A total of 42 studies met the initial inclusion criteria, of which 25 met the quality criteria and were included in the analysis. One additional study provided precision data only. The total number of participants was unclear.
In terms of study quality, the majority of the included studies (24 out of 25) were level 3; the remaining study was level 1.
Precision (one study).
Inter-examiner precision was poor; when assessed over seven clinical findings, only rebound tenderness was found to produce a kappa statistic of greater than 0.5 (k=0.54).
Accuracy of symptoms.
The presence of right lower quadrant pain had minimal impact on the likelihood of appendicitis (LR+ 1.2, 95% confidence interval, CI: 1.0, 1.5) while its absence decreased the likelihood (LR- 0.56, 95% CI: 0.43, 0.73), based on three level 3 studies.
The presence of fever increased the likelihood of appendicitis (LR+ 3.4, 95% CI: 2.4, 4.8) while its absence decreased the likelihood (LR- 0.32, 95% CI: 0.16, 0.64), based on data from the level 1 study; this was less useful when data from the four level 3 studies were considered (LR+ 1.2, 95% CI: 1.1, 1.4; LR- 0.53, 95% CI: 0.29, 0.97).
The other symptoms considered were less useful.
Accuracy of signs.
Rebound tenderness increased the likelihood of appendicitis (LR+ 3.0, 95% CI: 2.3, 3.9) while its absence decreased the likelihood (LR- 0.28, 95% CI: 0.14, 0.55), based on data from three level
The other signs considered were less useful.
Accuracy of symptom-sign combinations.
The Alvarado, or MANTRELS, score was the most commonly evaluated (three level 3 studies). A score of seven or higher increased the likelihood of appendicitis (LR+ 4.0, 95% CI: 3.2, 4.9) while a score of less than seven decreased the likelihood (LR- 0.20, 95% CI: 0.09, 0.41).
Accuracy of laboratory tests.
A white blood cell count of less than 10,000 per microlitre, or an absolute neutrophil count of less than 6,750 per microlitre, decreased the likelihood of appendicitis: LR- 0.22 (95% CI: 0.17, 0.30; based on four level 3 studies) and 0.06 (95% CI: 0.03, 0.16; based on one level 1 study), respectively.
The other laboratory tests considered were less useful.