Collecting SDOH Data Can Assess Risk of Medical Nonadherence, Improve HEI and Star Ratings


At the Academy of Managed Care Pharmacy (AMCP) 2024 annual meeting, a panel of presenters explored changes coming to Medicare that incorporate social determinants of health (SDOH) data to improve patient and health system outcomes.

A discussion around health equity and social determinants of health (SDOH) demonstrated the significance of these features in the context of Medicare’s Star Ratings and risk adjustment in health care, among other measures such as Z codes, during a panel at the Academy of Managed Care Pharmacy (AMCP) 2024 annual meeting.

Health Equity Word Cloud | image credit: Colored Lights -

Health Equity Word Cloud | image credit: Colored Lights -

CMS defines health equity as “the attainment of the highest level of health for all people.”1 Many factors culminating from an individual’s upbringing, environment, age and more, known as SDOH, can have profound impacts on one’s ability to achieve equitable health. Furthermore, achieving this equity can be further complicated by social risk factors (SRFs) such as poverty, education level, access barriers, and others—which are out of a provider’s control and unrelated to a patient’s quality of care.

To unpack the implications of these features, a group of presenters including Sujith Ramachandran, PhD, assistant professor of pharmacy administration and assistant director for the Center for Pharmaceutical Marketing and Management at the University of Mississippi School of Pharmacy; Lisa Lewerer, client quality principal, Prime Therapeutics; and Ben Urick, PharmD, PhD, health outcomes researcher, senior principal, Prime Therapeutics, led a session contextualizing disparities in data within the Star Ratings landscape (a measure of the performance of Medicare Advantage and Part D plans) and Health Equity Index (HEI).

From a policy perspective, Lewerer spoke to the changes taking place through the CMS and how it, from a provider data management (PDM) standpoint, can impact this work, starting with data and analytic information.

The collection of this information lays the foundation, Lewerer explained, for work in health equity because it sheds light on where inequities and disparities exist and tracks the influences that CMS efforts have on impactful change. The second priority is evaluating the causes of disparities within CMS programs to combat the operations and policies contributing to inequity. The third priority is centered around investing in diverse and bilingual staff and training to ensure the workforce better reflects the population it is serving. With this priority in mind, Lewerer emphasizes the overlap between member experiences and equity, especially in cases where language services can be applied to better facilitate care with a patient. The remaining priorities focus on efforts to increase member literacy and reduce barriers the impede individual access to resources.1

CMS has chosen SRFs, such as low income, dual eligibility, and disability status, for its HEI due to the persistent health disparities that arise and have widened in these areas.2 One of the changes taking place in CMS stems from the inclusion of these SRFs as they are going to impact scoring of the HEI. Index scores span from –1.0 to 1.0 per measure (total perfect score = 18.0) and are calculated with reference to the proportions of enrollees with targeted SRFs.

Another CMS change will come in the form of risk adjustment for adherence measures, beginning in 2026, for the Star Ratings of 2028. This change will include quality measures for the hypertension, diabetes, and hyperlipidemia categories, as well as add age and gender. Additionally, HEI will be implemented into Star Ratings in 2027.

Ramachandran continued the discussion by diving deeper into the data on risk adjustment. He coauthored a paper that found the majority (> 80%) of outcome measures submitted to the National Quality Forum (NQF), had a risk adjustment to the statistical model but 72% of process measures had no risk adjustment process between 2018 and 2019.3 "So it's very dependent on the type of the measure we are looking at,” he said.

An analysis conducted by NQF from 2017 to 2020 revealed risk adjustment errors occurred in 39% of all quality measures.4 Furthermore, while there was a conceptual rational for adjusting for an SRF in 38% of measures, only 1% applied this adjustment.

“So, these risk factors like insurance, race, or ethnicity and education were often considered for inclusion by measure developers, but were not actually adjusted for," Ramachandran said. "And the reasons that the quality measure developers used in their justifications were things such as small effect sizes in significant coefficients, complex causal pathways, lack of data, so on and so forth.”

He then discussed the HEI, how it is calculated, and the measures that influence its scores in the Star Ratings program. Ramachandran noted how all measures are included in the Medicare Part C and Part D Star Ratings program; however, the measures themselves change year-to-year.

First, the eligible measures are used to calculate scores in the health plan, but only for enrollees who have one of the SRFs. A case-mix adjustment is then performed before revisiting the eligibility of the included measures and checking for denominator requirements and reliability (needs a score at or above 0.7 at this step). Then, points are assigned based on tertiles for each contract and then those are put together with Star Rating measure-weights to calculate a final sum. This process is completed for every single measure.

A reward ranging from 0.0 to 0.4 can be applied to contracts with an overall positive HEI score or a percentage of enrollees with an SRF that equates to half the median figure for all contracts. The potency of the reward in this range depends on the proportion of enrollees with an SRF. Data from 2020-2021, Ramachandran noted, revealed that almost 42% of Part C contract enrollees were identified with an SRF, compared with just 14% in Part D.5

Urick built on Ramachandran’s talk on SRF to contextualize sources of SDOH-related data, Z scores, and nonadherence scores in health plans. Administrative claims, member files, geospatial data, and patient and provider reported data constitute the 5 resources for gathering forms of SDOH information.

