Date of Award


Document Type

Campus Access Dissertation


Health Services and Policy Management

First Advisor

Jan Probst


Hospital-acquired conditions (HACs) are major patient safety and quality concerns in the health care industry. In the United States, hospital-acquired infections occur in approximately 1.7 to 2 million hospitalized patients annually, or approximately 20%. (Klevens et al. 2007), (Burke 2003), (P. W Stone et al. 2010), (Scott 2009), (Inouye, Cynthia Brown, and Tinetti 2009) The overall goal of the present study is to evaluate whether the proportion of admissions for which a hospital-acquired condition is recorded has declined as a result of Medicare's nonpayment policy. It is hypothesized that the rate of HACs will decline more sharply after the implementation of Medicare's nonpayment policy compared to years prior to the policy change.

The dependent variable is the overall rate of hospital-acquired conditions (HACs) measured at the hospital level. The rate of HACs is the number of occurrences of each HAC per hospital over the total number of patients per hospital. The rate is multiplied by 1,000 so that inference is in terms of HACs per 1,000 patient discharges. The independent variable was time, which is associated with Medicare's nonpayment policy. Time is measured in quarters, with 20 quarters calculated over the five year study period. Quarters 1-15 comprise the "before" period prior to the policy change, and quarters 16-20 comprise the "after" period, after the policy change. There are two sets of control variables: hospital characteristics, including bed size (small, medium, or large), teaching status, (nonteaching or teaching), control and ownership (government or private, government public, private not-for-profit, private investor-owned, or private), location (rural or urban), and region (Northeast, Midwest, South, or West) and patient characteristics, including admission type, admission source, insurance status, age, and sex.

We conducted a time series analysis using the 2005, 2006, 2007, 2008 and 2009 Nationwide Inpatient Sample (NIS) datasets. The rates of HACs were calculated over a five year period on a quarterly basis resulting in a total of 20 time points. All analyses used NIS weights to allow generalization of inference from our sample to inference for the nation as a whole. A repeated measures Poisson regression growth curve model was estimated to analyze the rate of HACs over the five-year period of interest.

The mean rate of HACs was greater after the policy than before; however, when adjusting for hospital and patient characteristics, the rate of change over time declined. Before the policy the growth curve slope of HACs was positive, while after the policy the growth curve slope of HACs was slightly negative or almost flat. The rate of growth of HACs slightly declined after the policy, but the decline was not significant. The evidence for this study suggests that Medicare's nonpayment policy was not effective at reducing the rate of increase in hospital-acquired conditions. More time is needed to more accurately assess the change in rate of HACs and the true effectiveness of Medicare's nonpayment policy.