Snakebite in KwaZulu-Natal: the burden of disease and prediction of risk of adverse outcomes.

Abstract

BACKGROUND The total number of snakebites per annum in KwaZulu-Natal (KZN) is unknown. Yet it is believed that the burden that snakebites place on hospitals in areas with a high incidence of snakebite is significant. There are no official snakebite guidelines in South Africa or KwaZulu-Natal. The result is non-uniform management practices and in many cases inappropriate prescribing of antivenom, which may potentially be harmful given a high rate of allergic reactions to antivenom. In order to standardise practice along evidence-based lines, it is important to identify factors predictive of a poor outcome so that treatment can be appropriately targeted at those individuals. AIMS 1. To determine a figure for the annual incidence of snakebite, identify regional variations in incidence and estimate the burden of snakebite on public hospitals in the north-eastern province of KwaZulu-Natal, an area in which snakebite is endemic. 2. To report a five-year prospective experience with snakebite in a highly endemic area of South Africa and to identify factors predictive of severity. 3. To develop and validate a severity scoring system to facilitate the management of snakebite in South Africa by allowing early identification of patients at increased risk of a severe course, and thereby develop an improved algorithm for the management of snakebite. 4. To determine the site of expansion accounting for the swelling in patients bitten by a cytotoxic snake species, in particular to distinguish between muscle compartment swelling and superficial swelling, and to determine the clinical utility of bedside ultrasonic examination as a potential tool for identifying patients with possible compartment syndrome. METHODS The work is reported as four sub-studies, using a selection of methodologies appropriate to each, which are fully described in Chapters 2-7. In order to determine incidence of snakebite, we applied a novel method whereby incidence was extrapolated from antivenom supply data provided by the central provincial pharmacy depot, with the appropriate conversion factor being determined from a stratified a sample of 6 hospitals. We analysed prospectively captured data on all patients admitted to Ngwelezane Hospi Emergency Department from September 2008 to December 2013 with a diagnosis of snakebite. Using the need for an active treatment intervention (ATI), which we defined as antivenom administration or surgical intervention as a proxy for severity, we analysed our data for factors present on admission which were predictive of severity. In a subsequent study, we developed a severity score on a cohort of patients, which was then validated in a separate and subsequent cohort of patients. We developed a methodology for the assessment of depth of bite in patients bitten on a limb whereby the ratio of the thickness of the deep muscle compartment and the subcutaneous compartment of the bitten limb, measured by bedside ultrasound, were expressed proportionally, and compared with the ratio on the unaffected limb. This information was then used to identify the major site of swelling. RESULTS Incidence We estimated that 11% (95% CI: 8-14%) of snakebite presentations to hospital resulted in the administration of antivenom. By extrapolation, the overall incidence for KZN was 16/100 000. There was wide geographic variation, with the highest incidence, at 82/100 000 in the subtropical north east of the province. The estimated annual cost of snakebite in KZN was between USD 1 135 782 and USD 2 877 314. Analysis of a case series and prediction of severity 879 cases were analysed. Envenomation was identified in over two thirds of admissions. Cytotoxic snakebites accounted for 98% of envenomations. Only 4 cases of haemotoxic bleeding and 5 cases of neurotoxicity were admitted. Although we demonstrated a significant correlation between severity and prolonged INR, reduced platelet count, haemoglobin, reduced or elevated leucocyte count and elevated serum urea. However, their use as predictors of severity was limited by poor sensitivity and specificity. Clinical factors correlating with severity were the paediatric age group and a delayed presentation to hospital. In the prospective study, 146 of 879 snakebite admissions in the development cohort and 40 of 100 in the validation cohort reached the primary end point of an ATI. Six predictors of risk for ATI were identified from the development cohort: age <14 years, delay to admission >7 hours, white cell count > 10x109 cells/l, platelet count<92 x109/l, haemoglobin <7.1 g/dl, INR >1.2. Each risk predictor was assigned a score of 1; ROC curve analysis returned a value of more than 4 out of 6 as the optimal cut-off for prediction of an ATI (AUC 0.804; 95% CI 0.758-0.84). Testing of the score on the validation cohort produced a sensitivity of 22.5% and a specificity of 96.6%. The PPV and NPV were 81.8% and 65.2% respectively. Ultrasonic determination of the site of swelling in cytotoxic envenomation The majority of bites were in the upper limb (27/42). Tissue expansion was noted in both the sub-cutaneous and muscle compartments of the envenomed limbs. The site of swelling was predominantly in the subcutaneous tissues, while swelling in muscle compartment was limited (the mean expansion coefficient for subcutaneous tissues was 2.0 (CI: 1.7-2.3) versus 1.06 (CI: 1.0-1.1) respectively). The difference between the groups was significant (P<001). One case, confirmed as compartment syndrome, showed marked swelling in the muscle group and stood out as a clear outlier in terms of the expansion coefficient. CONCLUSIONS The burden of snakebite is substantial, and is felt unequally across the province. Furthermore, we propose that our method may be used to estimate the incidence of other diseases treated with a standard regimen and for which incidence figures are otherwise unknown. Two-thirds of patients who present to hospital with snakebite in north-eastern South Africa will have symptoms of envenomation, with the overwhelming majority manifesting cytotoxicity. Bites by neurotoxic and haemotoxic species are rare. We have identified a number of factors which may potentially be of value in predicting severity, but which are on their own of insufficient accuracy to be reliable. Basic ultrasound techniques may be used to identify the site and degree of tissue swelling from cytotoxic envenomation. It is a non-invasive, painless procedure that can assist the clinician to assess the injured limb and may also be of benefit to monitor the progression of swelling. Our scoring system, which we propose to name the Zululand Snakebite Severity Score (ZSSS), is a useful adjunct to clinical assessment in managing snakebite. A patient with a positive result has an 80% probability of progressing to the point where an ATI is indicated. Its value is greatest in those patients who fall in the mild to moderate clinical category. This score now requires validation on a wider scale across South Africa, to determine its accuracy in areas other than those in which it was tested.