July 16, 2020
The results are in from a multi-year OCHIN study designed to deepen our understanding of how social complexity influences patient health outcomes. The new report, “The Impact of Social and Clinical Complexity on Diabetes Control Measures in Community Health Centers,” was published this week in the Journal of the American Board of Family Medicine, and holds significant clinical, health systems, and policy implications.
Led by OCHIN researchers, with support from the Patient-Centered Outcomes Research Institute (PCORI), the ADVANCE Health Systems Demonstration Project found that increasing social complexity is significantly associated with higher rates of poor diabetes control among underserved patients. This is according to an analysis of electronic health records data from ADVANCE Collaborative and OneFlorida CDRN, two large clinical data research networks. The new findings underscore the vital importance of addressing social determinants of health (SDH, i.e. non-clinical factors like housing, food, or job insecurity), in order to reduce national health disparities and improve individual patient outcomes. The findings also suggest that community health centers and other providers that care for populations with greater social complexity may benefit from having their performance metrics and reimbursements adjusted, particularly in an era of value-based payment.
Read on to learn more about the Health Systems Demonstration Project in our Q&A with OCHIN Investigator Erika Cottrell, Ph.D., MPP. Additional collaborators on this study include: Abby Sears, Jean O’Malley and Katie Dambrun (OCHIN), Brian Park and Jennifer DeVoe (OHSU), Elizabeth Ann Shenkman and Hongzhi Xu (University of Florida), Mary Charlson (Cornell University), and Andrew Bazemore (American Board of Family Medicine).
Q&A with OCHIN Investigator, Erika Cottrell:
ADVANCE Health Systems Demonstration Project
Tell us a little about yourself. What do you enjoy most about working at OCHIN?
I’m trained as sociologist and have a joint appointment as a research investigator at OCHIN and assistant professor at Oregon Health & Science University (OHSU). My primary area of research is in social determinants of health, so all my projects are united by thinking about the non-medical factors that impact people’s health – things like their social and economic circumstances, where they live, family support, level of education, and challenges meeting basic needs like food, housing, or financial stability.
I have always been interested in questions of health equity. Part of what drew me to OCHIN is that we’re embedded in real world situations. We have a unique opportunity to conduct studies based on the practical needs of providers in our network and contribute valuable data and insights to help inform policy discussions about how to reduce health disparities and improve health equity among underserved populations.
Can you share a brief overview of the ADVANCE Health Systems Demonstration Project study and its research goals?
In 2014, OCHIN received funding from the Patient Centered Outcomes Research Institute (PCORI) to create a research-ready data network along with 12 other clinical data research networks (CDRN) across the country. Led by OCHIN, the ADVANCE Collaborative is the only network comprised of community health centers (CHCs) that focus on treating historically underserved communities. As part of this work, PCORI asked CDRNs to submit proposals for Health Systems Demonstration (HSD) Projects designed to help health care leaders better understand how the many features of health care delivery influence patient health outcomes. In the first phase of the ADVANCE HSD project, we engaged our national network of health systems leaders and providers and asked, “What research questions are most important for us to explore? What are the most pressing issues that your health centers are facing?”
A research question that quickly rose to the top was understanding whether and how patients’ social complexity (or SDH measures) may impact health care quality metrics for community health centers (CHCs) and providers. In this era of value-based payment, there is more and more discussion about providers being paid based on their performance on a defined set of patient health outcomes or clinical quality improvement measures. In this context, there is often an adjustment for the clinical complexity of a patient population (i.e. comorbidity or multiple chronic diseases), but many of our CHC providers are also caring for very socially complex patient populations (i.e. those facing adverse social determinants of health). As a result, providers and health systems leaders asked, “How can we account for both of these things—not just the clinical side, but the social context as well?” During the second phase of the HSD project, we collaborated with the OneFlorida CDRN to answer this very question. This new paper describes the findings of that study, which explored the impact of both clinical and social complexity on diabetes control and whether accounting for social (in addition to clinical) complexity improves the assessment of provider quality metrics.
What data or evaluation metrics did you use in this study?
For this project, patient-level clinical complexity was assessed using the Charlson Comorbidity Index (CCI), a validated measure that predicts the risk of mortality and resource utilization for patients with a range of comorbid conditions. Figuring out how to measure social complexity was a bit more complicated. Unlike information on diagnoses, vital signs, and medical history, data on patient social factors has not been recorded routinely in electronic health records (EHRs). This is beginning to change as national health leaders recommend screening for SDH in health care settings and systematically documenting the results in EHRs. Numerous health systems (including OCHIN) have started SDH screening, but it will take time before sufficient data is available across a large population of patients. However, community-level data that provides contextual information on the conditions of the neighborhoods where patients live is readily available from public data sources (e.g. U.S. Census, American Community Survey). These data can be geocoded and linked to individual patients via their residential addresses.
