Early prediction of severity of acute pancreatitis using peritoneal lavage.

Diagnostic peritoneal lavage was attempted in 96 out of a consecutive series of 168 attacks of acute pancreatitis, in order to evaluate its ability to predict severe disease. Lavage was successful in 89 instances, and resulted in relief of pain in many patients, and in one complication. Lavage indicated (by the presence of numerous organisms) that the diagnosis was incorrect in 1 patient with suspected acute pancreatitis; the correct diagnosis, biliary peritonitis, was revealed at post mortem examination. The severity of the attack was predicted by the clinician prior to lavage, and each attack was classified as mild or severe when the patient left hospital according to a simple clinical scheme. Although there was a strong correlations between severe attacks and the concentrations of G. O. T., albumin, protein, magnesium, potassium and phosphate in the lavage fluid, a simple and immediate classification based upon the physical features of the free peritoneal fluid and the lavage return fluid was adopted. The presence of brown free fluid, or more than 10 ml of free fluid, or mild-straw coloured lavage was taken as the lavage criterion of a severe attack. Of the 89 attacks with a successful lavage, 68 were classified as mild and 21 as severe according to the clinical outcome. Clinical assessment of the attack on the day of admission to hospital successfully predicted only 8 (38%) of the severe attacks, but 17 (76%) were correctly predicted by lavage, carried out at an average time of 7 hours after admission. False positive results of lavage were seen in only 4 (6%) of the 68 mild attacks. All attacks in which a shock-like illness developed, or which resulted in death (4) were predicted as severe by lavage. We feel that diagnostic lavage especially suited to select patients who may be expected to benefit from therapeutic peritoneal lavage.