Identifying and addressing bias in depression and anxiety quality measures

Kaiser Foundation Research Institute, Oakland CA

Investigators

Abstract

Abstract RFA MH-23-265 identifies a crucial need for validated outcome-focused quality measures to incentivize quality improvement, inform consumers’ choices regarding care providers, and motivate systematic assessment of outcomes in community practice. Quality measures based in patient-reported depression outcomes are now included in National Committee for Quality Assurance Healthcare Effectiveness Data and Information Set (HEDIS) “report cards” for healthcare systems and the Center for Medicare and Medicaid Services Merit-based Incentive Payment System (MIPS) program to adjust fee-for-service Medicare payments. Similar measures are proposed to assess and reward the quality of anxiety treatment. While outcome-based quality measures should reward more effective care, implementation must guard against delivering inaccurate incentives. Incentives intended to reward more effective care can have unintended negative consequences, penalizing those serving patients less likely to continue treatment or to improve during treatment Four specific aspects of existing depression measures may lead to inaccurate incentives: • Broad denominator definitions, including patients with widely varying treatment history and prognosis • Preference for measures of remission rather than proportional improvement or response • Focus on relatively narrow outcome windows as late as 12 months after initiating treatment • Considering all observations without recorded outcome scores to be treatment failures Data regarding current depression measures both demonstrate substantial differences between health systems serving different populations and raise concerns regarding specific design decisions that may exacerbate inaccuracies. While these problems could be addressed by detailed adjustment, benefits of more complex measures must be weighed against simplicity and transparency to measurement users. We propose to use data from 5 large health systems, including over 350,000 episodes of care for depression to evaluate inaccuracies in existing and proposed outcome-based quality measures. Specific aims include: Aim 1 – Identify patient characteristics that lead to different HEDIS/MIPS quality scores for those serving different patient populations by measuring to the populations they serve rather than the care they provide. Aim 2 – Adjust quality measures to account for differences in patient populations identified in Aim 1 and compare ranking from adjusted measures to existing HEDIS/MIPS measures. Aim 3 – Evaluate how altering specific aspects of existing depression measures can reduce inaccuracies in comparisons of care effectiveness without introducing unnecessary complexity. Aim 4 – Extend analyses for Aims 1-3 to proposed outcome-focused quality measures for anxiety. Aim 5 – In collaboration with a range of interested parties, develop proposals for improved outcome-focused quality measures for depression and anxiety.