Title: Evaluating the effect of healthcare providers through semi-Markov multi-state model and nonparametric discrete frailty
Authors: Francesca Gasperoni - MRC Biostatistics Unit, University of Cambridge (United Kingdom) [presenting]
Francesca Ieva - Politecnico di Milano (Italy)
Anna Maria Paganoni - MOX-Politecnico di Milano (Italy)
Chris Jackson - MRC Biostatistics Unit -Cambridge (United Kingdom)
Linda Sharples - Department of Medical Statistics - London School of Hygiene and Tropical Medicine - London (United Kingdom)
Abstract: Novel exploratory statistical methodology is introduced for investigating healthcare providers' effect on different patients outcomes through clinical administrative databases. The main purpose consists in identifying clusters of providers (latent populations) on the basis of patients' characteristics, considering the whole healthcare patients clinical history: re-admissions, discharges and death. We propose a semi-Markov multi-state model to describe the duration of hospital stay, time between hospital discharge and re-admission and time to death during admission and outside of hospital. Transition-specific hazards are modelled through a Cox proportional hazards model with a nonparametric discrete frailty term that is shared among patients hospitalised in the same provider. The inclusion of a nonparametric discrete frailty allows us to detect latent populations for each specific transition. The estimates are computed through a tailored Expectation-Maximization algorithm. As a direct consequence, we are able to identify the most frequent and most extreme latent populations across all transitions. This result is of interest for healthcare managers that can further investigate those providers associated to the most extreme latent populations' patterns. The proposed method is illustrated through an application to Heart Failure patients recorded in an administrative database from Lombardia, a northern region in Italy.