Hiroaki Osada1, Etsuko Yamakoshi. 1. Department of Surgery, Division of Chest Surgery, St. Marianna University School of Medicine, 2-16-1 Sugao, Miyamae-ku, Kawasaki 216-8511, Japan. h2osada@marianna-u.ac.jp
Abstract
OBJECTIVE: This study aimed at analyzing results of surgery for lung cancer in relation to hospital volume to see if a straightforward relation exists between hospital volume and surgical outcome. METHODS: Two data sets collected nationwide in Japan were retrospectively studied by statistical analysis: 18055 patients operated on during an average year for the most recent 4 years for an analysis of hospital mortality and 3233 patients operated on duraing 1989 for analysis of the 5-year survival rate. First, we examined the correlations of hospital volume with each outcome, which was estimated using the empirical Bayes (EB) method to stabilize any large variation due to the small sample. Then we estimated the volume effects using generalized regression models. Perioperative mortality and the 5-year survival rate in relation to hospital volume were measured. RESULTS: No statistically significant correlations between hospital volume and each outcome have been shown (EB mortality 1.51%, P = 0.0566; 5-year survival 48.9%, P = 0.333). Regarding the volume effect, only the lowest-volume subgroup showed a statistically significant higher perioperative mortality rate and a lower 5-year survival rate compared to the highest-volume counterpart, but many hospitals of the lowest-volume subgroup had operated on more patients with advanced-stage disease. CONCLUSIONS: No straightforward correlation was seen between hospital volume and surgical outcome. Hospital volume may not be a suitable single tool to predict the outcome of lung cancer surgery. We should reserve the conclusion that low-volume hospitals offer less effective surgery. We need a risk-adjusted database to study this topic.
OBJECTIVE: This study aimed at analyzing results of surgery for lung cancer in relation to hospital volume to see if a straightforward relation exists between hospital volume and surgical outcome. METHODS: Two data sets collected nationwide in Japan were retrospectively studied by statistical analysis: 18055 patients operated on during an average year for the most recent 4 years for an analysis of hospital mortality and 3233 patients operated on duraing 1989 for analysis of the 5-year survival rate. First, we examined the correlations of hospital volume with each outcome, which was estimated using the empirical Bayes (EB) method to stabilize any large variation due to the small sample. Then we estimated the volume effects using generalized regression models. Perioperative mortality and the 5-year survival rate in relation to hospital volume were measured. RESULTS: No statistically significant correlations between hospital volume and each outcome have been shown (EB mortality 1.51%, P = 0.0566; 5-year survival 48.9%, P = 0.333). Regarding the volume effect, only the lowest-volume subgroup showed a statistically significant higher perioperative mortality rate and a lower 5-year survival rate compared to the highest-volume counterpart, but many hospitals of the lowest-volume subgroup had operated on more patients with advanced-stage disease. CONCLUSIONS: No straightforward correlation was seen between hospital volume and surgical outcome. Hospital volume may not be a suitable single tool to predict the outcome of lung cancer surgery. We should reserve the conclusion that low-volume hospitals offer less effective surgery. We need a risk-adjusted database to study this topic.
Authors: Antonio E Martin-Ucar; David A Waller; Jane L Atkins; Daniel Swinson; Ken J O'Byrne; Mick D Peake Journal: Lung Cancer Date: 2004-11 Impact factor: 5.705