Prospective observational cohort study on grading the severity of postoperative complications in global surgery research


T. E.F. Abbott, Barts and The London School of Medicine and Dentistry
A. Patel, Barts and The London School of Medicine and Dentistry
Tahania Ahmed, Barts and The London School of Medicine and Dentistry
Rupert M. Pearse, Barts and The London School of Medicine and Dentistry
K. E. Greaves, The Royal London Hospital
James Haddow, Foundation for Blood Research
Emmanuel Futier, Université Clermont Auvergne
Matthieu Biais, Centre Hospitalier Universitaire de Bordeaux
Karem Slim, Centre Hospitalier Universitaire de Clermont-Ferrand
Scott Beattie, Toronto General Hospital
Pierre Alain Clavien, UniversitatsSpital Zurich
Nicolas Demartines, Centre Hospitalier Universitaire Vaudois
Lee A. Fleisher, University of Pennsylvania Perelman School of Medicine
Mike Grocott, University Hospital Southampton NHS Foundation Trust
Andreas Hoeft, Universitäts-Klinikum Bonn und Medizinische Fakultät
Peter Holt, St Georges Vascular Institute
Rui Moreno, Centro Hospitalar de Lisboa Central
Naomi Pritchard, School of Clinical Medicine
Andrew Rhodes, St George’s, University of London
Duminda Wijeysundera, Toronto General Hospital
Matt Wilson, Washington Hospital Center
Kirsty Everingham, The Royal London Hospital
Russell Hewson, The Royal London Hospital
Marta Januszewska, The Royal London Hospital
Mandeep Kaur Phull, The Royal London Hospital
Richard Halliwell, Westmead Hospital
Jennifer Cope, Westmead Hospital
Mark Shulman, The Alfred
Paul Myles, The Alfred
Marissa Ferguson, The Alfred
Michael MacMahon, The Alfred
Werner Schmid, Medizinische Universitat Wien

Document Type

Journal Article

Publication Date



British Journal of Surgery








Background: The Clavien–Dindo classification is perhaps the most widely used approach for reporting postoperative complications in clinical trials. This system classifies complication severity by the treatment provided. However, it is unclear whether the Clavien–Dindo system can be used internationally in studies across differing healthcare systems in high-(HICs) and low-and middle-income countries (LMICs). Methods: This was a secondary analysis of the International Surgical Outcomes Study (ISOS), a prospective observational cohort study of elective surgery in adults. Data collection occurred over a 7-day period. Severity of complications was graded using Clavien–Dindo and the simpler ISOS grading (mild, moderate or severe, based on guided investigator judgement). Severity grading was compared using the intraclass correlation coefficient (ICC). Data are presented as frequencies and ICC values (with 95 per cent c.i.). The analysis was stratified by income status of the country, comparing HICs with LMICs. Results: A total of 44 814 patients were recruited from 474 hospitals in 27 countries (19 HICs and 8 LMICs). Some 7508 patients (16.8 per cent) experienced at least one postoperative complication, equivalent to 11 664 complications in total. Using the ISOS classification, 5504 of 11 664 complications (47.2 per cent) were graded as mild, 4244 (36.4 per cent) as moderate and 1916 (16.4 per cent) as severe. Using Clavien–Dindo, 6781 of 11 664 complications (58.1 per cent) were graded as I or II, 1740 (14.9 per cent) as III, 2408 (20.6 per cent) as IV and 735 (6.3 per cent) as V. Agreement between classification systems was poor overall (ICC 0.41, 95 per cent c.i. 0.20 to 0.55), and in LMICs (ICC 0.23, 0.05 to 0.38) and HICs (ICC 0.46, 0.25 to 0.59). Conclusion: Caution is recommended when using a treatment approach to grade complications in global surgery studies, as this may introduce bias unintentionally.