Derivation and validation of a practical Bedside Score for the diagnosis of cholecystitis.

ABSTRACT

OBJECTIVE: We sought to develop a practical Bedside Score for the diagnosis of cholecystitis and test its accuracy against the Tokyo Guidelines (TG13). METHODS: We conducted a retrospective study of 438 patients undergoing urban, academic Emergency Department (ED) evaluation of RUQ pain. Symptoms, physical signs, ultrasound signs, and labs were scoring system candidates. A random split-sample approach was used to develop and validate a new clinical score. Multivariable regression analysis using development data was conducted to identify predictors of cholecystitis. Cutoff values were chosen to ensure positive/negative predictive values (PPV, NPV) of at least 0.95. The score was externally validated in 80 patients at a different hospital undergoing RUQ pain evaluation. RESULTS: 230 patients (53%) had cholecystitis. Five variables predicted cholecystitis and were included in the scores gallstones, gallbladder thickening, clinical or ultrasonographic Murphy's sign, RUQ tenderness, and post-prandial symptoms. A clinical prediction score was developed. When dichotomized at 4, overall accuracy for acute cholecystitis was 90% for the development cohort, 82% and 86% for the internal and external validation cohorts; TG13 accuracy was 62%-79%. CONCLUSIONS: A clinical prediction score for cholecystitis demonstrates accuracy equivalent to TG13. Use of this score may streamline work-up by decreasing the need for comprehensive ultrasound evaluation and CRP measurement and may shorten ED length of stay.