Research Abstracts Online
January 2009 - March 2010
University of Minnesota Twin Cities
Department of Pediatrics
PI: Peter B. Scal
Pediatric Hospitalizations: Is There a Link Between Safety/Quality and Hospital Competition?
Nearly all the empiric research evaluating the link between competition, cost, and quality is based on studies of adult conditions, and its relevance to pediatrics is questionable. A better understanding of the market structure of pediatric care and its effect is of critical importance because, if it is different from the adult market, then different policy options may be justified.
The specific aims of this project are to develop pediatric-specific measures of hospital competition and market structure and to evaluate the relationship between market structure and hospital outcomes (e.g.: quality of care or medical errors). The researchers are developing a computer model that tests for association between market structure measures and select pediatric patient safety indicators (PDIs). Commonly used measures of hospital competition fail to address the endongeneity of hospital volume, quality, and market share and thus may result in biased results. This model corrects for potential bias by predicting choice of hospitals among all potential choices. Choice sets are based on the distance from the patient’s zip code and services utilized. For each of the 400,000 hospitalizations there are numerous "hospital choices.” Once a measure of competition is developed, the researchers will use a multi-level logistic regression model specifying a PDI as the dependent variable and the level of competition as the principle independent variable.
It is unclear whether in the overall less competitive pediatric market the level of competition will be associated with adverse outcomes and quality of care. Empiric testing is necessary. The development of these market structure measures will allow more informative evaluations of the competition/quality/cost/volume question using advanced econometric analytic strategies that address common potential biases.
Peter Graven, Graduate Student