Method: MEPS's- Household Component (2004), a nationally representative survey of the U.S. civilian noninstitutionalized population, was used to evaluate the study objective. The analysis controlled for the complex survey design when estimating standard errors. Determinants of health insurance coverage were assessed using logistic model, while count utilization data were analyzed by endogeneity corrected negative binomial models. The predictors of health insurance coverage were ascertained aprior from literature review and the null hypothesis was rejected at alpha level 0.05. The impact of lack of coverage on resources explicitly, emergency room visit (ER), inpatient hospitalization days (IPh), ambulatory hospital stays (AmbH), number of hospital discharges (nDIS), office-based provider visits (ObP), and office-based physician visits (OMD) were quantified. We hypothesized that ER, IPh, nDIS and AmbH will be higher whereas ObP and OMD visits would be lower in the uninsured. Data were analyzed using Stata 10.0 and SAS 9.1 software.

Results: The mean age was 45.9 (±17.5) years, with a majority of females (51.9%), white (81.7%), married (56.0%) and high school diploma holders (46.5%). Demographic characteristics predicting insurance coverage included age, gender, race, marital status, family size, education and income. From the health behavior and comorbidities domains, perceived mental health and arthritis indicator were retained. Access to care domain retained; office setting of provider, difficulty reaching provider, inability to get necessary medical and dental care, and living in metropolitan area as predictors. Employment domain retained; holding multiple jobs, belonging to multi-location firm, employment in private firm, number of employees at firm, union status, and occupational category as predictors.

Incidence-rate ratios for ER (IRR=1.45, p=0.003), ObP (IRR=0.45, p<0.001), and OMD (IRR=0.31, p<0.001); supported the hypothesis that the uninsured had higher relative risk of utilization of catastrophic care and lower relative risk of utilization of routine care. Rate ratio of AmbH (IRR=1.30, p=0.83) was not statistically significant, while rate ratios of IPh (IRR=0.39, p=0.01), nDIS (IRR=0.35 p<0.001) were against apriori intuition.

Conclusion: Uninsured have higher relative risk of emergency room visits which maybe a consequence of lower utilization of routine care. Focusing interventions on determinants of insurance coverage such as improving access to routine care may reduce high cost catastrophic utilization.