A toolkit for monitoring the effectiveness of interventions to subpopulations

A recent letter from the Centers for Medicare & Medicaid Services (CMS) to State Medicaid directors reminded me again of the critical importance of data-driven decision making and outcomes measurement to continuously assess what is working and what can be improved. This is critically important to Medicaid agencies and state budget writers, as state spending on Medicaid comprises an ever-increasing share of overall state budgets. As of 2020, this share was 19 percent across all states.¹

Across the country, Medicaid agencies and contracted health plans are pursuing innovative strategies to improve outcomes for their enrolled populations, including the use of in lieu of services or settings (ILOS) and services targeting unmet health-related social needs (HRSNs). Central to most states’ strategies are the use of federal waivers and leveraging full risk contracts with health plans. States, and by extension health plans, employ value-based contracting principles to signal priorities and help them achieve improved outcomes. They are leveraging data and analytics to support measurement of progress toward those goals. These and other levers have the potential to improve quality and/or the bend of the cost curve. However, progress may not be evenly distributed, and high level trends often do not reveal quality deficits and/or excess costs for certain subpopulations.

A November 2021 report done by 3M clinical and economic research outlines a measurement framework for evaluating health care equity. In this report, the authors evaluated the impacts of socioeconomic status (SES) on the performance of health systems. Using individual claim level data for fee-for-service Medicare enrollees and the SES component² of the Center for Disease Control and Prevention (CDC) Social Vulnerable Index (SVI), they evaluated the impact of county level SES on the risk-adjusted performance of health care delivery systems across nine outcome measures in four categories.³  

Category 1: Population focused performance measures⁴

• Potentially preventable emergency department visits (PPV) 
• Potentially preventable hospital admissions (PPA)

Category 2: Post-acute care performance measures⁵

• Potentially preventable readmissions (PPR) within 30 days post discharge 
• Potentially preventable return emergency department visits (PPRED) within 30 days post discharge

Category 3: Inpatient quality performance measures⁶

• Potentially preventable complications (PPC) during a hospital stay 
• Surgical mortality deaths within 30 days of inpatient procedure

Category 4: Service volume performance measures⁷

• Hospital admissions from the emergency department: Short, lower severity, non-surgical inpatient admissions from the emergency department 
• Admissions to a post-acute care facility: Admissions to a skilled nursing or rehabilitation facility within four days post discharge 
• Ambulatory visits: Any ambulatory visit with an evaluation and management code

The results of this study revealed a very consistent pattern:

• Areas of lowest SES had substantially higher than expected rates for measures in categories 1-3. Conversely, they had lower than expected rates for measures in category 4.
• Areas of highest SES had substantially lower than expected rates for measures in categories 1-3. Conversely, they had had higher than expected rates for measures in category 4.

Even though the study group was made up of Medicare beneficiaries, the authors provided a workable analytical framework for Medicaid agencies and/or health plans to identify health care inequities by leveraging a manageable set of broad-based and risk adjusted outcome measures. The study enabled views into actual performance relative to expected performance for each performance measure, which is critical to understanding underperformance as well as best practices. Using these key outcome measures, the authors identified a straightforward and systematic approach to targeting potential disparities, which can provide the foundation for development of data-informed interventions and/or investments.

Medicaid agencies can also stratify the analyses in different ways to shed light on potential access and/or quality deficits, by time period, geographic region, enrollee race/ethnicity and enrollee age grouping or gender. Augmenting this analytical framework to include information on enrollee clinical and social risk could provide a state Medicaid agency, social service agencies and legislators with a comprehensive and effective framework to evaluate health equity, quality outcomes and cost effectiveness of various interventions.

Matthew Ferrara is the state program manager for the regulatory and government affairs team at 3M Health Information Systems.

¹ Medicaid Expenditures as a Percent of Total State Expenditures by Fund

² SES includes these factors: below poverty, unemployed, income, no high school diploma

³ An additional measure-ambulatory potentially preventable complications (AM PPC)- could also be included to assess outpatient hospital performance.  This software was not available at the time of this study

⁴ Category 1 measures are risk adjusted using 3M Clinical Risk Groups

⁵ Category 2 measures are risk adjusted using 3M Discharge APR DRG with Severity of Illness Subclass

⁶ Category 3 measures are risk adjusted using the 3M Admission APR DRG with Severity of Illness Subclasses

⁷ Category 4 measures: Hospital Admissions risk adjusted using 3M Admission APR DRG with Severity of Illness Subclasses; Admission to Post Acute Facility risk adjusted using 3M Discharge APR DRG with Severity of Illness Subclasses; Ambulatory visits risk adjusted using 3M Clinical Risk Groups