Multiple logistic regression analysis of plasma paraquat concentrations as a predictor of outcome in 375 cases of paraquat poisoning.

Abstract

Successful prediction of who may survive paraquat poisoning can prevent inappropriately aggressive treatment in those who have little hope of survival and those only minimally poisoned. We examined case records of patients admitted to one poisoning treatment unit over the last 5 years, and the English and French language literature on paraquat poisoning. Data were recorded from all patients where outcome and timed plasma paraquat concentrations were present. Of 375 patients (113 M, 62 F, 200 unknown), mean age 38.3 years (range 1-87 years), 49 had evidence of renal toxicity, and 41 received haemodialysis or charcoal haemoperfusion; 61 developed pulmonary sequelae; and 44 had lesions in the upper gastrointestinal tract. Median time from ingestion to death in the 241 deaths reported was 270 h (range 3-720 h). We plotted log(plasma paraquat concentration) against log(h since ingestion). The predicted probability of survival for any specified time and concentration was exp(logit)/[1 + exp(logit)], where logit = 0.58-2.33 x log(plasma paraquat)-1.15 x log(h since ingestion). This equation may be helpful in predicting who will survive after ingestion of paraquat up to at least 200 h after ingestion, and can now be used as a research tool for studies on efficacy of treatment of paraquat poisoning.