Identifying the offending species in snakebite: a clinical score for use in community based epidemiological surveys

Abstract

INTRODUCTION: Hospital data on snakebite fails to account for victims who do not reach hospital or who seek alternative treatment. In community surveys, identification of biting species is very difficult because the offending snake may sometimes not even be seen, is infrequently captured or killed, and antigen detection cannot be used in this setting. OBJECTIVE: To develop and validate a clinical score to identify biting species. METHODS: We identified 10 features relating to bites of the 6 venomous snakes in Sri Lanka (3 relating to circumstances of a bite and 7 clinical features of envenoming). Based on these, we developed a score applying different weightings to individual features depending on the species of snake. We then constructed 60 artificiaLdata sets containing different clinical scenarios, typical and atypical, for bites of a particular species. They were used to make adjustments to the score and determine cut-off values for identification (total ≥60/100 + difference of ≥5/100 from second highest score). The score was prospectively validated on 134 cases of snakebite with definite species identification, six months after the incident. RESULTS: Data were available from 25 cobra bites, 32 hump-nosed viper bites, 27 krait bites, 30 Russel's viper bites and 20 saw-scaled viper bites. Sensitivity/specificity of the score were: cobra 76/99, kraits 85/99, RV 70/99, HNV 97/83, SSV 100/83. 114 (85%) cases were correctly identified. 17 (12.7%) were misidentified, and 3 (2.3%) could not be identified. Conclusions: This score identifies the offending snake species with reasonable accuracy, and may be suitable for use in epidemiologicai surveys.