Original article Operative mortality after colorectal resection in the Netherlands A. F. Engel 1 , J. L. T. Oomen 1 , D. L. Knol 2 and M. A. Cuesta 3 1 Department of Surgery, Zaans Medical Centre, Zaandam, and Departments of 2 Clinical Epidemiology and Biostatistics and 3 Surgery, Vrije Universiteit Medisch Centrum, Amsterdam, The Netherlands Correspondence to: Dr A. F. Engel, PO Box 210, 1500 EE, Zaandam, The Netherlands (e-mail: Engel.a@deheel.nl) Background: The aim of this study was to quantify factors related to operative mortality after colorectal resection in the Netherlands. Methods: Multilevel logistic regression modelling was used. Institutional effects were calculated with and without adjustment for specific patient (age, sex, urgency of operation) and hospital (number of procedures, type of hospital) characteristics. All adult Dutch patients who underwent primary colorectal resection between 1994 and 1999 were included, except those who had (sub)total colectomy or local rectal resection. Results: A total of 67 594 patients underwent colorectal resection. The in-hospital mortality rate was 7·0 per cent (elective 3·9 per cent, acute 14·3 per cent). Acute operation (odds ratio 3·89) and age (odds ratios 2·63, 5·23 and 10·13 for patients aged 50–69, 70–79 and 80 or more years respectively compared with those aged less than 50 years) had the strongest effects, followed by male sex (odds ratio 1·48) and type of hospital. There was no difference in operative mortality rate between low-, medium- and high-volume hospitals. Conclusion: In the Netherlands, advanced age and acute operation are by far the most important factors related to operative mortality after colorectal resection. Male sex and type of hospital have only a modest effect, and there is no discernible effect of hospital volume. Presented in part to a meeting of the Dutch Association of General Surgeons, Veldhoven, The Netherlands, May 2003, and at the United European Gastroenterology Week, Prague, Czech Republic, September 2004 Paper accepted 26 June 2005 Published online 4 November 2005 in Wiley InterScience (www.bjs.co.uk). DOI: 10.1002/bjs.5153 Introduction Colorectal resection forms a large part of the general surgical workload in the Netherlands. The need for outcome studies is increasing because of the increasing cost of healthcare and the role of funding agencies in the financing and distribution of healthcare. Accurate information is needed to guide long-term decisions regarding the structure and control of healthcare systems. The outcome of surgery may be one of the reasons underlying a change in healthcare distribution. In the USA, outcome of surgery, mainly measured as the operative mortality related to both hospital and surgeon experience in certain procedures, has become an important but also controversial subject. A volume – outcome relationship for some complex surgical procedures may exist, but conflicting studies have also been presented 1,2 . Some of the evidence that dates from the 1980s and 1990s is flawed because appropriate statistical analyses were not used 3 . In earlier studies no statistical allowance was made for the clustering of patients within one hospital. To allow for clustering of patients with shared characteristics in one hospital, multilevel logistic regression (MLR) modelling may be used. This is a relatively new technique that has been used to analyse pupil performance in a school system and patient outcome in a healthcare system 4–7 . The aim of this study was to quantify factors related to operative mortality after colorectal resection in the Netherlands, using MLR modelling. Patients and methods In the Netherlands, independent hospital-based medical registration departments collect a set of data after discharge or death of a hospitalized patient. Data sets with incomplete fields are not processed and cannot be filed. The complete Copyright  2005 British Journal of Surgery Society Ltd British Journal of Surgery 2005; 92: 1526–1532 Published by John Wiley & Sons Ltd