A semi-parametric model for clustering hospitals by similarity in patients’ outcome: a study of cesarean sections rates in Sardinia

CANNAS, MASSIMO;CONVERSANO, CLAUDIO;MOLA, FRANCESCO;Sironi, E.

2014-01-01

---

Abstract

The differences in rates of cesarean deliveries across hospitals cause concern and debate about the appropriateness of many interventions. This problem is particularly relevant in Italy, which has one of the highest intervention rates in Europe. Using data from hospital abstracts on deliveries that occurred in Sardinia over a two-year period, we fit a semiparametric logistic regression model with a Dirichlet process prior for the random effects. The model is useful to assess whether the observed differences in cesarean rates across hospitals can be justified by case-mix differences across hospitals. Moreover, the discrete nature of the random effects is exploited in order to obtain an optimal clustering of the hospitals affecting decisions on cesarean section in a similar way.