Value of stool cultures.

An evaluation of stool cultures in a large teaching hospital where this test had been requested for 2468 patients in 1977 showed that only 58 patients (2.4%) had positive stool cultures (Salmonella or Shigella). Likelihood ratios computed from data from the medical records of 54 patients with positive stool cultures and a control group of 189 patients with negative cultures suggested that the best clinical predictor of a positive stool culture was the combination of persistent diarrhoea (duration more than 24 h), fever, and either blood in the stool or abdominal pain with nausea or vomiting. These findings were present in 20 of 52 patients with positive cultures, but in no patients with negative cultures. 53 patients had a positive result on the first culture done; 1 patient required 2 cultures. Microscopic examination for faecal leucocytes was not a useful predictor. In only 10 of 227 patients did culture results (positive or negative) lead to appropriate changes in treatment. The requirement that request forms be signed by a doctor, the limitation to no more than 2 stool cultures per illness, and consideration of the clinical predictors and their likelihood ratios for positive stool cultures would promote a more discerning use of this low-cost but low-yield diagnostic test.