High-Quality State Data Needed to Better Address Preventable Hospitalizations and Other Health Care Inequities Among Adult Medicaid Enrollees 

Hospitalizations are stressful and costly health events, but they can often be prevented if patients have access to timely, high-quality outpatient and preventive care. A recent report by the Urban Institute provided new data on preventable hospitalization rates among nonelderly adults with Medicaid coverage. The analysis examined differences in preventable hospitalization rates by race and state and how preventable hospitalization rates differ by whether people receive Supplemental Security Income (SSI), a program providing monthly payments to people with disabilities and that automatically qualifies individuals for Medicaid in many states. The analysis was limited to 21 states because of a lack of high-quality data in the other states.  

In this blog, we review the findings and show how high-quality state-level data can be used by researchers and policymakers to identify and address health care disparities. We also discuss bright spots in North Carolina’s data to highlight opportunities for other states to similarly improve their Medicaid data.  

Research Highlights Disparities in Hospitalizations  

Using data from the 2018 and 2019 Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF) from the Centers for Medicare and Medicaid Services (CMS), the Urban Institute team examined preventable hospitalization rates among Medicaid-covered adults for the three most common types of preventable hospitalizations: asthma/chronic obstructive pulmonary disease (COPD), diabetes, and heart failure. These data, which are nationally standardized for research purposes, include enrollment and demographic information, health care claims, and managed care encounter records for all Medicaid/Children’s Health Insurance Program (CHIP) enrollees in all US states and territories. Key findings from the analysis included the following: 

• For all three conditions, pooled across 21 states, preventable hospitalization rates were substantially higher for adults eligible for Medicaid through their eligibility for SSI compared with adults eligible for Medicaid through other pathways (e.g., via eligibility for parents with low incomes or ACA Medicaid expansions for adults with incomes up to 138% of the federal poverty level). 
• Preventable hospitalization rates for heart failure were substantially higher for Black, non-Hispanic Medicaid-enrolled adults than for white, non-Hispanic Medicaid-enrolled adults in the pooled sample of 11 states where differences by race could be assessed. The differences by race persisted even after adjusting for age, sex, managed care status, rurality, and state. Rates of asthma/COPD and diabetes preventable hospitalizations were higher among Black, non-Hispanic enrollees compared to white, non-Hispanic enrollees but the differences were smaller than those for heart failure.  
• Preventable hospitalization rates for all three conditions varied significantly by state. This was true even among adults eligible for Medicaid through SSI—for whom the eligibility thresholds are similar across most states—evidence that state variation is not entirely explained by differences in Medicaid eligibility rules. 

Although our findings highlight differences in preventable hospitalizations that suggest underlying social drivers or disparities in access to health care, many states had to be excluded from the analysis due to poor data quality, especially related to the inpatient claims and enrollment data files within TAF. Ultimately, the analysis was only able to analyze TAF data from 21 states: Alabama, Arizona, District of Columbia, Delaware, Indiana, Kansas, Louisiana, Michigan, Mississippi, Montana, North Carolina, New Jersey, New Mexico, Pennsylvania, South Carolina, South Dakota, Texas, Utah, Wisconsin, West Virginia, and Wyoming.   

Recommendations for State Medicaid Data Teams 

Investing in the quality of state Medicaid data quality can not only improve states’ ability to conduct analyses of their own Medicaid programs, but also improve the quality and usability of a states’ TAF data, which is available to external researchers like the Urban Institute team, who obtain a Data Use Agreement through CMS. There are a number of ways that Medicaid data teams can provide better and more complete data. 

Improve quality of race/ethnicity data 

The Urban research team was interested in examining racial and ethnic differences in preventable hospitalization rates which, because of data quality issues, further limited their analysis of Black-White disparities to only 11 states: Delaware, Indiana, Michigan, Mississippi, North Carolina, New Jersey, New Mexico, Pennsylvania, South Dakota, Texas, and Wyoming. In all other states, the race/ethnicity variable in the TAF file was either missing for more than 20 percent of enrollees, or the share of adults classified as non-Hispanic Black or non-Hispanic White was considerably different from external benchmarks from the American Community Survey. As described in the report, analysis of other racial and ethnic groups was not undertaken due to data quality concerns for these groups. 

Race/ethnicity data in the TAF originates from Medicaid applications, where it is then cleaned and submitted by states to CMS and standardized into eight mutually exclusive categories for the TAF. However, since states cannot legally require Medicaid applicants to report their race or ethnicity, it can be challenging to produce and submit complete race/ethnicity data.  

There are several strategies states could consider to improve the quality and completeness of their race/ethnicity data. For example, states could conduct additional data cleaning, impute an enrollee’s race/ethnicity from a prior year in the event that it’s missing in a given year, or pull data from other sources when missing from a Medicaid application. Using an integrated enrollment system may improve the completeness of race/ethnicity data; for example, North Carolina is able to collect enrollment data from Medicaid applications within DHHS itself, rather than rely on other departments to collect, store, or transfer such data.  

Examine preventable hospitalizations and health outcomes using state-specific data  

The Urban Institute’s analysis of preventable hospitalizations used publicly available claims-based algorithms provided by the Agency for Healthcare Research and Quality. States could use this software to examine preventable hospitalizations across other characteristics available directly in their Medicaid data, or by linking to other datasets within the state that may shed light on a wider range of factors contributing to health inequities. For example, linking Medicaid data with data on access to social services outside the health care system or exposure to environmental pollutants could identify shortcomings in other resources and possible root causes of disparities such as housing instability, food insecurity, or poor air quality. Addressing these social needs may help people avoid hospitalizations and other poor health outcomes.  

In states with Medicaid managed care, state data could also be used to identify managed care plans that have particularly high rates of preventable hospitalizations overall or for particular groups and work directly with those plans to improve outcomes and reduce inequities.  

How North Carolina Sustains High Medicaid Data Quality 

North Carolina has consistently high TAF data quality in 2018 and 2019 according to the CMS Data Quality Atlas, including both race/ethnicity and inpatient claims volume. The research team reached out to North Carolina’s Information technology (IT) Division within the North Carolina Department of Health and Human Services (DHHS) to learn more about how they have achieved data quality success.  

The IT team described the benefits of running an in-house data team with a commitment to producing and submitting high-quality data to CMS. A constant line of communication and collaboration between the IT team and other policy teams across the agency enables the data team to ensure that their systems can adapt to changes in North Carolina’s Medicaid program. For example, throughout North Carolina’s transition to Medicaid managed care in 2021, the IT team worked closely with managed care teams to provide ongoing guidance and support to ensure that encounter records were submitted by plans in a way that facilitated integration with their existing fee-for-service system. In states that rely heavily on capitated managed care in Medicaid — which is the norm — it likely requires additional steps to maintain high-quality data. 

Looking ahead 

The Urban Institute’s report sheds light on differences in preventable hospitalizations among Medicaid enrollees according to SSI status, race/ethnicity, and state. However, more work and improved Medicaid data quality are essential to continue to assess root causes of disparities in preventable hospitalizations — and to inform effective policy responses and actionable steps that Medicaid programs and state policymakers could take to reduce hospitalizations and promote equitable access to health care.