For this study, we used a measure called the Social Deprivation Index (SDI), a composite measure of seven demographic characteristics: percent living in poverty, percent with less than 12 years of education, percent single parent households, percent living in rented housing unit, percent living in overcrowded housing unit, percent of households without a car, and percent non-employed adults under 65 years of age. Each patient was assigned an SDI score at the finest geographic resolution possible. A higher score indicates higher levels of social deprivation. The median SDI score in the U.S. is 50, whereas the median SDI for patients in the OCHIN network is 80, underscoring that OCHIN CHCs serve patients with higher levels of socially complexity relative to the general U.S. population.
What were some of the findings in the recent phase of this study?
A common performance metric is the percent of a provider’s patient panel with poorly controlled diabetes (defined as having an HbA1c score >9%). Although it is pretty easy to imagine that treating a patient with diabetes who has a well-paying job, access to healthy food, and stable housing is a much different task than treating a patient without insurance, with unstable housing, and who is experiencing food insecurity, we needed a good way to quantify this. In this study, we were able to demonstrate that even after accounting for clinical complexity, higher levels of social complexity were associated with a higher likelihood of poorly controlled diabetes. For every 10-point increase in the SDI score, the odds of poorly controlled diabetes increased by 5 percent.
For every 10-point increase in the SDI score, the odds of poorly controlled diabetes increased by 5 percent.
Mirroring the method that is used by the Centers for Medicare & Medicaid Services for adjusting provider performance metrics to account for patient clinical complexity, our analyst Jean O’Malley calculated provider performance metrics with an additional adjustment for social complexity. In particular, we explored what would happen if we assumed that instead of caring for patients with a median SDI of 80 (the median for OCHIN patients) a provider had a panel of patients with a median SDI of 50 (the national U.S. median)? Using this assumption, our models predicted that 25% of providers would have a 1-2% improvement in the assessment of their diabetes control performance metrics, 45% would have a 2-5% improvement, and 5% would have more than a 5% improvement in their assessed metrics. To clarify, it’s not that their patients’ health outcomes would necessarily improve, but the assessment of their providers’ performance would improve after adjustments to account for social complexity. Adjusting performance metrics for social complexity may help ensure that CHCs are not penalized for caring for more socially or economically disadvantaged populations, including the growing number of patients on Medicaid, and/or racial and ethnic minorities who have faced long-standing and systemic health and social inequities.
Did you encounter any challenges or are there any limitations in this study?
Systematic screening and collection of social determinants of health data is just beginning, so one limitation is that we relied on community-level SDH data. An area for future research will be to look at how reliable community-level information is compared to asking patients about their social risk and experiences directly. In terms of making policy and payment decisions about groups of patients, this type of community-level data is a good place to start; but it shouldn’t negate drilling down to what needs patients actually have and how to address those on a more individual or clinic-based level.
Based on what you’ve learned so far, what recommendations can be made about how to improve care delivery or payment among providers who serve more socially complex patients?
Our findings indicate that providers caring for patients with greater social risk factors may benefit from having their performance metrics adjusted for the social complexity of their patient populations. I think there’s an opportunity for really positive change, if the way we pay for health care shifts (i.e. by not penalizing providers who are caring for more complex patients and by giving them more flexibility to work with patients in different ways).
Some clinics, in Oregon and Massachusetts for example, have shifted to alternative payment models, where instead of tying reimbursement to the number of office visits with a billable provider, clinics are paid a per member per month capitated rate to provide care for a given population of patients. The idea is that removing the link between payment and volume of visits will enable clinics to focus on providing holistic, patient-centered care, rather than simply turning out more visits. For example, across the OCHIN network, health centers are exploring virtual care delivery and telehealth options for connecting patients with a community health worker or other services, so they don’t always have to come in for an office visit with a billable health care provider. These innovative solutions that don’t disincentivize treating uninsured or socially complex patients are even more important in the age of COVID-19.
What’s next?
There’s always so much more to uncover in our research. This is the start. For the third phase of this study, we partnered with Kaiser Permanente Northwest’s Center for Health Research because we wanted to compare our outcomes and findings to those of a well-insured population. A recent article in Street Roots describes this third phase in more detail. We’re now in the middle of analyzing and getting those results ready for publication in future papers. While the funding is over for this project, OCHIN plans to continue refining these models and leveraging our data, including patient-level SDH screening data, to advance research and inform the development of policies and programs to improve health equity.