Bayesian analysis of the reliability of peritoneal lavage

The popular methods of reporting the reliability of a test--sensitivity and specificity--do not tell the practicing physician what the chance is that a patient has the disease if the results from the particular test for that disease are positive or negative. Bayesian analysis, however, does answer this question. In this study, the data from 29 studies of peritoneal lavage were analyzed using the Bayesian method. The probability of an intra-abdominal injury being present if findings from lavage were positive ranged from 0.448 to 1.0. The probability of an intra-abdominal injury being present if findings from lavage were negative ranged from zero to 0.286. These probabilities varied with the criteria for positivity, technique used and type of trauma. In general, results of Bayesian analysis showed that findings from peritoneal lavage were most reliable in diagnosing the patient with blunt trauma when the standard positive criteria and open technique were used.