In the United States, appendectomy is the most common surgical procedure in children, with more than 70,000 performed each year. To aid in diagnosis, computed tomography (CT) commonly is used, but the method raises concerns about radiation exposure. The exposure is not benign: according to one estimate, 500 in 600,000 children undergoing CT will develop cancer as a direct result of radiation. For this reason, physicians desire a reliable clinical decision tool that can avoid indiscriminate use of CT. To derive and validate clinical criteria to diagnose appendicitis without using CT, Kharbanda and colleagues conducted a 15-month study of children presenting to the emergency department with abdominal pain.

Of 4,140 patients presenting with abdominal pain, 601 patients three to 18 years of age were enrolled; of these, 35 percent were diagnosed with appendicitis. After examination, 221 patients (84 percent) underwent CT, ultrasonography, or both. Seventeen of these patients taken to the operating room did not have appendicitis, whereas four of six patients assessed as healthy and discharged from the hospital actually had appendicitis. Two decision rules were developed: one rule used a logistic regression tool and the other a recursive partitioning tool. Applying these rules to the study population would have reduced CT use by 23 and 20 percent, respectively. The logistic regression rule would have missed two patients with appendicitis and the recursive partitioning tool would have missed one patient. None of these cases involved a perforated appendix.

The logistic regression rule used a six-part score to identify low-risk patients: nausea (2 points), focal right lower quadrant pain (2 points), migration of pain (1 point), difficulty walking (1 point), rebound tenderness or pain with percussion (2 points), and an absolute neutrophil count greater than 6.75 × 1,000 per μL (.0067 × 10⁹ per L [6 points]). With a score of 5 or fewer points, this rule had a sensitivity of 96.3 percent and negative predictive value of 95.65 percent. The recursive partitioning rule, which identifies low-risk patients as those who have a combined absolute neutrophil count less than 6.75 × 1,000 per μL, absence of nausea (which included emesis or anorexia), and absence of maximal tenderness in the right lower quadrant, also effectively ruled out appendicitis. The sensitivity for this rule was 98.1 percent with a negative predictive value of 97.5 percent.

The authors recommend the recursive partitioning rule as the better option because of its simplicity. They conclude that children identified by either model as being at low risk should be observed rather than undergo CT.