Predicting outcomes in paraquat poisonings

The purpose of this study was to analyze the factors in patients with paraquat poisoning at admission which were able to predict the chance of survival. From January 1983 to September 1988, 24 patients were admitted to Ramathibodi Hospital because of paraquat poisoning. Nine were male and 15 were female, aged 29.9 +/- 18 y (range from 12 to 73 y). The amount of paraquat ingested was 57.4 +/- 107.9 ml. Patients were treated with Fuller's Earth and forced diuresis. The serum paraquat levels obtained in 15 patients were 0.17 +/- 0.34 ug/ml at 24.7 +/- 6.6 h after ingestion. Seventeen patients (71%) died. A prediction rule was developed by discriminant analysis and multiple logistic linear regression to predict outcome in paraquat poisoning. Age, amount of paraquat ingested, and white blood cell count at admission were identified as positively correlated with mortality. The discriminant function which classified the patients into the expired group if the function was negative, and vice versa, was 0.027 age + 0.022 ingested amount + 0.0002 WBC - 4.06. The model was validated by the jackknife method and prospectively. Seventy-nine percent of patients in the study group were classified correctly when using the jackknife method. Prospectively, The prediction rule correctly classified all 9 newly admitted paraquat poisoning patients. We conclude that age, ingested amount, and white blood cell count at admission are prognostically significant in paraquat poisoning. The prediction rule based on the above prognostic factors can appropriately forecast outcome in paraquat-poisoned patients.