Development of a practical prediction scoring system for severe acute organophosphate poisoning

Prediction of acute organophosphate poisoning severity using machine learning techniques

Poisoning with organophosphate compounds is a significant public health risk, especially in developing countries. Considering the importance of early and accurate prediction of organophosphate poisoning prognosis, the aim of this study was to develop a machine learning-based prediction model to predict the severity of organophosphate poisoning. The data of patients with organophosphate poisoning were retrospectively extracted and split into training and test sets in a ratio of 70:30. The feature selection was done by least absolute shrinkage and selection operator method. Selected features were fed into five machine learning techniques, including Histogram Boosting Gradient, eXtreme Gradient Boosting, K-Nearest Neighborhood, Support Vector Machine (SVM) (kernel = linear), and Random Forest. The Scikit-learn library in Python programming language was used to implement the models. Finally, the performance of developed models was measured using ten-fold cross-validation methods and some evaluation criteria with 95 % confidence intervals. A total of 1237 patients were used to train and test the machine learning models. According to the criteria determining severe organophosphate poisoning, 732 patients were assigned to group 1 (patients with mild to moderate poisoning) and 505 patients were assigned to group 2 (patients with severe poisoning). With an AUC value of 0.907 (95 % CI 0.89-0.92), the model developed using XGBoost outperformed other models. Feature importance evaluation found that venous blood gas-pH, white blood cells, and plasma cholinesterase activity were the top three variables that contribute the most to the prediction performance of the prognosis in patients with organophosphate poisoning. XGBoost model yield an accuracy of 90.1 % (95 % CI 0.891-0.918), specificity of 91.4 % (95 % CI 0.90-0.92), a sensitivity of 89.5 % (95 % CI 0.87-0.91), F-measure of 91.2 % (95 % CI 0.90-0.921), and Kappa statistic of 91.2 % (95 % CI 0.90-0.92). The machine learning-based prediction models can accurately predict the severity of organophosphate poisoning. Based on feature selection techniques, the most important predictors of organophosphate poisoning were VBG-pH, white blood cell count, plasma cholinesterase activity, VBG-BE, and age. The best algorithm with the highest predictive performance was the XGBoost classifier.

Latent class trajectories of biochemical parameters and their relationship with risk of mortality in ICU among acute organophosphorus poisoning patients

Acute poisoning is a global public health challenge. Several factors played role in high mortality among acute organophosphorus poisoning (OP) poisoning patients including clinical, vitals, and biochemical properties. The traditional analysis techniques use baseline measurements whereas latent profile analysis is a person-centered approach using repeated measurements. To determine varying biochemical parameters and their relationship with intensive care unit (ICU) mortality among acute organophosphorus poisoning patients through a latent class trajectory analysis. The study design was a retrospective cohort and we enrolled data of 299 patients admitted between Aug’10 to Sep’16 to ICU of Dr. Ruth K. M. Pfau, Civil Hospital, Karachi. The dependent variable was ICU-mortality among OP poisoning patients accounting for ICU stay, elapsed time since poison ingestion, age, gender, and biochemical parameters (including electrolytes (potassium, chloride, sodium), creatinine, urea, and random blood sugar). Longitudinal latent profile analysis is used to form the trajectories of parameters. In determining and comparing the risk of ICU-mortality we used Cox-Proportional-Hazards models, repeated measures and trajectories were used as independent variables. The patients’ mean age was 25.4 ± 9.7 years and ICU-mortality was (13.7%, n = 41). In trajectory analysis, patients with trajectories (normal-increasing and high-declining creatinine, high-remitting sodium, normal-increasing, and high-remitting urea) observed the highest ICU-mortality i.e. 75% (6/8), 67% (2/3), 80% (4/5), 75% (6/8), and 67% (2/3) respectively compared with other trajectories. On multivariable analysis, compared with patients who had normal consistent creatinine levels, patients in normal-increasing creatinine class were 15 times [HR:15.2, 95% CI 4.2–54.6], and the high-declining class was 16-times [HR 15.7, 95% CI 3.4–71.6], more likely to die. Patients in with high-remitting sodium, the trajectory was six-times [HR 5.6, 95% CI 2.0–15.8], normal-increasing urea trajectory was four times [HR 3.9, 95% CI 1.4–11.5], and in extremely high-remitting urea trajectory was 15-times [HR 15.4, 95% CI 3.4–69.7], more likely to die compared with those who were in normal-consistent trajectories of sodium and urea respectively. Among OP poisoning patients an increased risk of ICU-mortality were significantly associated with biochemical parameters (sodium, urea, creatinine levels) using latent profile technique.

Clinical Analysis of Acute Organophosphorus Pesticide Poisoning and Successful Cardiopulmonary Resuscitation: A Case Series

Acute organophosphorus pesticide poisoning (AOPP) with cardiac arrest has an extremely high mortality rate, and corresponding therapeutic strategies have rarely been reported. Therefore, this study aimed to explore the prognostic factors and effective treatments of AOPP-related cardiac arrest. This retrospective study was conducted in our department in the years 2018–2021. We conducted a descriptive analysis of the clinical manifestations, rescue strategies, and prognosis of patients with AOPP who had experienced cardiac arrest and successful cardiopulmonary resuscitation. This study included six cases of patients with AOPP in addition to cardiac arrest; in four cases, cardiac arrest occurred <12 h after ingestion, and in two, cardiac arrest occurred more than 48 h after ingestion. Five patients had not undergone hemoperfusion therapy before cardiac arrest, and all six were treated with atropine during cardiopulmonary resuscitation and subsequent pralidoxine. Four patients recovered and were discharged from the hospital, one died in our department, and one was transferred to a local hospital and died there 2 h later. The last two patients had severe pancreatic injuries and disseminated intravascular coagulation. This, along with their death, might have been related to their prognosis. Cardiac arrest can occur in patients with severe AOPP for whom antidote administration was insufficient or not timely. Application of atropine and pralidoxine in a timely manner after cardiac arrest following AOPP is the key to successful treatment. This study provides useful guidelines for the treatment of similar cases in the future.

