4
Administrative Databases
Administrative data collected in the course of program operations, no matter how good and complete, are unlikely to be able to fully substitute for survey data in that, although they can depict the population enrolled and served in public programs, they cannot delineate the population that is truly uncovered. An uncovered population as delineated by program data consists of children who may well have coverage through other (mostly private) programs and those who have fallen through the cracks and have not been enrolled in the programs.
Nonetheless, the counts of actual children enrolled and served that are the product of the collection of program operating data are far better counts of the served population than can be obtained through surveys, for the reasons given in Chapter 3. This chapter discusses the major sources of administrative data on children covered in public programs, assesses their strengths and weaknesses, and addresses steps being taken by the federal and state agencies responsible for collecting and compiling the data to improve their quality, timeliness, and representativeness over time.
MAJOR ADMINISTRATIVE DATABASES
The three major administrative databases maintained by the federal government are the CMS-64 Quarterly Expense Report to the Centers for Medicare & Medicaid Services (CMS), the Medicaid Statistical Information System (MSIS), and the Children’s Health Insurance Program (CHIP) Statistical Enrollment Data System (SEDS). Each of these is discussed below.
CMS-64 Quarterly Expense Report
Form CMS-64 is a statement of expenditures for which states are entitled to federal reimbursement under Title XIX of the Social Security Act. It reconciles the monetary advance made on the basis of Form CMS-37, filed previously for the same quarter. Consequently, the amount claimed on Form CMS-64 is a summary of expenditures derived from source documents, such as invoices, cost reports, and eligibility records. At the workshop, David Rousseau made the point that the CMS-64 is primarily collected to reconcile payments to states from the federal treasury for services delivered to Medicaid beneficiaries. It has a great deal of information about overall spending but not very much about what is driving that growth, because, he pointed out, there is no information on utilization, enrollment, or the types of beneficiaries that use that service, and managed care spending is not separated into the services.
Medicaid Statistical Information System
The purpose of the Medcaid Statistical Information System (MSIS) is to collect, manage, analyze, and disseminate information on eligibles, beneficiaries, utilization, and payment for services covered by state Medicaid programs (see www.cms.gov/MSIS [September 2010]). The MSIS data serve multiple purposes, including health care research and evaluation activities, program utilization and expenditures forecasting, analyses of policy alternatives, responses to congressional inquiries, and matches to other health-related databases. States provide CMS with quarterly computer files containing, among other data items, specified data elements for persons covered by Medicaid (eligible files). If a person is covered by Medicaid (or CHIP) for at least 1 day during the reporting quarter, their demographic and monthly enrollment data resides in this file. Claims records contain information on the types of services provided, providers of services, service dates, costs, types of reimbursement, and epidemiological variables. The data files are subjected to quality assurance edits to ensure that the data are within acceptable error tolerances, and a distributional review verifies the reasonableness of the data. Once accepted, valid tape files are created that serve as the historical source of detailed Medicaid eligibility and paid claims data maintained by CMS.
Statistical Enrollment Data System
States submit quarterly and annual CHIP statistical data to CMS through the SEDS automated reporting system (U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services, 2010, p. 47). Using forms provided by CMS, states report unduplicated
counts of the number of children under age 19 who are enrolled in separate CHIPs and Medicaid expansion CHIPs.
ASSESSMENT OF THE MAJOR ADMINISTRATIVE DATABASES
Rousseau pointed out that the databases do a good job of measuring what they are intended to measure. For example, the measures of program enrollment are detailed and extensive. Year-to-year enrollment is the primary driver of Medicaid spending growth and is, in fact, the only driver of Medicaid’s increasing share of overall national health care spending. The data systems permit more detailed analysis of enrollment data. But there are issues with timeliness and completeness, which has driven organizations such as the Kaiser Family Foundation to collect Medicaid and CHIP administrative data directly from the states to track enrollment growth on the same time line as spending data from CMS become available, mostly because enrollment data lag spending data by at least a year. Kaiser collects administrative data directly from all 50 states and the District of Columbia in partnership with researchers at Health Management Associates. The organization finds that having enrollment data from the same period as the latest spending data allows it to analyze spending trends far sooner.
Depending on the questions that are asked of the administrative data, the sources change and the answers differ in ways that may be confusing. For example, Medicaid and Medicaid expansion CHIP are reported in MSIS. If the data user is looking for the population that is “ever enrolled by fiscal year,” the source is MSIS Medicaid State Summary Mart; the population “ever enrolled in a calendar year” comes from the Medicaid Analytic eXtract (MAX), a special repository of research data derived from MSIS; and the population “enrolled at a point-in-time” is also available from MAX (enrolled in any given month). The “ever enrolled” and “point-in-time” estimates are themselves quite different: the “ever enrolled” counts are inclusive annual (fiscal or calendar year) counts of Medicaid and CHIP enrollees, which include those enrolled for any time, from a single day up to a full year, and “point-in-time” estimates are the count as of a single, specified day. These files may or may not represent the state CHIP programs, since stand-alone state CHIP enrollments are reported at state option. A total of 44 states have either combination (Medicaid and stand-alone) or stand-alone only CHIPs, but 21 do not report all stand-alone CHIP enrollees in MSIS, so users have to look elsewhere to get a total for both CHIP elements. This is an important issue, because MSIS includes all Medicaid and Medicaid expansion CHIP enrollments (M-CHIP), but enrollment in the state CHIP (S-CHIP), which was more than two-thirds (69 percent) of enrollment in 2008, is not consistently
included in the MSIS. This limits the ability to look at children’s enrollments in Medicaid and CHIP in a comparable way.
