Diagnostic Accuracy of History, Physical Examination, Laboratory Tests, and Point‐of‐care Ultrasound for Pediatric Acute Appendicitis in the Emergency Department: A Systematic Review and Meta‐analysis

BACKGROUND: Acute appendicitis (AA) is the most common surgical emergency in children. Accurate and timely diagnosis is crucial but challenging due to atypical presentations and the inherent difficulty of obtaining a reliable history and physical examination in younger children. OBJECTIVES: The aim of this study was to determine the utility of history, physical examination, laboratory tests, Pediatric Appendicitis Score (PAS) and Emergency Department Point-of-Care Ultrasound (ED-POCUS) in the diagnosis of AA in ED pediatric patients. We performed a systematic review and meta-analysis and used a test-treatment threshold model to identify diagnostic findings that could rule in/out AA and obviate the need for further imaging studies, specifically computed tomography (CT) scan, magnetic resonance imaging (MRI), and radiology department ultrasound (RUS). METHODS: We searched PubMed, EMBASE, and SCOPUS up to October 2016 for studies on ED pediatric patients with abdominal pain. Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2) was used to evaluate the quality and applicability of included studies. Positive and negative likelihood ratios (LR+ and LR-) for diagnostic modalities were calculated and when appropriate data was pooled using Meta-DiSc. Based on the available literature on the test characteristics of different imaging modalities and applying the Pauker-Kassirer method we developed a test-treatment threshold model. RESULTS: Twenty-one studies were included encompassing 8,605 patients with weighted AA prevalence of 39.2%. Studies had variable quality using the QUADAS-2 tool with most studies at high risk of partial verification bias. We divided studies based on their inclusion criteria into two groups of "undifferentiated abdominal pain" and abdominal pain "suspected of AA." In patients with undifferentiated abdominal pain, history of "pain migration to right lower quadrant (RLQ)" (LR+ = 4.81, 95% confidence interval CI = 3.59-6.44) and presence of "cough/hop pain" in the physical examination (LR+ = 7.64, 95% CI = 5.94-9.83) were most strongly associated with AA. In patients suspected of AA none of the history or laboratory findings were strongly associated with AA. Rovsing's sign was the physical examination finding most strongly associated with AA (LR+ = 3.52, 95% CI = 2.65-4.68). Among different PAS cutoff points, PAS ≥ 9 (LR+ = 5.26, 95% CI = 3.34-8.29) was most associated with AA. None of the history, physical examination, laboratory tests findings, or PAS alone could rule in or rule out AA in patients with undifferentiated abdominal pain or those suspected of AA. ED-POCUS had LR+ of 9.24 (95% CI = 6.24-13.28) and LR- of 0.17 (95% CI = 0.09-0.30). Using our test-treatment threshold model, positive ED-POCUS could rule in AA without the use of CT and MRI, but negative ED-POCUS could not rule out AA. CONCLUSION: Presence of AA is more likely in patients with undifferentiated abdominal pain migrating to the RLQ or when cough/hop pain is present in the physical examination. Once AA is suspected, no single history, physical examination, laboratory finding, or score attained on PAS can eliminate the need for imaging studies. Operating characteristics of ED-POCUS are similar to those reported for RUS in literature for diagnosis of AA. In ED patients suspected of AA, a positive ED-POCUS is diagnostic and obviates the need for CT or MRI while negative ED-POCUS is not enough to rule out AA.