Amid the COVID-19 pandemic, it has become more apparent than ever that community resources — such as access to food, safe and affordable housing, and green space — are critically important to people’s ability to stay healthy. There is heightened recognition that structural-level disparities existing across communities influence who becomes sick, who is able to social distance, and who has access to testing and treatment. Yet although it is generally recognized that where people live has serious consequences on their health outcomes, payers have not had clear financial incentives to invest in communities, largely because of questions around long-term return-on-investment.
Despite an abundance of evidence showing that the characteristics of a community affect health outcomes, there is a gap in evidence linking community factors to medical spending by payers. Melissa Sherry, PhD, director of Population Health Innovation and Transformation at Johns Hopkins HealthCare, sought to shed light on the relationship between community-level factors and medical spending in her recently completed doctoral thesis. The Center for Health Care Strategies (CHCS) spoke with Dr. Sherry, a member of CHCS’ Complex Care Innovation Lab, to learn more about her research and the implications of her findings, particularly in the context of the current COVID-19 crisis.
Q: What was the impetus for your research into how where people live impacts medical spending?
A: Through my experiences at Hopkins with special need and complex need populations, I noticed that there wasn’t a lot of information on how community-level factors relate to medical spending, even though we know that availability of food, housing, transportation, etc. influences peoples’ ability to prevent and manage disease. Further, convincing payers to invest in community resources to reduce medical spending requires evidence demonstrating the link between community factors and medical spending outcomes.
Q: What were the main research questions you were interested in exploring?
A: My first research question focused on methods for aggregating a large amount of existing community-level data. Using measures covering various community domains, I created a high-level community resource index that could be broken down into sub-indices (representing domains like crime, education, or housing). I wanted to show methods that are reproducible and usable in other community and research settings. I then used the aggregated community resource index to answer my second research question, which looked at the relationship between high, medium, and low resource communities and medical spending in a Medicaid population. My third research question explored which of the sub-indices — or different components of community resources — are significantly related to medical spending.
Q: What were the key results of your study?
I think my research points out something obvious: ensuring people’s basic needs are met by the communities in which they live may have benefits for everyone involved.
A: We found that consistently — even after adjusting for different variables used in models predicting medical spending — the community resource index was significantly associated with medical spending in the Medicaid population in Baltimore. While there is good evidence that where you live affects your health status, it was interesting that even when we controlled for morbidity, there were still higher levels of medical spending in low resource communities as compared to higher resource areas.
One of my goals was also to show that there’s an abundance of publicly available data out there that we can turn into indices useful for community-level research. We found that crime, housing, employment and workforce, and the overall community resource index were all independently significantly associated with medical spending, even after adjusting for other factors. Overall, I think my research points out something obvious: ensuring people’s basic needs are met by the communities in which they live may have benefits for everyone involved.
Q: What are the takeaways from your research for how to best support complex populations?
The U.S. has always struggled with a lack of investments in public health and social services as compared to other developed countries, and we continue to see how this leads to poor health outcomes and equity issues across populations.
A: We need to further investigate why community factors are linked to medical spending and use that information to build programs and policies that improve community infrastructure and resources. There have been a lot of interesting conversations going on in this area recently, with the Camden Coalition of Healthcare Providers recent randomized controlled trial not having the expected results earlier this year and with health disparities being uncovered by COVID-19 in the current moment. While the complex care field in the U.S. has made strides in connecting patients to social services, the Camden outcomes and, more recently repercussions from COVID-19, have exposed the problem to be much bigger than connecting individuals to services — particularly if those services aren’t even available in the first place. The U.S. has always struggled with a lack of investments in public health and social services as compared to other developed countries, and we continue to see how this leads to poor health outcomes and equity issues across populations.
One silver lining from the COVID-19 pandemic is that it has created a strong push for advocacy to make sure that the resources needed across communities exist and are accessible. How do we ensure better housing, adequate food supplies, and green spaces to safely exercise? My research highlights a need to think about the holistic circumstances in which people live when we want to improve health outcomes and reduce spending.
Q: If other health researchers or systems were interested in doing something similar, what are the key datasets they would need? Is it possible to conduct this research without access to claims and utilization data?
A: Communities have a lot more data than we think they do. However, a lot of it is not aggregated or formatted in a way that is off-the-shelf ready, so it requires effort to bring together. The American Community Survey data and census data are rich resources that can be used to measure community-level factors, and both are publicly available online. The Area Deprivation Index is also available online, and already aggregated at the community level. In my research, I was also able to use the Baltimore Neighborhood Indicators Alliance, which brings together a large number of data sources measured at a community level. Health care researchers should explore whether their communities have similar kinds of data resources that they can leverage.
It can be more challenging to get the outcomes data you need for a project like this because there is identifiable patient information involved. Putting the right protections in place to protect the data, keeping it on secure servers, and de-identifying everything to the extent possible (along with IRB approval) would make it more likely for health plans to be willing to provide data to researchers.
Q: How, if at all, did this work change your perception of where health care systems and plans should be investing time and resources to support complex populations?
A: One of the arguments we hear from payers regarding investments in community resources needed to protect and improve health is “we’re just a health plan. We can’t solve all of these problems.” And they’re right! My recommendation is to act as a convener that brings multisector organizations and community voices to the table to partner around improving communities. By facilitating the creation of platforms from which communities, health systems/payers, social services organizations, and public health entities can come together to tackle community improvements related to health and well-being, there is an opportunity to coordinate funding streams and create more efficiencies across sectors. And as my research indicates, payers may find that these efforts pay off by improving health outcomes and lowering costs for people with complex needs.
In terms of best investments, another exciting opportunity for health systems and plans is investing in community data exchanges. In Maryland, we have an incredible state health information exchange, but historically we have not placed emphasis on linking community and social data. There are groups across the U.S., like UniteUs, that have partnered on a large scale with states, health systems, and other funders to do some amazing work in creating community data exchanges that connect community resources to each other and to health systems and plans, creating a resource, referral, and tracking network for social needs. There’s power in these data to highlight gaps in the system and areas where advocacy and funding are needed to increase availability of services. These data can also be used by community organizations to advocate for themselves and increase funding streams to provide more services. Further, these data would be very valuable for researchers examining the links between community resource availability and health outcomes.
Q: Would the ability to link community and health system data help our ability to respond to a future pandemic?
Connecting public health and community resource data would allow a public health workforce to capture data on COVID-19 testing, contact tracing, and could include social needs assessments to determine needs for food or energy assistance that are required for individuals to effectively shelter-in-place.
A: Yes! These community exchanges could also have serious public health benefits. Connecting public health and community resource data would allow a public health workforce to capture data on COVID-19 testing, contact tracing, and could include social needs assessments to determine needs for food or energy assistance that are required for individuals to effectively shelter-in-place. Linking these public health data to a community network of resources would allow for instantaneous referrals to social service organizations that could support the needs of people sheltering in place, and would allow us to understand gaps in services available that put individuals at risk.