Objective:To investigate the predictive value of early indicators changes in blood test on the prognosis of patients with acute paraquat poisoning.Methods:The clinical data of patients with acute paraquat poisoning admitted to emergency department of Shengjing Hospital of China Medical University from January 2012 to June 2019 were retrospectively analyzed. The changes of blood test indexes within 24 hours after admission were collected, including white blood cell count (ΔWBC), neutrophils count (ΔNE), lymphocytes count (ΔLY), monocytes count (ΔMO), arterial partial pressure of oxygen (ΔPaO 2), arterial partial pressure of carbon dioxide (ΔPaCO 2), arterial blood pH (ΔpH), bicarbonate radical (ΔHCO 3-), base excess (ΔBE), lactate (ΔLac), total protein (ΔTP), albumin (ΔALB), alanine aminotransferase (ΔALT), aspartate aminotransferase (ΔAST), total bilirubin (ΔTBil), direct bilirubin (ΔDBil), blood urea nitrogen (ΔBUN), serum creatinine (ΔSCr), serum calcium concentration (ΔCa 2+), and serum potassium concentration (ΔK +). Multivariate Logistic regression was used to analyze the risk factors of prognosis in patients with acute paraquat poisoning, and receiver operating characteristic (ROC) curve was drawn to analyze the predictive value of ROC curve for the death of patients with paraquat poisoning. Results:A total of 251 patients with acute paraquat poisoning were included, with 99 cases dead, and the mortality was 39.4%. The increase of the markers including ΔWBC, ΔLac, ΔALT, ΔAST, ΔTBil, ΔDBil, ΔBUN, ΔSCr and ΔK + within 24 hours of admission in the death group were significantly higher than that in the survival group; the decrease of the markers including ΔPaCO 2, ΔHCO 3-, ΔBE, ΔTP, and ΔALB in the death group were significantly greater than those in the survival group. The variables with statistical significance in the above single factor analysis were included in the multivariate Logistic regression analysis. The results showed that ΔLac, ΔSCr and ΔK + were independent risk factors for the prognosis of patients with acute paraquat poisoning [odds ratio ( OR) and 95% confidence interval (95% CI) were 1.662 (0.997-2.772), 1.045 (1.010-1.083) and 4.555 (1.190-17.429), respectively, all P < 0.05]. The area under the ROC curve (AUC) of ΔLac, ΔSCr and ΔK + for predicting death of patients with acute paraquat poisoning was 0.639 (95% CI was 0.505-0773), 0.811 (95% CI was 0.704-0.917), and 0.649 (95% CI was 0.519-0.779), respectively. When the cut-off of ΔLac was 1.85 mmol/L, the sensitivity was 87.9%, the specificity was 47.7%, and the diagnostic accuracy was 70.2%; when the cut-off of ΔSCr was 37.75 μmol/L, the sensitivity was 84.4%, the specificity was 77.9%, and the diagnostic accuracy was 80.5%; when the cut-off of ΔK + was 0.42 mmol/L, the sensitivity was 36.6%, the specificity was 90.7%, and the diagnostic accuracy was 68.3%. The efficiency of combination of ΔLac, ΔSCr, and ΔK + was greater than a single indicator in predicting death of patients with acute paraquat poisoning, with AUC of 0.911, and 95% CI of 0.834-0.989. Conclusions:ΔLac, ΔSCr, ΔK + within 24 hours of admission were all independent risk factors for the prognosis of patients with acute paraquat poisoning. ΔSCr > 37.75 μmol/L within 24 hours of admission would predict a poor prognosis in the patients with acute paraquat poisoning. Combined analysis of ΔLac, ΔSCr, and ΔK + can predict the prognosis of paraquat poisoning patients more accurately than single index.