TABLE 2.
Recent Quasi-Experimental Studies of the Effects of Health Insurance Coverage on Health Outcomes
| Study | Dataa | Methodological Approach | Principal Findingsb | Limitations |
|---|---|---|---|---|
| Card, Dobkin, and Maestas 2004; The Impact of Nearly Universal Insurance Coverage on Health Care Utilization and Health: Evidence from Medicare. | Cross-sectional survey data from the 1992–2001 NHIS; mortality data from NCHS Multiple Cause of Death files. | Regression discontinuity analysis of general health status by age. | Medicare eligibility after age 65 associated with significant 12% relative reduction in sociodemographic disparity in general health status but no evidence of deceleration in mortality rates at age 65. | Comparisons by prior insurance status or preexisting conditions not possible with cross-sectional data; only one self-reported general health outcome assessed; differential changes in health trends not assessed; regression discontinuity design not suited to identifying delayed mortality effects in general population. |
| Card, Dobkin, and Maestas 2007; Does Medicare Save Lives? | Cross-sectional state hospital discharge data from California from 1992 to 2002. | Regression discontinuity analysis of mortality by age among acutely ill adults hospitalized for nondeferrable conditions. | Medicare eligibility after age 65 associated with abrupt absolute decrease in seven-day mortality of 1% (20% relative reduction) that persisted for at least two years after admission. | Comparisons by prior insurance status not possible with cross-sectional data; alternative explanations for survival gains could not be tested directly. |
| Decker and Rapaport 2002; Medicare and Inequalities in Health Outcomes: The Case of Breast Cancer. | SEER cancer registry data from 1980 to 1994 with follow-up mortality data. | Difference-in-differences comparisons of stage of diagnosis and survival for white and black women with breast cancer before and after age 65. | Medicare eligibility after age 65 associated with significant decrease in probability of late detection for white women but not black women; coverage estimated to increase five-year survival rate for both black and white women diagnosed with early-stage disease, but differential effect for black women not significant. | Comparisons by prior insurance status not possible; persistent racial and ethnic disparities in outcomes among insured adults may have reduced differential effects; outcomes assessed for breast cancer only. |
| Decker 2005; Medicare and the Health of Women with Breast Cancer. | SEER cancer registry data from 1980 to 2001 with follow-up mortality data. | Difference-in-differences comparisons of stage of diagnosis and survival for white, black, and Hispanic women with breast cancer before and after age 65. | Medicare eligibility after age 65 associated with absolute decrease of 3.4% in probability of late detection for Hispanic women and 1.8% decrease for white women, but differential effect not significant; 11% relative reduction in mortality risk after age 65 did not differ by race or ethnicity. | Comparisons by prior insurance status not possible; persistent racial and ethnic disparities in outcomes among insured adults may have reduced differential effects; outcomes assessed for breast cancer only. |
| Decker and Remler 2004; How Much Might Universal Health Insurance Reduce Socioeconomic Disparities in Health? A Comparison of the US and Canada. | Cross-sectional survey data from the 1997–1998 NHIS and the 1996–1997 NPHS. | Difference-in-differences-in-differences comparison of general health status by age in the U.S. and Canada. | Medicare eligibility after age 65 associated with a significant differential reduction of 4.0 percentage points in probability of fair or poor health for low-income U.S. adults; socioeconomic disparity in general health among nonelderly adults reduced by more than half. | Comparisons by prior insurance status or preexisting conditions not possible with cross-sectional data; only one self-reported general health outcome assessed; differential changes in health trends not assessed; precise age at which discontinuity occurred could not be assessed; fluctuating disparities in health within countries suggest explanations other than Medicare coverage. |
| Dor, Sudano, and Baker 2006; The Effect of Private Insurance on the Health of Older, Working Age Adults: Evidence from the Health and Retirement Study. | Longitudinal survey data from the 1992–1998 HRS. | Instrumental variables analysis using state-level marginal tax rates, unemployment rates, and unionization rates as instruments for health insurance coverage. | Having private insurance at baseline associated with significantly better health scores for a summary index of five general and physical health measures. | Validity of instruments cannot be tested directly; self-reported health outcomes; continuity of coverage not assessed. |
| Finkelstein and McKnight 2005; What Did Medicare Do (and Was It Worth It)? | Mortality data from NCHS Multiple Causes of Death files. | Difference-in-differences comparisons of mortality before and after 1965 by age (young elderly who became covered by Medicare in 1965 versus near-elderly who did not) and by geographic variation in insurance rates before 1965. | No discernable impact of the introduction of Medicare in 1965 on overall mortality for elderly adults. | Coverage gains occurred more than 40 years ago; subsequent medical advances likely to have improved the effectiveness of health care. |
