Accounting for Socioeconomic Status in HEDIS Measures

Increased attention to socioeconomic status (SES) and its potential effect on receiving health services, raised questions about how to account for a health plan’s proportion of members with low SES when measuring quality. The Improving Medicare Post-Acute Care Transformation (IMPACT) Act of 2014 (H.R. 4994) directs the Secretary of Health and Human Services to study the effect of individuals’ SES on quality, resource use and other measures for individuals in the Medicare program.

The Centers for Medicare & Medicaid Services commissioned NCQA to assess whether Medicare Advantage plans that enroll a disproportionate number of vulnerable beneficiaries are systematically disadvantaged by programs that use performance measures to assess quality. Because previous studies suggested the existence of within-plan(1) differences between low- and high-SES beneficiaries, NCQA explored SES effects in select HEDIS^® Medicare measures.

To determine whether plan measure performance is sensitive to the proportion of low-SES enrollees, NCQA used its established process for reevaluating HEDIS measures. Our work was guided by a Technical Expert Panel on Socioeconomic Status in Health Care.

Our qualitative assessment included interviews with health care disparity experts about how to address SES in measurement and with health plans about their barriers and strategies for achieving good performance on their low-SES populations.

We also evaluated how the proportion of beneficiaries in a plan that is high- or low-SES affects its performance relative to similarly adjusted plans. We explored whether other factors may be responsible for any observed SES differences in performance.

We posted a proposed strategy based on the results of our work for wide public comment.

In the qualitative analysis, interviewees cautioned against using case-mix or risk adjustment for the measures, given the potential to mask differences in care between low- and high-SES populations. This concern was echoed by our advisory panels and those submitting public comments. The literature reviews suggested there was a conceptual basis for addressing SES in the measures except the two that assess whether a clinician asked about fall risk and physical activity.

In the quantitative analysis, adjusting for SES did not have a meaningful impact on results for most of the measures in our study. When adjusting for disparity in performance between low- and high-SES populations, plan ranks were not substantially impacted. When accounting for clinical and demographic factors, we found that low-SES beneficiaries were as likely, or more likely, to receive recommended care as high-SES beneficiaries.

However, an SES disparity did persist after accounting for clinical and demographic factors for four measures: Breast Cancer Screening, Colorectal Cancer Screening, Comprehensive Diabetes Control-Eye Examand Plan All-Cause Readmissions. NCQA proposed stratifying Medicare performance rates by beneficiary status for these measures.

We discussed results with independent advisory panels who agreed that NCQA should account for SES for measures for which a clear and consistent disparity existed. To further assess feasibility of this proposal, we modeled beneficiary-level stratified reporting in our testing data. Results suggested that stratified reporting can show within-plan differences given a plan’s beneficiary profile. We received broad support for this strategy in public comment. Generally, the commenters suggested this approach could increase transparency and allow for a more complete picture of performance among plans.

Overall, our findings supported accounting for beneficiary SES in measure performance for Breast Cancer Screening, Colorectal Cancer Screening, Comprehensive Diabetes Care—Eye Exam, Plan All-Cause Readmission (measure-specific conclusions are shown in the table below).

With transparency in mind, NCQA determined that stratifying Medicare plan performance can account for LIS/DE and disability status in measurement. Stratified reporting has the potential to highlight disparities, facilitate fairer plan comparisons, and provide a more complete picture of performance on key health care services.

NCQA proposed stratifying Medicare plan performance by beneficiary LIS/DE or disability status, for Breast Cancer Screening, Colorectal Cancer Screening, Comprehensive Diabetes Care – Eye Exam, and Plan All-Cause Readmissions in HEDIS 2018. NCQA proposed that plans use the CMS Monthly Membership Detail Data File to get the data and calculate the strata.
Medicare plans would report four mutually exclusive performance rates and an aggregate total rate for each measure:

• Beneficiaries who have neither low-income subsidy, dual eligibility nor disability status.
• Beneficiaries who have low-income subsidy or dual eligibility status only.
• Beneficiaries who have a disability status only.
• Beneficiaries who have low-income subsidy, dual eligibility, and disability status.

We conducted additional testing to ensure that the proposed files and data were available, and that the calculations were appropriate for the strata selected. In the testing process, NCQA partnered with HEDIS auditors, vendors and Medicare Advantage plans that report quality measures. We recruited four plans to conduct feasibility testing. We started with informational interviews and confirmed that the plans receive the Monthly Membership Detail Data Files from CMS, and that the files contain fields to identify the beneficiaries. We also tested the proposed logic to confirm it was correct and usable.

Based on our work with vendors, auditors and health plan data, NCQA confirmed that the plans will use the Monthly Membership Detail Data Files, but we made further revisions to the logic for plans to calculate the SES strata in the four HEDIS measures.