2017).
One key question emerging in this area, said Fichtenberg, is how to identify patients who are interested in assistance and will benefit from it. A number of studies on social risk screening have found the numbers of people who report specific social risk factors on the screening tools is substantially greater than the numbers of people who are interested in receiving help related to those risk factors, and it is not clear why that is. It is possible, she said, that patients who screen positive but do not want help might not prioritize that social risk factor in the context of other competing stressors or priorities. Other possibilities may be that the screening
tools are too sensitive, that patients do not think the health care system can help them with these needs, or that they do not want help from their care team. What is clear, based on consistent findings from multiple studies, is that current screening tools identify a large number of people who are not interested in help from the health system.
Another area needing research concerns the patient acceptability of health care–based social risk screening, including whether screening might stigmatize patients or lead to response-based discrimination, said Fichtenberg. Some suggested that social risk screening should occur only in the context of an ongoing relationship with a case manager, community health worker, or patient navigator. However, Fichtenberg and her colleagues have completed a study involving 1,000 patients at 10 sites, the results of which will be available in fall 2019, which found patient acceptability to be high. Participants voiced some concerns that they were slightly less comfortable with information about their social risks being stored in the electronic health record.
In the area of implementation, adoption, and sustainability of interventions, Fichtenberg said more research is needed on how to make data collection feasible and sustainable for different health care organizations. Studies here should focus on understanding how health information technology can support implementation, adoption, and sustainability and what kind of training is needed to help the clinical workforce develop the competence and comfort to screen patients for social risks. Research is also needed to understand how policy and payment models can incentivize implementation, adoption, and sustainability once these other questions are answered.
There are many information gaps regarding adjustment activities, largely because they have received little attention from the research community. One open question, for example, has to do with how, for whom, and when health care organizations should individualize care based on information about non-medical, health-related social needs and risks. One recent study in this area did find that providers who were given access to data about social risks reported they had changed how they provided care; however, more research is needed. Research is also needed to examine the effects of care adjustment activities on the experience of care, health outcomes, health equity, and utilization. It is particularly important, she noted, to ensure that any information collected is not used in any way to discriminate or lower standards of care.
In the area of assistance, the SIREN team reviewed nearly 5,000 papers published between 2000 and February 2017 on social needs interventions (Gottlieb et al., 2017b). Of these, only 67 evaluated a social risk-related intervention and were designed to isolate the effect of those intervention components. Among the 67, only 30 percent examined health outcomes,
20 percent examined utilization and prior outcomes, and only 10 percent were judged to be of high quality.
One of the biggest gaps in knowledge about effectiveness, according to Fichtenberg, concerns how these interventions affect health, for which patients, via what mechanisms, and in what contexts. She has also found no studies that look specifically at the effects of an intervention on health equity. The assumption is that these interventions often focus on the most vulnerable patients and so they will reduce health inequities. “We know that if you are not specifically trying to improve health equity, you can, in fact, not have an impact on health equity or [you can] have the opposite impact, so it is important to look at that outcome,” she explained.
Research on the effect of interventions on utilization and health care costs is generating evidence, and Fichtenberg noted that the Commonwealth Fund is in the process of finalizing a review of the literature that identified 50 studies in this area. Many questions remain, however, because of the heterogeneity of the interventions tested and the outcomes measured. She added that it is also important to look at patient and provider experience of care, which have cost and quality implications. She noted there are a few studies showing that perceived organizational capacity to address social determinants of health in primary care settings is associated with lower provider burnout, which could also affect health systems’ bottom lines.
Once the available evidence shines light on interventions that work to address social needs, there will still be a need for research on how to scale effective interventions, said Fichtenberg. This research should look at what supports are needed and whether health information technologies, training, policies, and payment incentives can help drive the adoption of effective interventions. Another area of needed research concerns how these interventions will affect social service organizations.
In summary, said Fichtenberg, interventions directed at social risks are promising but have not yet been studied sufficiently. Effectiveness studies should include a wider array of outcomes, and in particular, should focus on health and health equity outcomes. Finally, she added, implementation science studies are needed to understand how to best translate effective interventions into real-life practice settings.
Kelly Doran began her presentation by acknowledging that the “tidal wave of interest” in addressing social needs through the health care setting is exciting and that many of the individuals at the workshop were conducting the research needed to address the gaps she was about to discuss. The first gap, one that Fichtenberg discussed, regards research
on developing, validating, and standardizing the tools that researchers should use to measure social determinants of health and social needs. “Having a good set of standard tools would allow us to better compare our results across studies, and it would also be helpful for researchers for whom social determinants are not their primary interest,” said Doran. “Maybe they are doing cardiovascular disease or diabetes research but would be able to take a set of measures if they were easy and incorporate them into their studies.” She noted that the National Institute on Drug Abuse’s Clinical Trials Network has a set of common data elements, available online, recommended for use by researchers interested in substance use disorders.