| Hadley and Waidmann 2006; Health Insurance and Health at Age 65: Implications for Medical Care Spending on New Medicare Beneficiaries. | Longitudinal survey data from the 1992–1998 HRS. | Instrumental variables analysis using spouse's prior union status, immigrant status and years in the U.S., and involuntary job loss as instruments for health insurance coverage. | Continuous insurance coverage associated with significantly fewer deaths among the near-elderly before age 65 (2.8% absolute decrease in death rate) and significant upward shift in distribution of general health states among those who survived (3.3% and 4.1% absolute increases in probability of excellent and very good health, respectively). | Validity of instruments cannot be tested directly; self-reported health outcomes. |
| Lichtenberg 2002; The Effects of Medicare on Health Care Utilization and Outcomes. | Cross-sectional survey data from the 1987–1991 NHIS; vital status data from SSA life tables. | Regression discontinuity analyses of disability and mortality by age. | 13% relative reduction in bed days and 5.1% absolute decrease in ten-year mortality risk associated with Medicare eligibility after age 65. | Effects not disaggregated by predictors of insurance status; comparisons by prior insurance status not possible with cross-sectional data; potentially spurious results due to data limitations of SSA life tables; formal testing of effects not consistently conducted; alternative explanations not addressed. |
| McWilliams et al. 2007a; Health of Previously Uninsured Adults after Acquiring Medicare Coverage. | Longitudinal survey data from the 1992–2004 HRS. | Comparison of health trend changes at age 65 by prior insurance status. | Medicare eligibility after age 65 associated with differentially improved health trends for previously uninsured with cardiovascular disease or diabetes in summary health (p= .006), change in general health (p= .03), mobility (p= .05), agility (p= .003), and adverse cardiovascular outcomes (p= .02); differential improvement also significant for depressive symptoms (p= .002) but not summary health (p= .17) for previously uninsured without these conditions. | Self-reported health outcomes; subject to bias from differential mortality among previously uninsured or coincidental changes in time-varying predictors of health among comparison groups. |
| Pauly 2005; Effects of Insurance Coverage on Use of Care and Health Outcomes for Nonpoor Young Women. | Cross-sectional survey data from the 1996 MEPS. | Instrumental variables analysis using firm size and marital status as instruments for health insurance status. | Associations between insurance coverage and probability of fair or poor health not significant in either naïve or instrumental variables analyses. | Validity of instruments cannot be tested directly; imprecise estimates; only one self-reported general health outcome assessed. |
| Polsky et al. 2006; The Health Effects of Medicare for the Near-Elderly Uninsured. | Longitudinal survey data from the 1992–2004 HRS. | Comparison of health trend changes at age 65 by prior insurance status. | Medicare eligibility after age 65 associated with significant improvements in health trajectories for both previously insured and previously uninsured adults; differential increase in probability of being in excellent or very good health after age 65 not significant for previously uninsured adults (absolute increase +1.8%; 95% CI: −2.6,7.0). | Only one self-reported general health outcome assessed; subject to bias from differential mortality among previously uninsured or coincidental changes in time-varying predictors of health among comparison groups. |
| Volpp et al. 2003; Market Reform in New Jersey and the Effect on Mortality from Acute Myocardial Infarction. | Cross-sectional state and national hospital discharge data from New Jersey, New York, and the NIS from 1990 to 1996. | Difference-in-differences comparisons of mortality rates for hospitalized patients with acute myocardial infarction in New Jersey and New York before and after state reforms in New Jersey reduced subsidies for hospital care for the uninsured and introduced price competition. | New Jersey health care reform associated with no significant changes in mortality for insured patients in New Jersey relative to New York or the nation, but with a significant differential increase of 3.7 to 5.2 percentage points in mortality rates for uninsured patients in New Jersey. | Subject to bias from coincidental changes in state-specific predictors of mortality in insured and uninsured populations; mortality for only one condition assessed; analysis limited to one state and may not generalize to national population of uninsured. |
| Volpp et al. 2005; The Effects of Price Competition and Reduced Subsidies for Uncompensated Care on Hospital Mortality. | Cross-sectional state hospital discharge data from New Jersey and New York from 1990 to 1996. | Difference-in-differences comparisons of mortality rates for hospitalized patients with six other acute conditions in New Jersey and New York before and after state reforms in New Jersey reduced subsidies for hospital care for the uninsured and introduced price competition. | New Jersey health care reform associated with relative increases in mortality for uninsured New Jersey patients with congestive heart failure (p < .05) and stroke (p > .05) compared with uninsured New York patients; mortality trends similar in New Jersey and New York for patients with other conditions, regardless of insurance status. | Subject to bias from coincidental changes in state-specific predictors of mortality in insured and uninsured populations; analysis limited to one state and may not generalize to national population of uninsured. |
Notes:
a
Abbreviations of data sources: HRS (Health and Retirement Study), MEPS (Medical Expenditure Panel Survey), NCHS (National Center for Health Statistics), NHIS (National Health Interview Survey), NIS (Nationwide Inpatient Sample), NPHS (National Population Health Survey), SEER (Surveillance, Epidemiology, and End Results), SSA (Social Security Administration).
b
Point estimates, 95% confidence intervals (CI), or p-values presented as reported in original articles.