Perception of surgical complications among patients, nurses and physicians: a prospective cross-sectional survey

227 patients, 143 nurses and 245 physicians independently rated the severity of 30 common post-operative complications on a numerical analogue scale from 0 (not severe at all) to 100 (extremely severe) while being blinded towards the Clavien-Dindo classification. We considered a difference in ratings of >10 to be clinically important in distinguishing between grades of severity and groups. We evaluated the level of reproducibility of responses by calculating intraclass correlation coefficients (ICC) and compared scores across severity grades and between groups using the generalized estimating equations.Reproducibility of the ratings was good for all three groups (ICCpatients 0.71 (95%-CI 0.64-0.76), ICCnurses 0.83 (0.78-0.87) and ICCphysicians 0.87 (0.83-0.90)). The participants' perceptions of the severity of complications reflected the Clavien-Dindo classification (median of grade I: 20 (IQR 10-30), grade II: 40 (31.3-52.5), grade IIIa: 50 (40-60), grade IIIb: 70 (60-75), grade IVa: 85 (80-90) and grade IVB: 95 (90-100)). Although patients' perception differed significantly from those of physicians (average difference -8.7 (95%-CI -10.4 to -6.9, p < 0.001) and nurses (difference -2.8 (-4.8 to -0.8, p = 0.007) they did not reach our thresholds for clinical importance.The severity of post-operative complications is perceived similarly by patients, nurses and physicians and reflects the Clavien-Dindo classification well. Our results support the use of Clavien-Dindo classification system as part of the shared or informed decision making process.A well-known editorial in the Lancet highlighted the poor methodology and lack of convincing outcome measures in most surgical studies [1]. Mortality and a variety of markers for morbidity are commonly used but there is an ongoing debate on how to define and standardize post-operative complications. This is well illustrated by a systematic review that found more than 40 different definitions of anastomotic leaks in 107 differen