November 1, 2003
Pamela Farley Short, Deborah R. Graefe, Ph.D.
"Battery-Powered Health Insurance: Stability in Coverage of the Uninsured," Pamela Farley Short, Ph.D, andDeborah R. Graefe, Ph.D., Health Affairs 22, 6 (November 2003): 24455
New analysis of insurance coverage over time finds that a total of 84.8 million people under age 65 were uninsured for at least one month in the four-year period 1996 to 1999. This amounts to 38 percent of the population tracked during these four years. Reflecting turnover in the uninsured population, the number of people who were uninsured was nearly double the number uninsured at any one point in time. It is also nearly double a recent Census Bureau estimate of uninsured Americans in 2002 (43.6 million). In addition to those uninsured for all or most of the four years, a substantial portion who had a time without coverage experienced repeated lapses in coverage, exposing them to major financial and health risks.
Commonwealth Fund–supported study, "Battery-Powered Health Insurance? Stability in Coverage of the Uninsured,"
published in Health Affairs
(Nov./Dec. 2003), researchers Pamela Farley Short and Deborah R. Graefe of Pennsylvania State University analyzed monthly insurance status over the four years 1996 to 1999 using data from the 1996 panel of the Survey of Income and Program Participation (SIPP). SIPP tracked coverage for 40,000 households representing 226 million Americans under age 65. The authors' analysis reveals a complex picture of insurance instability over time, with implications for the design of health policy reforms.Patterns of Uninsured Gaps over Time
The finding that 84.8 million people under age 65 were uninsured for at least one month uncovers considerable turnover and instability in coverage. Most of those uninsured had one or more changes in coverage over time, often adding up to substantial time uninsured during the four years. The majority of the uninsured were uninsured for more than 12 months over the four-year period, although most new uninsured spells lasted for less than one year.
Those with any time uninsured divided into several distinct longitudinal patterns.
- A third of those uninsured—28.2 million people—were repeatedly uninsured, meaning two or more lapses in insurance and at least two times insured. The majority of these individuals (70%) were uninsured more often than they were insured during the four years.
- 10.1 million were without coverage for the entire four years. Nearly 5 million were uninsured for most of the four years, with only temporary coverage when insured. Most people in this group (87%) were without insurance for more than two of the four years.
- About 8.5 million of those surveyed were "scrambling for coverage." Even though they had only one gap in coverage, they changed insurance type several times.The uninsured also include those transitioning in or out of coverage or with an isolated gap during these four years. About 10 million of those surveyed began with no insurance and then became insured for the remainder of the four years.
As part of the initiative to determine whether health plans can obtain data on members' race and ethnicity and whether those data can be used to generate reports on quality of care stratified by race/ethnicity, researchers collaborated with African American, Hispanic, and Asian and Pacific Islander community leaders, as well as a wide range of technical experts. The advisory groups identified four basic components of a quality-of-care report card for health plans:
- HEDIS effectiveness-of-care measures.
- Information about plan members' experiences with care obtained from the Consumer Assessment of Health Plans Survey (CAHPS).
- Surveys of patients with chronic disease (asthma or diabetes) or newly diagnosed prostate cancer, as a way to develop measures of provider-patient communication.
- Survey of plans' cultural and linguistic competence.At the project's outset, data on race and ethnicity were not available at the plan level, so alternative data sources had to be devised. The project was divided into two phases.
The first phase of the project focused on identifying minority populations. Some plans used three general measures to calculate HEDIS or other quality-of-care measures for different racial/ethnic groups:
- A self-report item on race/ethnicity in surveys was used to split survey participants into racial and ethnic groups, and analyses were done comparing responses across groups.
- Software that relies on surnames to distinguish Hispanic members from non-Hispanic members was used to calculate HEDIS scores for those two groups of plan enrollees.
- Information from providers' medical records and electronic encounter databases helped plans assign patients to racial and ethnic groups. These data were then used to analyze disparities in processes of care.The project's advisory groups favored self-reports as the best method of assigning persons to race/ethnicity groups. However, until health plans have self-reported data on race and ethnicity for all of their members, other methods—even with their inherent weaknesses—can be used as proxies, the researchers suggest.
The project's second phase involved working with the eight health plans to demonstrate the feasibility of separately reporting HEDIS, CAHPS, and other quality measures by race/ethnicity. The plans used five methods for obtaining such data in order to prepare comparative quality reports:
- Self-reported data from CAHPS and chronic disease surveys.
- State Medicaid data files.
- Medical record information obtained during chart-review stage of the HEDIS hybrid method.
- GUESS (Generally Useful Ethnicity Search System) software for estimating Hispanic ethnicity based on surname.
- Geocoding, with race/ethnicity of selected persons imputed based on street address and Census data.All plans were able to incorporate race/ethnicity data into their procedures for HEDIS and CAHPS in 2001 and nearly all were able to generate stratified reports. Lastly, the researchers said that there are no legal barriers, except possibly in four states, that would bar health plans from using data on race and ethnicity for the purposes of improving health care quality.