There are also troubling differences in the totals that are obtained from the different administrative data sources. For example, Rousseau found that, after adjusting for differences in scope and design, since the late 1990s, MSIS spending is consistently below CMS-64 spending totals. Differences between the CMS-64 and the MSIS by state range from 45 to –13 percent, with six states reporting more than 10 percent higher spending in the MSIS than in the CMS-64, which potentially means they are losing out on some federal dollars. In total, the differences range from about $5 billion to more than $16 billion per year. These differences are noted even after excluding spending from the CMS-64 that is not intended to be put into the MSIS, such as administration or a disproportionate share of hospital spending or accounting adjustments.
The differences between administrative data counts and survey estimates are sometimes explainable. Rousseau mentioned that one difference between the National Health Interview Survey (NHIS) and MSIS enrollment data for the same time period is that the MSIS includes retroactive eligibility—persons who were deemed not eligible at the point in time but who have become eligible over time. Another difference is in definitions, a common difference being varying age cutoffs to describe the population.
A summary of the strengths and weaknesses of the three main administrative databases is found in Table 4-1.
STEPS BEING TAKEN TO IMPROVE THE ADMINISTRATIVE DATABASES
David Baugh summarized the intensive program under way at CMS to rationalize and improve these administrative databases, particularly the MSIS data. These efforts, according to Rousseau, are bearing fruit, in that MSIS data quality continues to improve, with far fewer beneficiaries of unknown eligibility and far less spending on unknown services.
Baugh discussed an initiative called the Medicaid and CHIP Business Information and Solutions (MACBIS) council, which provides leadership and guidance in support of efforts toward a more robust and comprehensive information management strategy for Medicaid, CHIP, and state health programs. The MACBIS is charged with promoting consistent leadership on key challenges facing state health programs, including quality, access, value, and integrity; improving the efficiency and effectiveness of the federal-state partnership; making data on the Medicaid, CHIP, and state health programs more widely available to stakeholders; and reducing duplicative efforts within CMS and thus minimizing the burden on states. Under the auspices of MACBIS, the agency is also developing an
TABLE 4-1 Strengths and Weaknesses of Primary CMS Administrative Databases for CHIP
Strengths |
Weaknesses |
CMS-64 |
|
|
|
Medicaid Statistical Information System (MSIS) |
|
|
|
Statistical Enrollment Data System |
|
|
|
SOURCE: Data from Rousseau (2010). |
enhancement to the MSIS known as MSIS Plus, which expands reporting to support anti-waste, fraud, and abuse programs and other activities.
New funding has been made available to improve the MSIS data through an expansion of MAX data activities funded by the American Recovery and Reinvestment Act (ARRA). CMS also coordinates a project known as the Medicaid and CHIP Fee-for-Service, Managed Care, Eligibility, and Provider Data Improvement Project, which is identifying the data elements that CMS needs in order to manage Medicaid, CHIP, and other state programs.
Another initiative is to improve the identification of program enrollees by obtaining identification crosswalks from Medicaid agencies. This work has been under way for some time, but to date, corrections have been implemented in MAX but not MSIS. The agency has also turned attention to verification of Social Security numbers on a limited basis using the
Social Security Administration High Group List. CMS is exploring more complete verification in concert with the Social Security Administration, the Census Bureau, and the states.
Duplication in the count of enrollees across the states and over time is an issue for CMS and data users. Baugh estimated that about 1 million of the 60 million Medicaid/CHIP enrollees are duplicates. The duplicates will be reduced by creation of a MAX Enrollee Master file (MAXEM), on which work is now under way.
Lastly, he reported that under the umbrella of the MAX expansion project, funded by ARRA, there is increased coordination between CMS and State Technical Advisory Groups to describe CHIPs, identify their issues, challenges, and concerns, and develop criteria for selecting volunteers. As part of this initiative, CMS is working to develop state-specific technical assistance plans in 10 to 15 pilot states, which include regular communications and working closely with data task teams, with the objective of improving MSIS reporting. However, at this time, state participation in these quality improvement programs is voluntary, regulations have yet to be promulgated, and states have limited resources to make improvements that have been identified.
SUMMARY
In summarizing his views on the status of administrative records for estimating children’s health insurance coverage, Rousseau observed that administrative data are a good resource for research and benchmarking, but they are not a perfect gold standard. For one thing, the data are collected for other purposes (e.g., CMS-64) and not for specific research questions, so they may not be able to fully illuminate some key issues, such as the lack of managed care information in MSIS. However, they are increasingly available and researchers can and do use them for a variety of purposes. Despite their limitations, administrative data complement survey data and are a critical component in all work of CHIP and Medicaid researchers.
Administrative data are critical for state-level analysis of subpopulations for which surveys are often limited by sample size and other methodological issues. Real-time administrative data on enrollment are critical to decision makers and analysts. CMS has made great strides in improving the availability and timeliness of administrative data, but more work needs to be done to ensure data quality and make enrollment data contemporary with spending. The new health reform legislation increases the need for timely, high-quality administrative and survey data.