By and large, said Doran, most studies on social needs in the context of health care remain cross-sectional and come with challenges, one of which concerns confounding variables. For example, a cross-sectional study might find that food insecurity is associated with bad health, but food insecurity may just happen to travel along with housing instability, which could be the actual variable associated with bad health. If the study design does not adjust for that association, the conclusion will be incorrect. Solving the problem of confounding variables requires more information on the true social needs that drive health outcomes, the pathways through which those drivers are affecting health, and potential mediators and moderators of those pathways.
Another shortcoming of cross-sectional studies is they cannot provide much information about causality or directionality. Longitudinal research would help with the directionality issue, said Doran, and there has been some work along those lines using administrative data sets, though many of the administrative data sets commonly used by health services researchers lack robust social determinants data. There are, however, ways of combining health services data with data from other sources that could provide useful information, she noted. For example, she collects patient data at a baseline time point on social determinants and gets consent from the patients to link their data going forward with administrative health records to see what the effects those social determinants have on downstream outcomes.
Doran reiterated Fichtenberg’s call for more intervention research. “We are at the point now where we need more information on what works,” said Doran. To date, most of the research on interventions has examined process and social need outcomes, with less work on downstream health outcomes. To give a parallel biomedical example, studies of a statin drug would have an intermediate outcome of a reduction in serum low-density lipoprotein cholesterol levels, but the ultimate outcome would be how the drug affects cardiovascular health or mortality. Without the latter outcome, the former does not necessarily have any
intrinsic value. In contrast, in the social needs area, intermediate outcomes can have intrinsic value. For example, if an intervention addresses housing insecurity, that alone has benefit even if the desired final outcome among those in the health sector is better health outcomes.
On a final note, Doran recognized that it is important to involve experts and stakeholders from sectors outside of health in the design and conduct of research. In addition, she believes that the health care sector needs to do more thinking about what its endgame is. For example, if all of this work results in more money going to the health care sector so that interventions addressing housing and food insecurity need to be funded through health care because that is where all the money is, that is not really a win, said Doran. “I think we can invest in health care system–based social interventions, and I do think there is real value there, but how do we strike that balance with investing directly in social service systems?” she asked. “We need to do more thinking on what the best role is for us as the health care system.”
Genevieve Kenney and Timothy Waidmann started their work on data and research gaps by assessing the role that Medicaid and Medicaid-managed care plans play in helping enrollees address their non-medical, health-related social needs and then identifying the driving forces behind the activities and investments that are occurring. As part of this effort, they conducted key informant interviews with Medicaid officials, managed care plan representatives, academics, national policy experts, and other key stakeholders from around the country. “We found a veritable explosion of innovation and activity, and many exciting examples have been shared today,” said Kenney, “but we also found consensus that rigorous evidence is sorely lacking to guide investments in policy development.”
Noting that Fichtenberg identified a number of unresolved issues and pressing research needs with respect to the specifics of the who, what, and how of screening for social needs, as well as on the effects of plan and provider screening for those needs, Kenney added that there is a need for a systematic assessment of the extent and nature of unmet health-related social needs at a population level for different subgroups in Medicaid and Medicare and writ large. “An analysis that gives us the grounding on where the major holes are in our safety net and where investments in other safety net programs is needed,” said Kenney.
In the area of housing, only one in five people who meet the eligibility criteria for housing subsidies can actually get them given the constraints on public resources. This, said Kenney, raises fundamental questions
about how much payers and plans can do to help people address their housing needs, as well as other unmet social needs, if those funding constraints are not addressed. Moreover, if addressing how unmet social needs, individually and in combination, affect health is not challenging enough, it is important to understand the negative effects that unmet health-related social needs have on sectors outside of health care. This is where the “wrong pockets” problem comes into play consistently, which argues for the importance of understanding how addressing specific unmet health-related social needs affects different outcomes that matter over time, as well as how they affect spending across sectors and different levels of government, said Kenney. Given that current government budgeting time frames are often limited to a single year, understanding the medium- and long-term implications of these interventions will be critical if government is going to incentivize investments that have significant albeit longer-term payoffs over the intervention life cycle.
