Previous articles in this series introduced the importance of social determinants of health (SDoH) and explained how it can be captured and classified. In this article, I’ll show how many hospitals, payers, and other institutions are acting on that data.
Of course, the staff of a clinic or hospital don’t go around looking for apartments the way they actively look for a long-term care facility. Instead, their case managers partner with community institutions to help patients. But the clinic or hospital is a crucial linchpin in connecting the patient to this help. According to Dr. Victor Lee, vice president of clinical informatics at Clinical Architecture, the job is harder because most community institutions lack the sophisticated digital systems that are commonplace in modern commercial institutions.
According to Karl Ulfers, co-founder and vice president of growth at DUOS, governments offer excellent listings of resources with which care managers and clinicians can address housing, food, transportation, and other SDoH. For instance, Centers for Medicare & Medicaid Services has 50 data sets pointing to available help; states and the Department of Veterans Affairs has other data sets. These are all highly structured and can be pulled into the DUOS system programmatically, although Ulfers says they have to alter the language and convert it bureaucratese to human-friendly text.
Community-based programs also offer information about benefits on their web sites, although this is less likely to be structured and often calls for some screen scraping to retrieve.
Putting this information about resources together with information about patients’ eligibility, DUOS can offer members of Medicare Advantage programs resources to improve SDoH. This past Spring, DUOS replaced their traditional Elasticsearch-based search engine with a much better one based on a large language model from Open AI.
The Advantage programs all have supplemental benefits, although Ulfers says that few members know about the benefits, and fewer know how to trigger them until DUOS helps them. A typical benefit is an allocated number of rides to medical appointments per year.
SDoH in the Field
Two examples of clinical institutions putting SDoH to use today are the Medical Home Network (MHN) of Chicago and One Brooklyn Health.
MHN worked with ClosedLoop to find patients who needed assistance with basic health activities such as finding food and scheduling wellness visits. According to Andrew Eye, CEO of ClosedLoop, the process led to success with 90% of the patients they worked with.
One Brooklyn Health employed CipherHealth to automate outreach for an Equity Index Survey, with the goal of advancing health equity by amplifying the patient’s voice and reducing health disparities. The ten-question survey was created with the help of SUNY Downstate Health Sciences University, and was administered to inpatients, ambulatory patients, emergency department patients, and maternal care patients one day after discharge.
On their first round of outreach, 366 completed surveys were received out of a total 4,200 patients enrolled from an array of different locations within the system. Those answers helped to identify gaps in the system. Bolstered by this success, they have expanded the program, expecting to administer the survey to 10,000 patients with 1,000 completed surveys. With this data, One Brooklyn Health can segment their own patient populations, identify trends in the moment, and make care delivery changes based on accurate socioeconomic data.
Surgo Health collects not only the SDoH data we’ve seen so far, but people’s beliefs and biases about the healthcare system, their social norms, and other subjective information that affects their access to and use of healthcare.
Like other organizations in this series, Surgo Health conducts surveys and mines various data sets to inform their analyses. Dr. Sema Sgaier, founder of Surgo Health, points out that different factors are important for different conditions. For instance, vulnerability to COVID-19 was closely associated with housing density (with residents of nursing homes and prisons clearly suffering the most).
Surgo Health prefers direct contact with people, asking them whom they trust, whom they live with, how far they have to go to get treatment, etc. The company employs visits, phone calls, and online contacts to get the data. To determine what questions to ask, the company consults behavioral scientists.
Because getting this data is hard, the company also relies on geographic proxies to categorize groups of people. Results are analyzed through both standard statistical methods and machine learning to extend insights to larger populations. Machine learning creates predictions for people who were not in the original sample.
One of their reports links maternal health outcomes to vulnerabilities in the mother’s communities, with a focus on disparities between white and black women. They obtained data down to the ZIP code for this survey, which was conducted in Washington, DC. It turned out that lack of transportation was a major factor affecting maternal health, so they created a partnership with Uber Health and achieved very positive outcomes.
Surgo Health is working with pharma to achieve more diversity in clinical trials. Because drop-outs from clinical trials cause some of the biggest costs in pharma, the company tries to predict why subjects won’t participate: lack of trust in pharma, lack of expectations that they’ll benefit from the trial, difficulty traveling to the site, and so on.
Like many other vendors, Equality Health is used by primary care providers and others to make value-based care work. According to Dr. Sherri Onyiego, Medical Director for the Texas Market there, they can also help the providers feed SDoH data to payers.
Equality Health actually integrates their staff with the practices they serve, as well as their technology. The company looks at each site’s workflows and teaches the practices how to embed the portal into these workflows, aiming not to disrupt existing staff behavior.
The analytics are also useful to track improvements over time in performance and earnings.
mPulse Mobile says that it “offers digital engagement solutions that provide critical health resources and support to patients and members throughout their health journey while overcoming barriers, creating accessibility, and delivering better health outcomes at scale. A combination of plan data, mPulse’s internal and external data sets, and experiential data informs tailored intervention using omnichannel conversations powered by Natural Language Understanding (NLU).” Some of this data involves SDoH, such as indications of isolation. Therefore, according to Eden Brownell, director of behavioral science, mPulse Mobile is incorporating SDoH data into their service.
They use AI and NLP and work with clients to create effective auto-responders to patient inquiries. When an auto-responder is not sufficient, they can elevate the issue to a care manager or nurse.
Wellness Through SDoH
AdhereHealth is a platform for patient care that focuses on access to medicines, but also takes SDoH into account. During their initial, holistic interview with the patient, according to CEO Jason Rose, they ask about food, shelter, access to care, health literacy, and other important life factors—even about access to water, which is now a major issue in many parts of the U.S. and the world. They can assess the effects of SDoH on medication adherence and other aspects of health, find possible solutions, and call the patient back if the measures of health have not improved.
To me, AdhereHealth is an activist approach that contrasts with the wellness programs offered by many employers and payers. A typical wellness program gives you discounts and other inducements, but puts the burden on you to take care of your health. The program doesn’t actually help you overcome the SDoH barriers to health, but a program such as AdhereHealth can do so.
Rose says that AdhereHealth was a “pioneer” in using inputs from patients and providers, along with public and claims data, to take action on SDoH. They measure about 50 factors and re-evaluate the patient every day. Rose thinks that AdhereHealth’s analytics also improve the clinician experience by making this data easily available and turning up issues such as drug contraindications and problems with nonadherence that can lead to poor outcomes.
Avanade‘s platform for connection and relationship building in health can also identify SDoH and find solutions for transportation, food, translators, financial support, etc.
Lucem Health is starting to integrate SDoH data into its various AI-powered Reveal programs. Conrad Gudmundson, vice president of strategy, sees increased quality and quantity of SDoH data in the field, and plans to use it for such tasks as predicting malnutrition and onset or progression of diseases.
Jessica Probst, senior specialist, real world evidence at OM1, would also like SDoH to be considered routinely by regulators in making decisions. She says they are currently hesitant to use it.
The next and final article in this series will hone in on some specific types of SDoH addressed by the medical profession.
