The effect of relationship status on health with dynamic health and persistent relationships.

The dynamic evolution of health and persistent relationship status pose econometric challenges to disentangling the causal effect of relationships on health from the selection effect of health on relationship choice. Using a new econometric strategy we find that marriage is not universally better for health. Rather, cohabitation benefits the health of men and women over 45, being never married is no worse for health, and only divorce marginally harms the health of younger men. We find strong evidence that unobservable health-related factors can confound estimates. Our method can be applied to other research questions with dynamic dependent and multivariate endogenous variables.

The dynamics of medical care use in the British household panel survey.

We explore whether medical care use is persistent over a long panel using 18 waves of the British Household Panel Survey. Of particular interest is high medical care use because a few high users account for a disproportionate amount of use while many individuals use no medical care in a given year. If health is a primary driver of medical care demand, and we control for health, then past medical care use should be uninformative for future use. However, we find that conditional on health, other covariates and unobservable heterogeneity, medical care use remains significantly persistent. "No use" and "high use" are more strongly persistent, and persistence is generally stronger for women, those in poor health, and at older ages. We find that unobservable heterogeneity explains between 10% and 25% of the variation in medical care use. This heterogeneity is significantly correlated with both medical care use and health over our long panel. These findings have implications for the econometric modeling of medical care demand and suggest that policies aimed to reduce aggregate medical care spending by improving health, particularly the health of seniors, may be less effective than projected using static models.

Cost of hospital-acquired infection.

The authors assessed the costs of hospital-acquired infections using rigorous econometric methods on publicly available data, controlling for the interdependency of length of stay and the incidence of hospital acquired infection, and estimated the cost shares of different payers. They developed a system of equations involving length of stay, incidence of infection, and the total hospital care cost to be estimated using simultaneous equations system. The main data came from the State of New Jersey UB 92 for 2004, complimented with data from the Annual Survey of Hospitals by the American Hospital Association and the Medicare Cost Report of 2004. The authors estimated that an incidence of hospital acquired infection increases the hospital care cost of a patient by $10,375 and it increases the length of stay by 3.30 days, and that a disproportionately higher portion of the cost is attributable to Medicare. They conclude that reliable cost estimates of hospital-acquired infections can be made using publicly available data. Their estimate shows a much larger aggregate cost of $16.6 billion as opposed to $5 billion reported by the Centers for Disease Control and Prevention but much less than $29 billion as reported elsewhere in the literature.