An International Statistical Classification of Diseases, 10th Revision (ICD-10), Z code is an administrative source of structured SDOH information that can be found on the diagnosis field of a claim. The fact that Z codes are included on a medical claim is one of the main benefits they provide for identifying members with SRFs.

“If you look sort of far enough in the electronic information sent from pharmacies for adjudication, you can find ICD-10 codes on pharmacy claims, and you can actually see Z codes on pharmacy claims. So, we've been able to look at that within our data as well,” he said. "So, the accessibility of Z codes is really, really high. And they're easily therefore recorded into broader managed care reporting efforts."

He goes on to explore the cons associated with Z-codes, which include inconsistencies that can occur between pharmacies and the fact that they cannot provide billable information; however, common Z codes are linked to and can identify SDOH features, such as education, employment, social environment, and more, which can help a clinic provide care.6

"[Z codes] are a nice to have, but they're not a need to have," Urick said.

In the context of Medicare Part D, member files can outline a number of demographics, including age, gender, race/ethnicity, low-income subsidy status, disability status, and dual eligibility. This resource is readily available and highly predictive of SDOH, he adds, but has limited use for commercial lines of business and there need to be careful considerations when it comes to risk adjustment.

Geospatial data is related to member files and access through a member’s address. These sources come from measures such as the American Community Survey (ACS), which collects data on numerous SDOH factors such as education, housing, racial makeup, etc; the Social Vulnerability Index (SVI), which incorporates ACS data to assess household characteristics, racial and ethnic minority statuses, housing type and transportation; and the Area Deprivation Index (ADI), which also utilizes ACS data when considering factors like employment, housing quality, and more.

Patient- and provider-reported SDOH data is also useful; however, Urick points to barriers in this collection that create considerable challenges despite the development of various screening tools: “But taking information from these tools, transmitting it over to a health client side, standardizing it in a way that actually margins the claims. And then making that through useful is an incredible challenge.”

With these sources available, Urick posed the question as to whether or not SDOH information can be used to target Medicare members and improve adherence scores. To explore this, Urick and colleagues conducted a study to evaluate the extent to which SDOH factors correlate with medication adherence in the hopes of developing a risk score related to nonadherence outcomes. This was a retrospective study that incorporated a plethora of the previously mentioned SDOH variables. A linear probability model was used with dichotomous adherence constituting the dependent variable across adherence measures in diabetes, hypertension, and cholesterol medications to create a summary score that could be applied across these variables. The generated scores, they found, proved to be applicable across adherence measures and beneficial identifying members that had nonadherence. Notably, age was a stronger predictor of this outcome, whereas disability or low-income subsidy status were less predictive.7

“This is something that clients can actually use to try identify members’ risk for nonadherence, how to solve problems with members, and improve performance in those specific scores—the HEI and the risk-adjusted score that are now being rolled out,” Urick concluded as he reemphasized the value of using historical data to improve, for example, Star Ratings moving forward.


1. CMS Framework for Health Equity. Centers for Medicare & Medicaid Services. 2022. Accessed April 18, 2024.

2. Disparities in Health Care in Medicare Advantage Associated with Dual Eligibility or Eligibility for a Low-Income Subsidy and Disability. Centers for Medicare & Medicaid Services. May 2023. Accessed April 18, 2024.

3. Ramachandran S, Maharjan S, Nsiah I, Urick BY, Carr A, Foster M. Review of the national quality forum's measure endorsement process. J Healthc Qual. 2023;45(3):148-159. doi:10.1097/JHQ.0000000000000378

4. Social Risk Trial Final Report. National Quality Forum. July 2021. Accessed April 18, 2024.

5. Centers for Medicare & Medicaid Services. Medicare Program; Contract Year 2024 Policy and Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit Program, Medicare Cost Plan Program, Medicare Parts A, B, C, and D Overpayment Provisions of the Affordable Care Act and Programs of All-Inclusive Care for the Elderly; Health Information Technology Standards and Implementation Specifications; 2022. Accessed April 18, 2024.

6. Using Z-Codes: The Social Determinants of Health (SDOH) Data Journey to Better Outcomes. Centers for Medicare & Medicaid Services. June 2023. Accessed April 18, 2024. 

7. Thompson K, Urick B, Quam C, Gleason PP. Medication nonadherence social determinants of health score: development and use. Poster presented at: ACMP 2024; April 15-18; New Orleans, LA.

Related Videos
Monica Kraft, MD, ATSF.
Michael Arzt, MD.
Julie Linton, MD, FAAP.
Jan Hedner, MD, PhD.
Dr Migvis Monduy
Kimberly Westrich, MA, chief strategy officer of the National Pharmaceutical Council
Amy Shapiro, MD
Adam Benjafield, PhD.
Paul Frohna, MD, PhD, PharmD.
Debra Boyer, MD, MHPE, ATSF.
Related Content
CH LogoCenter for Biosimilars Logo