Fig 1.
Map of India with state of Tamil Nadu.
The union territory of Puducherry (town), showing location of the study site with highlighted adjacent districts of the state of Tamil Nadu from where patients were enrolled. (Map not to scale. Maps created using https://www.datawrapper.de/).
Table 1.
Clinical characteristics of Derivation and Validation cohorts.
Table 2.
Variables in the final multivariable regression model at step 5 of backward elimination with regression coefficients, adjusted odds ratio, p value, confidence intervals and points allotted in the score.
Table 3.
Calculation of VENOMS score.
Table 4.
Final model with regression equation, intercept, and regression coefficients.
Fig 2.
A: Mortality risk plotted against each point of the score for the derivation cohort (n = 248) showing a sigmoid curve with steep increase in mortality at score was greater than 6. B: Mortality prediction estimates for validation cohort (n = 140).
Fig 3.
A: Model discrimination in derivation cohort using a receiver operator characteristic curve (ROC) showing area Under Curve (AUC/c-index) of 0.948 (95% CI 0.920–0.976). A cut-off of 6 had a sensitivity of 90% and specificity of 83% for predicting mortality. B: Model performance in validation cohort using a ROC showing AUC/c-index of 0·90 (95% CI 0·85–0·97).
Table 5.
Accuracy of VENOMS score in predicting mortality in the validation cohort of patients with viper envenomation (n = 140).
Fig 4.
Predicted versus observed mortality risk in the validation cohort.
Calibration plots showing a slope of 0.7, intercept (CITL) of 0.4 and a c-index (AUC) of 0.92. E:O: ratio of expected to observed mortality. Graph created using pmcalplot in STATA, Stata/IC 16 for Windows.