Health Data Sets: Building Blocks for Research and Analysis

Now more than ever, health care data systems and data quality are critically important, as delivery systems shift and the health system changes. Mathematica Policy Research is leading the way in helping policymakers build and analyze the data sets needed for studying the changing health care environment. We continue to find innovative ways to ensure data quality to support thoughtful and policy-relevant analysis.

Building Federal-Level Files
Linking Data Sets
Collecting Survey Data
Analyzing Existing Data Sets
Reviewing and Implementing Data Systems
Using State and Local Data Sets
Focusing on High-Quality, Secure Data
Providing Experienced Systems Staff

Building Federal-Level Medicaid Research Files

To make decisions about health care delivery and health reform, policymakers need complete, accurate, and timely data. For many years, we have been involved in developing Medicaid files for the Centers for Medicare & Medicaid Services (CMS). Currently, we are providing technical assistance to CMS and all 50 states in developing the Medicaid Statistical Information System (MSIS) files for 1999 and beyond. We are assisting the states in file preparation, reviewing state crosswalks, and identifying data quality problems. We continue to provide support to CMS in the conversion of MSIS files into research files.

Linking Data Sets

Creating analytic databases needed to support major policy evaluations often involves linking data from several disparate sources that do not share unique and reliable identifiers. We have broad experience linking these types of data sets. For example, our study of the impact of the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) linked data from state Medicaid and Vital Records systems with administrative data from WIC systems to assess the effects of WIC participation on birth outcomes and Medicaid costs. We also evaluated the feasibility of creating a nationwide database of linked Medicaid and Vital Statistics data for the National Center for Health Statistics. As part of the CMS Multi-State Dual Eligibles project, we linked Medicare and Medicaid data on enrollees in 12 states and constructed a linked claims and eligibility database.

Collecting Survey Data

We are highly experienced in collecting and using survey data to support health policy analysis. Our large survey division develops and conducts policy-relevant health care surveys of general populations, special populations, and providers. Data from these surveys support our own studies and benefit other researchers in the form of public use files.

Analyzing Existing Data Sets

Not only are we helping to build the MSIS system, but we are also using data gathered through the system to study state Medicaid managed care enrollment patterns and service use for people needing long-term care in their homes or community settings and for people with mental illness. Many of our studies use health data drawn from existing sources. We have worked with data from major national cross-sectional and longitudinal surveys, such as the Current Population Survey, Survey of Income and Program Participants, National Health and Nutrition Examination Survey, Medical Expenditure Panel Survey, National Health Interview Survey, National Long Term Care Surveys, and others. We also have extensive experience in using administrative data from the Medicare program for studies ranging from the costs of treating HIV-positive individuals to the effectiveness of demonstration programs for frail elderly people.

Reviewing and Implementing Data Systems

Successful program evaluation and monitoring must be based on high-quality operational or administrative data systems. We have both reviewed existing health-related data systems and designed and implemented new systems to support research and program operations. For example, we reviewed the Drug and Alcohol Services Information System, which includes the Treatment Episode Data Set, the National Facility Register, and the Uniform Facility Data Set, for the Substance Abuse and Mental Health Services Administration. We then recommended ways to improve information reporting from these systems. We also evaluated procedures for collecting, editing, and coding health insurance data in the National Health Interview Survey, to improve the timeliness of data release without sacrificing accuracy. Other studies have focused on helping states transform the data they collect, manage, and maintain into useful information for developing and evaluating health policy. Our systems and programming staff have also designed and created efficient database systems for many health- and non-health-related applications. For example, we designed an AIDS data collection and reporting system for Ryan White grantees and helped state and local grantees implement it.

Using State and Local Data Sets

Much of the data needed for research and evaluation is maintained in different systems at the state and local level. For example, Medicaid claims and eligibility files; Vital Records files; substance abuse and disability data; and welfare, child support and enforcement, unemployment, and criminal justice files are all maintained by separate state and county agencies. Because we have used data from all these agencies in our projects, we are well aware of the quality, uniformity, and linking issues that these data present. Data from providers, such as hospital discharge data sets and managed care encounter data, have also been an important source of information in our studies.

Focusing on High-Quality, Secure Data

High-quality data are at the heart of any superior research project. We have taken the lead in quality assurance for administrative data from many program-run management information systems. In the national evaluation of Healthy Start, we played a major role in analyzing Minimum Data Set quality. For CMS, we are also reviewing and validating encounter data from Medicaid managed care. Because acquiring individually identifiable data, particularly in the health field, can raise a wide range of privacy concerns, we have developed policies and procedures in which our clients can have confidence. These procedures have satisfied the high standards of the many agencies that supply data for our evaluation and analysis of health care delivery systems.

Providing Experienced Systems Staff

Our highly trained systems analysts and programmers can work independently to address clients’ data needs. We encourage our systems staff members to share programs and ideas to maximize efficiency and minimize duplication of effort. Our comprehensive resource manual for health programmers describes conventions, discoveries, and techniques.

Back to Top