In late spring 2021, the Centers for Medicare & Medicaid Services (CMS) released data revealing that many Americans had forgone primary, preventive, and mental health services during the pandemic. More recent data show people returning to the doctor for at least some types of services, but the lower use of behavioral health services continues to be an area of concern, particularly because of the extraordinary stress and trauma people experienced during the pandemic.
Trends cited in the CMS announcement stemmed from Mathematica’s analysis of T-MSIS Analytic File (TAF) data. T-MSIS and TAF are the most comprehensive national data set on beneficiaries, service use, and expenditures for Medicaid and the Children’s Health Insurance Program (CHIP). CMS could glean near-real-time insight into health care use for Medicaid and CHIP beneficiaries during the pandemic, which was only possible because of more than two decades of collaboration with Mathematica and states that improved the quality, accessibility, and timeliness of Medicaid and CHIP data.
The path to producing timely data started in the early 1990s, when states began submitting electronic Medicaid claims and eligibility data files through the national Medicaid Statistical Information System (MSIS). When Congress mandated MSIS reporting for all states in 1999, these files became the only national, uniform, and comprehensive data collection system for Medicaid and CHIP. Mathematica was responsible for processing these data and using them to develop a research file known as the Medicaid Analytic eXtract.
By 2017, all states transitioned to a new system called T-MSIS, which enhances and expands the previous system in several ways. T-MSIS data are more timely, more complex, and considerably richer than MSIS data. Transitioning to T-MSIS presented a large challenge for states, and any new data system raises concerns about the quality of the data. Mathematica worked with CMS to develop a comprehensive strategy for identifying data quality issues. We now use more than 2,000 data quality checks as states submit their files to CMS, prioritizing these checks into T-MSIS Priority Items and working with states to fix them.
As one part of our team works with states to improve data quality, another works to transform the data into a new research file: the TAF. These data are easier to use than raw T-MSIS data and are designed for staff at the Center for Medicaid and CHIP Services and external researchers. Before making the TAF data more widely available, CMS wanted to ensure that we knew the strengths and limitations of the data for analytic purposes. This required another broad set of data quality and usability checks, ultimately resulting in the creation of the Data Quality Atlas, a publicly available tool for TAF users.
It was this sequence of events that put us in a position to respond when, in March 2020, CMS needed to understand how COVID-19 was affecting Medicaid and CHIP, programs that cover some of the most vulnerable populations in the country. By then, states were submitting quality T-MSIS data, and we had plenty of evidence that the data were usable. Within one month, we produced TAF for more than three years of T-MSIS submissions, collecting data through February 2020, and revamped the production process to produce updated TAF data monthly rather than quarterly.
From the new data, we developed tables tracking COVID-19-related conditions and acute care use among Medicaid and CHIP beneficiaries as well as a data visualization tool that enabled us and CMS staff to see how trends changed on a month-by-month basis. We continued to build out the measures and the functionality of the tool so that we can now see important patterns in more than 20 monthly measure sets of enrollment and service use with only a two-month lag. Today, we continue to see new patterns emerge as the pandemic unfolds.
Although using Medicaid data to monitor trends has proven useful during the pandemic, we know that the most recent months of data are incomplete. Claims can take up to a year to be submitted, particularly claims for long-term care facilities, inpatient stays, and beneficiaries covered by managed care organizations. Anyone using TAF data to answer urgent questions should keep this important limitation in mind.
The past 18 months have shown TAF data to be a critical resource for monitoring Medicaid and CHIP, but using the data for program monitoring is just the beginning. As we continue to demonstrate the utility of the data, we will expand how we use them to continue supporting CMS’s triple aim of better care for individuals, better health for populations, and lower cost. Just as we couldn’t have imagined how TAF data would be useful during a pandemic, we can’t anticipate all the potential needs for these data in the future. But we can ensure that accessible, high quality, and timely data will be available when those needs arise.