Categorisation of Complications and Validation of the Clavien Score for Percutaneous Nephrolithotomy

Background: Although widely used, the validity and reliability of the Clavien classification of postoperative complications have not been tested in urologic procedures, such as percutaneous nephrolithotomy (PCNL). Objective: To validate the Clavien score and categorise complications of PCNL. Design, setting, and participants: Data for 528 patients with complications after PCNL were used to create a set of 70 unique complication-management combinations. Clinical case summaries for each complication-management combination were compiled in a survey distributed to 98 urologists, who rated each combination using the Clavien classification. Outcome measurements and statistical analysis: Interrater agreement for Clavien scores was estimated using Fleiss' kappa (kappa). The relationship between Clavien score and the duration of postoperative hospital stay was analysed using multivariate nonlinear regression models that adjusted for operating time, preoperative urine microbial culture, presence of staghorn stone, and use of postoperative nephrostomy tube. Results and limitations: Overall interrater agreement in grading postoperative complications was moderate (kappa = 0.457; p < 0.001). Agreement was highest for Clavien score 5 and decreased with lower Clavien scores. Higher agreement was found for Clavien scores 3 and 4 than in subcategories of these scores. Postoperative stay increased with higher Clavien scores and was unaffected by inherent differences between study centres. A standard list of post-PCNL complications and their corresponding Clavien scores was created. Conclusions: Although the Clavien classification demonstrates high validity, interrater reliability is low for minor complications. To improve the reliability and consistency of reporting adverse outcomes of PCNL, we have assigned Clavien scores to complications of PCNL. (C) 2012 European Association of Urology. Published by Elsevier B.V. All rights reserved.