Influence of study design on reported mortality and morbidity rates after abdominal aortic aneurysm repair

Background The mortality and morbidity rates of elective abdominal aortic a neurysm (AAA) surgery, as reported over the past 12 years, were graded and analysed by levels of evidence. Methods Articles on elective AAA surgery published between 1985 and 1996 we re retrieved and classified into five levels of evidence. Level 1 contains prospective studies and is subdivided into population-based (level 1a) and hospital-based (level 1b) studies. Level 2 includes retrospective studies, subdivided into population-based studies (level 2a), hospital-based studies (level 2b) and hospital-based studies concerning a specified group of sele cted patients (level 2c). Operative mortality and systemic and local/vascul ar complication rates with 95 per cent confidence intervals were calculated for each level of evidence. Results Seventy-two articles describing a total of 37 654 patients could be included: two level la studies (692 patients), nine level 1b studies (1677 patients), 13 level 2a studies (21 409 patients), 32 level 2b studies (12 019 patients) and 10 level 2c studies (1857 patients). The mean 30-day mort ality rates of the two population-based levels were similar: 8.2 (95 per ce nt confidence interval 6.4-10.6) per cent for the prospective (1a) and 7.4 (7.0-7.7) per cent for the retrospective (2a) series. These figures were si gnificantly higher than the remarkably similar hospital-based mortality rat es: 3.8 (3.0-4.8) per cent for the prospective (1b). 3.8 (3.5-4.2) per cent for the retrospective (2b) and 3.5 (2.8-4.4) per cent for selected patient group (2c) studies. The most frequent complication was of cardiac origin. In the population-based series the cardiac complication rates were 10.6 (8. 5-13.2) and 11.1 (9.1-13.6) per cent for levels 1a and 2a respectively. Thi s compared well with 12.0 (10.5-13.9) per cent for the prospective hospital -based series (level 1b). The cardiac complication rates in the retrospecti ve hospital-based studies were significantly lower: 8.9 (8.4-9.5) and 6.1 ( 4.9-7.6) per cent for levels 2b and 2c respectively. Conclusion There is a clear and consistent disagreement in reported mortali ty rates between hospital-based and population-based studies of elective su rgery for AAA. Prospective studies give the best documentation of postopera tive morbidity.