Definitions Matter: Exploring Uninsurance Duration Estimates and Characteristics Associated with Long-term Uninsurance

The majority of literature on uninsured spell duration and the dynamics of uninsurance has been drawn from the 1996-1999 Survey of Income and Program Participation (SIPP). Furthermore, little is known regarding the effects of the definition of uninsurance duration estimates (e.g. total length of uninsurance versus uninsurance spell length) on the characteristics of the sample. Our study explores the associations between individual demographic and health related characteristics and 1) length of uninsured spell duration (USD) and 2) total length of time uninsured (TLT). The duration of uninsurance spells were estimated using event history analysis. TLTs were estimated by combining the total number of months uninsured in the 2005-2006 Medical Expenditure Panel Survey (MEPS). We excluded participants under 18 or over 64 years of age from our sample due to their general eligibility for government provided health insurance. The relationship between USD and individual risk factors were explored using multivariate life course techniques. Similar comparisons were conducted on TLT using bivariate and multivariate analysis. New uninsured spells within our study (median/mean= 11 months) were higher than those drawn from the 1996-99 SIPP (median = 5 months), the 2004-05 SIPP (median=8.9 months) and the 2004-05 MEPS (median=10.2 months). Individual characteristics associated with longer length of a new spell included: lower poverty and education level, minority race/ethnicity, intermittent employment, rural residence, and a history of stroke. The general adult population had a mean TLT of 6.7 months while participants with any lapse of insurance had a mean uninsurance length of 15.9 months. Individual characteristics associated with longer total length of uninsurance included: lower poverty and education, Black/Hispanic race/ethnicity, intermittent employment, non-North residence, poorer self-perceived physical health, and greater number of chronic conditions. Our estimates of USD from the MEPS are higher than those previously documented. In addition, we found the mean total length of uninsurance (15.9 months) was higher than the average length of uninsurance spell (11 months). Although individual predictors were similar for both outcomes, it is imperative that we understand the dynamics of the uninsured when using different definitions of uninsurance duration. Our findings indicate that the length of uninsurance spell has risen since previous estimates and that identifying multiple lapses of insurance status within a year is imperative in generating a complete picture of the uninsured. Based on these findings, policymakers may need to favor solutions that focus on the problems of the long-term uninsured, rather than on short-term uninsured.