文本框: A CASE CONTROL STUDY TO IDENTIFY PREDICTIVE FACTORS ASSOCIATED WITH POSITIVE BACTERIOLOGICAL STOOL CULTURE IN PAEDIATRIC PATIENTS ADMITTED FOR ACUTE DIARRHOEA Chung Pui Hong HONG KONG Objective: To determine the biological yield of bacteria stool culture in North District Hospital for patients aged from three months to five years for acute diarrhoea. To design clinical predictors for positive bacterial growth in general and in a more specific group of severe bacteria (Campylobacter, Shigella and pathogenic E. coli) which can potentially reduce the number of stool samples sent for culture. Method: One-year retrospective case-control study in North District Hospital. Subjects: Patients from three months to five years old who were admitted to the Paediatric department of North District Hospital through the Accident and Emergency Department from 1/1/1999 to 31/12/1999. Results: 378 patients were recruited. 71(18.8%) got positive bacterial culture in stool. Sex, age, abdomen pain, maximum white cell count, maximum absolute neutrophill count and fever were not related to bacterial culture positivity. Blood in stool (OR 5.49: 95%CI2.27 - 13.27), mucus in stool (OR 3.54: 95%CI 1.63-7.68), absence of vomiting (OR 6.43: 95%CI 2.8-14.76), bowel opening (BO) more than 7 times per day (OR 2.94: 95%CI 1.02-8.47) and season (summer- OR 3.38: 95%CI 1.06-10.81, and autumn - OR3.79: 95%CI 1.09-13.13) were found to be statistically significant in relation to positive bacterial growth in stool. Negative predictive value for individual clinical predictor varies from 86-93.4%. This logistic model has an overall predictive correctness of 88.06%. 25(6.6%) belonged to the severe bacteria infection group (Campylobacter, Shigella, pathogenic E.coli). Presence of any one of the clinical predictor (mucus in stool, blood in stool, absence of vomiting, BO more than 7 times per day, admitted on summer or autumn) will be good in predicting positive severe bacteria growth in stool. For this model, Sensitivity and Negative predictive value is 100%. Specificity is 24.3%, Positive predictive value is 9.7%). Conclusion: The results from this retrospective analysis were used to develop guidelines to reduce unwarranted stool cultures. 1973