Kenney noted that the current approach to evaluating interventions to address social needs undervalues investments that produce positive effects and savings across multiple sectors and across different levels of government. Investments in systems that combine administrative data from different sectors such as Medicaid, child protective services, criminal justice, education, housing assistance, and income supports—an approach described earlier by Ross Owen—can provide information about the extent to which the same individuals are touched by more than one program and potentially more than one caseworker. They can also support assessments of how interventions in one sector affect cause and outcomes in other sectors, she said. This is an area that could use more research, both in terms of how such systems get launched and how they are maintained. It will also be critical to share knowledge and experiences regarding the design, implementation, and maintenance of these systems as a means to encourage more cross-sector approaches to investing in interventions.
Kenney suggested another area where the evidence base is scant and where there is a need to understand trade-offs. The issues concern payers and the designation of specific services that plans can cover and be reimbursed for versus letting plans decide for themselves how to use a flexible pot of money to meet the needs they observe in their clients. Data are also needed to guide how targeted investments have to be to create a sustainable, convincing business case for payers and plans. An important question here is whether substantial health care sector investments in addressing social needs will mean consistently deprioritizing investments that yield benefits and savings for other sectors or that take time to materialize. In particular, she is concerned that there is a real risk that investments in children will be deprioritized. She asked in her concluding comments:
To the extent that health payers and plans do focus on helping their enrollees meet particular social needs that are pressing, is there a risk that other population groups will see their access to those services decline, and could it undermine the ability of payers and plans to meet the medical needs of their enrollees?
Seth A. Berkowitz opened the discussion by asking the panelists to talk about the changes needed to the current process of funding research. Doran said that if the goal is to have rigorous research on social needs and how to address them, the field needs rigorous funding similar to what other health and medical research gets. One impediment is that funding from the National Institutes of Health is organized largely by disease or body systems, and that is not how social needs work. There is also a need to bring people together to develop consensus guidelines around what measures researchers should be using and which outcomes should be prioritized, particularly given that these are the early days of social needs research so it would be good to get everyone on the same page now, said Doran.
Timothy Waidmann said in addition to learning more about what works in a clinical sense, there is the need to understand what works in terms of implementation. For example, the model developed by Len Nichols and Lauren Taylor that treats social determinants as a public good as a means of generating the will to finance healthy communities has theoretical promise and the support of many economists, but there needs to be research to see whether this approach is feasible (Nichols and Taylor, 2018). He noted that this approach could address the “wrong pockets” problem.
After saying she agreed with both Kenney and Waidmann, Fichtenberg said that more studies that included the same outcome could accelerate learning in this field and allow comparative effectiveness analyses. “This is something that funding agencies could help encourage or even require,” she said. What would also help advance the field more rapidly would be for researchers to synthesize and share results. This is something that she and her colleagues at SIREN think about, but they have yet to come up with a great way of capturing lessons learned across the field, particularly regarding implementation. Waidmann offered one big concern:
We think we can solve one problem, but if the real cause of a disparity in health outcomes between people with and without social needs is not the social need itself but the lack of resources to fill needs on one’s own, we are not going to solve the health disparity problem by addressing one type of unmet need—even with a very targeted, well-designed,
and approved intervention because something else is just going to come along and rise up to take its place.
Doran commented that one reason to study return on investment to the health care system for interventions having to do with housing or food insecurity is that it might sway policy makers and others who believe in the “pull oneself up by one’s bootstraps” philosophy and do not support those types of assistance on their own merit. “I think that for different parties you need to make different arguments, and different things are going to move and drive different people,” she said. In the same vein, Fichtenberg said that she and her colleagues at SIREN are increasingly turning their attention to research showing the important role the health care sector plays at the community level. She noted that there are a number of health care organizations across the country that are looking at ways they can use their reserve dollars to invest in economic development and affordable housing. Health care organizations can also be strong advocates at local, state, and federal levels for changes to address the underlying social inequities that are creating the health-related social risks that contribute to escalating health care spending.
Ellen Marie Whelan from the Centers for Medicare & Medicaid Services (CMS) commented that many states are using their existing Medicaid authorities to provide home- and community-based services, nonemergency transportation, and food. She wondered if there was a way CMS could play a role in linking these efforts to researchers who could help the states figure out what is working and what is not and determine how to scale successful interventions. Fichtenberg replied that SIREN would be on board for that and would be more than happy to work with CMS to bring about that kind of connection. Kenney was also enthusiastic about the idea and suspected philanthropy might be interested in supporting that type of collaboration between practitioners and researchers. Doran pointed out that the best time to get a researcher involved is before the intervention is rolled out, and she also commented that there is a role for some entity to get the knowledge that researchers have into the hands of the states who are making Medicaid and other policy decisions.