New clinical prediction rule (the Pediatric Appendicitis Laboratory Score) accurately rules out appendicitis in children

Référence

Benito J, Fernandez S, Gendive M, et al. A new clinical score to identify children at low risk for appendicitis. Am J Emerg Med 2020;38(3):554-561.

Sommaire

These authors prospectively enrolled children 2 years to 14 years of age who visited a single emergency department for suspected appendicitis. The research team gathered clinical and laboratory data in a structured manner. From the 361 children they enrolled, the authors randomly selected 278 to develop a clinical prediction rule, and then validated the rule on a mixed cohort of the 83 remaining children an additional 172 children seen in the same emergency department during an earlier time frame. The diagnosis of appendicitis was confirmed using pathology for those undergoing surgery and clinical follow-up 15 days after the index visit. Clouding the picture a bit, the authors also used ultrasound during the diagnostic process. Among the children in the derivation cohort, 35.9% had acute appendicitis compared with 49% in the validation group. After performing feats of statistical gymnastics, the authors created the PALabS, which consists of 6 parts: nausea (3 points), maximal pain in the right lower quadrant (4 points), absolute neutrophil count above 7500 per microliter (7 points), white blood cell count above 10,000 per microliter (4 points), C-reactive protein above 10 mg per liter (2 points), and calprotectin above 50 nanograms per milliliter (3 points). Overall, in the derivation group, the PALabS was accurate (area under the receiver operator characteristic curve 0.88). The researchers also applied 2 other scores to the derivation sample, the Pediatric Appendicitis Score and the Kharbanda Score. They found these scores to have less diagnostic accuracy. In the validation cohort, a PALabS of 6 or less was 99.2% sensitive (95% CI 95.6 - 99.9) and had a negative likelihood ratio of 0.03 (0.00 - 0.18). Although the authors did not report the specificity of the PALabS, based on my calculations, it was approximately 32% and the positive likelihood ratio was 3.2. In other words, this looks like a SnNOut (a very sensitive test that rules out the disease when the result result is negative). Now this clincial prediction rule needs to be tested in an independent cohort of patients.