M Rabilloud1, R Ecochard, J Estève. 1. Unité de Biostatistiques du Département d'Information Médicale, Hospices Civils de Lyon, Lyon, France. muriel.rabilloud@chu-lyon.fr
Abstract
OBJECTIVE: Ranking maternity hospitals on prophylactic caesarean section rates. STUDY DESIGN: The study population was a sample of 8470 women having delivered in the 86 maternity hospitals of the Rhône-Alpes region (France) in 1990. A two-level logistic model and the Gibbs sampling method were used to estimate the adjusted rate and rank of each maternity hospital with their 95% credible interval. RESULTS: Ranking of maternity hospitals was substantially modified after adjusting for differences in case-mix. The 95% credible intervals of the ranks were generally wide and overlapping. No maternity hospital could be confidently placed in the upper quarter nor in the lower quarter of the league table. CONCLUSION: The use of ranks based on rates to compare hospitals may be seriously misleading; thus, league tables need an accompanying reading guide. Furthermore, some rules should be respected, including the need to make accurate case-mix adjustments, to reduce the rate instability in smaller hospitals, and to highlight the uncertainty associated with ranks.
OBJECTIVE: Ranking maternity hospitals on prophylactic caesarean section rates. STUDY DESIGN: The study population was a sample of 8470 women having delivered in the 86 maternity hospitals of the Rhône-Alpes region (France) in 1990. A two-level logistic model and the Gibbs sampling method were used to estimate the adjusted rate and rank of each maternity hospital with their 95% credible interval. RESULTS: Ranking of maternity hospitals was substantially modified after adjusting for differences in case-mix. The 95% credible intervals of the ranks were generally wide and overlapping. No maternity hospital could be confidently placed in the upper quarter nor in the lower quarter of the league table. CONCLUSION: The use of ranks based on rates to compare hospitals may be seriously misleading; thus, league tables need an accompanying reading guide. Furthermore, some rules should be respected, including the need to make accurate case-mix adjustments, to reduce the rate instability in smaller hospitals, and to highlight the uncertainty associated with ranks.
Authors: Chris Feudtner; Jay G Berry; Gareth Parry; Paul Hain; Rustin B Morse; Anthony D Slonim; Samir S Shah; Matt Hall Journal: Pediatrics Date: 2011-09-02 Impact factor: 7.124
Authors: Maria P Fantini; Elisa Stivanello; Brunella Frammartino; Anna P Barone; Danilo Fusco; Laura Dallolio; Paolo Cacciari; Carlo A Perucci Journal: BMC Health Serv Res Date: 2006-08-15 Impact factor: 2.655