Prognostic value of the neutrophil-to-lymphocyte ratio in acute organophosphorus pesticide poisoning

This study investigates the ability of blood neutrophil-to-lymphocyte ratio (NLR) to predict acute organophosphorus pesticide poisoning (AOPP). Clinical data of 385 patients with AOPP were obtained within 24 h of admission, and NLR values were calculated based on neutrophil and lymphocyte counts. The patients were divided into two groups - good and poor - based on prognosis. Poor prognosis included in-hospital death and severe poisoning. The factors affecting prognosis were analyzed by logistic regression analysis, and the prognostic value of NLR was evaluated using the area under the receiver operating characteristic curve (AUC). Univariate logistic regression analysis showed that NLR levels, serum cholinesterase, and creatinine levels were good predictors of AOPP. Multivariate logistic regression analysis showed that high NLR was an independent risk factor for severe poisoning (adjusted odds ratio [AOR], 1.13; 95% CI, 1.10-1.17; p < 0.05) and in-hospital mortality (AOR, 1.07; 95% CI, 1.03-1.11; p < 0.05). NLR values >13 and >17 had a moderate ability to predict severe poisoning and in-hospital mortality, respectively (AUC of 0.782 [95% CI, 0.74-0.824] and 0.714 [95% CI, 0.626-0.803], respectively). Our results show that high NLR at admission is an independent indicator of poor prognosis in AOPP and can be used to optimize treatment and manage patients.

Trajectories of vital status parameters and risk of mortality among acute organophosphorus poisoning patients - a latent class growth analysis

Background

Acute organophosphorus (OP) poisoning is one of the major causes of mortality among patients presenting to emergency departments in developing countries. Although various predictors of mortality among OP poisoning patients have been identified, the role of repeated measurements of vital signs in determining the risk of mortality is not yet clear.

Therefore, the present study examined the relationship between trajectories of vital signs and mortality among OP poisoning patients using latent class growth analysis (LCGA).

Methods

This was a retrospective cohort study using data for 449 OP poisoning patients admitted to Civil-Hospital Karachi from Aug’10 to Sep’16. Demographic data and vital signs, including body temperature, blood pressure, heart rate, respiratory rate, and partial-oxygen pressure, were retrieved from medical records. The trajectories of vital signs were formed using LCGA, and these trajectories were applied as independent variables to determine the risk of mortality using Cox-proportional hazards models. P-values of < 0.05 were considered statistically significant.

Results

Data for 449 patients, with a mean age of 25.4 years (range 13–85 years), were included. Overall mortality was 13.4%(n = 60). In trajectory analysis, a low-declining systolic blood pressure, high-declining heart rate trajectory, high-remitting respiratory rate trajectory and normal-remitting partial-oxygen pressure trajectory resulted in the greatest mortality, i.e. 38.9,40.0,50.0, and 60.0%, respectively, compared with other trajectories of the same parameters. Based on multivariable analysis, patients with low-declining systolic blood pressure were three times [HR:3.0,95%CI:1.2–7.1] more likely to die compared with those who had a normal-stable systolic blood pressure. Moreover, patients with a high-declining heart rate were three times [HR:3.0,95%CI:1.5–6.2] more likely to die compared with those who had a high-stable heart rate. Patients with a high-remitting respiratory rate were six times [HR:5.7,95%CI:1.3–23.8] more likely to die than those with a high-stable respiratory rate. Patients with normal-remitting partial oxygen pressure were five times [HR:4.7,95%CI:1.4–15.1] more likely to die compared with those who had a normal-stable partial-oxygen pressure.

Conclusion

The trajectories of systolic blood pressure, heart rate, respiratory rate and partial-oxygen pressure were significantly associated with an increased risk of mortality among OP poisoning patients.

Correlation and prognostic significance of serum amylase, serum lipase, and plasma cholinesterase in acute organophosphorus poisoning

Background:

Organophosphorus (OP) are substances that are originally produced by the reaction of alcohols and phosphoric acid. These OP compounds are the main components of herbicides, pesticides, and insecticides. These are easily available in developing country like India; there is lack of awareness about these chemicals which results in high morbidity and mortality.

Aims and Objectives:

To estimate levels of amylase, lipase, plasma cholinesterase in acute OP poisoning. To assess severity of OP poisoning by using plasma cholinesterase levels and correlating it with other two markers. Predicting the severity of acute OP poisoning by using these biochemical markers.

Materials and Methods:

A hospital-based observational study was conducted on 100 subjects who were clinically diagnosed of acute OP poisoning. Subjects of either gender of all age-groups were included in the study. On admission, plasma cholinesterase, serum amylase, and serum lipase were measured. Based on plasma cholinesterase activity at the time of admission, subjects were divided into three groups. Group I-having 20-50% of plasma cholinesterase activity; Group II-10-20% of plasma cholinesterase activity; and Group III <10% of plasma cholinesterase activity.

Results:

Among 100 patients it was seen that serum amylase and serum lipase were negatively correlated with plasma cholinesterase levels and it was statistically significant. It was seen that serum amylase had the highest diagnostic accuracy for assessing severity of poisoning, 10 deaths were there in which 6 had <10% of plasma cholinesterase activity, 8 out of these 10 patients had elevated amylase level.

Conclusion:

OP poisoning is associated with elevated amylase level. Serum amylase, lipase can be used as an additional prognostic indicator along with plasma cholinesterase levels. Serum amylase could be considered as a better predictor of severity than lipase.