Decision analysis in the clinical and imaging diagnosis of acute cholecystitis

The importance of clinical, laboratory and imaging data in the diagnosis of acute cholecystitis (AC) was studied in 825 patients with right upper quadrant pain hospitalized in the Surgical Clinic of the Fundeni Hospital--Bucharest, between January 1, 1986 and June 30, 1988. A number of 21 parameters were analysed in each case. Of these 825 patients, 259 were considered after surgery as AC. These 259 cases were divided, after the microscopical examination of the surgically-obtained specimens, into two groups: 1) pathologically confirmed AC (137 cases) and 2) pathologically non-confirmed AC (122 cases). The importance of every parameter in establishing a histologically confirmed diagnosis of AC was determined by the diagnostic probability calculated according to Bayes'theorem. The hierarchy of the value of parameters in the diagnosis of AC was based on their capacity to distinguish between the cases histologically confirmed and those detected on surgery, but without microscopically demonstrated changes of AC. The same decision criterion was used in building the decision trees in the exploration of the cases of presumed AC. In the 825 cases with right upper quadrant pain, the main and most frequent cause was chronic calculous cholecystitis (31.8%), followed by AC pathologically confirmed (16.6%), AC non-confirmed (14.7%) and chronic acalculous cholecystitis (12.4%). The most useful parameters in distinguishing between pathologically confirmed AC and pathologically non-confirmed AC were: 1) sudden onset of pain; 2) mild resistance to abdominal palpation; 3) frank peritoneal irritation; 4) stone impacted in the gallbladder neck (ultrasonography); 5) fever; 6) palpable gallbladder; 7) lithiasis (ultrasonography); 8) gallbladder wall with double outline (ultrasonography). Ultrasonography supplied a diagnostic probability of 85% for the correct diagnosis of AC in cases without a clinical picture suggestive for AC. The decision tree analysis supported the same conclusion: only ultrasonography gives a good distinction between pathologically confirmed AC and pathologically non-confirmed AC.