Snakebites are an important public health and wilderness medical issue in Taiwan. Because of the warm and humid climate in Taiwan, there are more than 40 terrestrial snake species, of which 15 are venomous. Six of the venomous species are of high clinical importance, including Protobothrops mucrosquamatus (Taiwan Habu), Trimeresurus stejnegeri (Taiwan bamboo viper), Naja atra (Taiwan or Chinese cobra), Bungarus multicinctus (banded krait), Deinagkistrodon acutus (hundred pacer), and Daboia russelii siamensis (Russell’s viper).1 2

Naja atra belongs to the Elapidae family, and in addition to Taiwan, it inhabits southern China, Hong Kong, northern Laos, and northern Vietnam.3 Cobra venom contains a mixture of components, including cardiotoxin, cobrotoxin, haemotoxin, and phospholipase A2.4 Patients envenomed by a cobra experience varying degrees of neurotoxicity and cytotoxicity depending upon the proportions of the venom components. Due to evolution and geographical variations, different cobra species cause distinct clinical effects. For example, Naja philippinensis (northern Philippine cobra) causes a purely neurotoxic effect without local cytotoxicity.5 In contrast, N atra envenomation is associated with more cytotoxic effects.3 6 7 Although an equine-derived bivalent F(ab)2 antivenin has been produced by the Centers for Disease Control, ROC (Taiwan) to neutralise the venom of N atra, the surgical intervention rate remains high.1 8 The main objective of this study was to investigate the clinical presentations and predictors for surgery in patients with N atra envenomation. Due to high wound infection rates, the isolated bacteria from surgical wounds and the antimicrobial susceptibility were also analysed.

The Buddhist Tzu Chi General Hospital is the only tertiary care centre in eastern Taiwan. There are 1000 beds and the emergency department (ED) has more than 55 000 patient visits per year. This hospital is also the toxicant, drug information, and antidote control centre for eastern Taiwan. A retrospective study was conducted to analyse data from patients admitted to the ED with N atra envenomation between 1 January 2008 and 30 September 2014.

A medical assistant was responsible for collecting the medical records of patients admitted with snakebite during the study period by using the computerised chart system and International Classification of Diseases, 9th Revision, Clinical Modification codes 989.5, E905.0, E905.9, E906.2, and E906.5. Two physicians (the first and fifth authors) independently reviewed the charts and categorised these patients as having venomous or non-venomous snakebites based on the patient’s presentation with or without toxic effects. For venomous snakebites, classification of the snake species was based on the identification of the snake brought in by the patient or identification by the patient from a picture. All the included patients had a compatible presentation and consistent antivenin use as recorded in the patient chart. Patients who were initially recognised as having venomous snakebites but did not receive antivenin treatment were excluded from the study because of the high probability of a dry bite or misidentification of the snake species. Patients with a toxic presentation who could not identify the snake species or who received more than one type of antivenin were recorded as having an unknown poisonous snakebite.

Here we only report patients who were bitten by N atra. To identify the early clinical predictors of surgery, we categorised the patients into surgical and non-surgical groups. All surgical interventions were performed after surgical consultation in the ED or after admission when patients presented with progressive signs suggesting tissue necrosis, necrotising fasciitis, or suspected compartment syndrome. The final diagnoses of necrotising fasciitis and compartment syndrome were made according to surgical pathological findings and intracompartmental pressure measurement, respectively. The surgical procedures included debridement, fasciotomy, skin graft, and digit or limb amputation. The potential clinical predictors of surgery in N atra envenomation included the patient’s age, gender, season of snakebite, co-morbidities, details of envenomation, site of snakebite, initial vital signs on arriving at the ED, clinical presentation, laboratory data, treatment, timing of initial antivenin therapy, and total dose of antivenin.

For the laboratory analyses, the initial data obtained in the ED were collected, including haematology, biochemistry, and coagulation profiles. In regard to clinical presentation, the local signs and symptoms, local complications, and systemic manifestations and complications were classified. Local signs and symptoms included swelling, ecchymosis, necrosis, numbness, and bulla formation. Local complications included necrotising fasciitis and suspected compartment syndrome. Systemic manifestations and complications included neurological symptoms, including ptosis, blurred vision, drooling, and paralysis of facial, limb, or respiratory muscles; leukocytosis, defined as a white blood cell count of >11.0 x 10⁹ /L; thrombocytopenia, defined as a platelet count of <150 x 10³ /mm³;2 prothrombin time (PT) prolongation, defined as PT of >11.6 seconds; activated partial thromboplastin time (aPTT) prolongation, defined as aPTT of >34.9 seconds (prolonged PT and aPTT were defined according to our clinical laboratory reference range); fibrinogen consumption, defined as a fibrinogen level of <1.8 g/L; elevated D-dimer level, defined as a D-dimer level of >500 µg/L; acute renal impairment, defined as a creatinine level of >123.8 µmol/L9; and rhabdomyolysis, defined as a creatine kinase level of >1000 U/L.10 Two physicians reviewed the charts of the enrolled patients and rechecked the accuracy of the data collection. If the patient’s initial vital signs were not measured or laboratory tests were not performed in the ED, this was recorded as a missing value in the database. Any discrepancy regarding the collected data was resolved through discussion with the third physician on the research team. The study protocol was approved by the institutional review board of the Buddhist Tzu Chi General Hospital (IRB102-38). All patient records and information were anonymised and de-identified prior to analysis.

To identify significant early clinical presentation and laboratory data associated with surgery in patients with N atra envenomation, the Student’s t test or the Mann-Whitney U test for continuous variables and Chi squared test for categorical variables were used to perform univariate analysis. A P value of <0.05 was considered statistically significant, and all statistical tests were two-tailed. For multivariate analysis, the categorical variables with a P value of <0.05 in the initial univariate analysis were selected and entered into a logistic regression forward stepwise Wald test to calculate the odds ratios (ORs). The Statistical Package for the Social Sciences (Windows version 12.0; SPSS Inc, Chicago [IL], US) was used to perform the statistical analyses.

Between 1 January 2008 and 30 September 2014, a total of 245 patients with venomous snakebites were recorded. Among these, 64 (26.1%) patients had P mucrosquamatus envenomation, 56 (22.9%) had T stejnegeri envenomation, 28 (11.4%) had N atra envenomation, five (2.0%) had B multicinctus envenomation, six (2.4%) had D acutus envenomation, seven (2.9%) had D r siamensis envenomation, and 79 (32.2%) had unknown poisonous snake envenomation.

The snakebites associated with the highest surgical intervention rates were N atra (60.7%), followed by D acutus (33.3%), and P mucrosquamatus (12.5%).

Of the 28 patients with a N atra bite, 20 (71.4%) were male. The mean (± standard deviation) age of patients was 52.3 ± 3.2 years. Of the patients, 22 (78.6%) were bitten in the summer or fall; 17 (60.7%) were bitten on an upper limb; and 17 (60.7%) with N atra envenomation received surgical treatment. These patients had a significantly longer duration of hospitalisation than non-surgical patients (27.5 ± 10.2 days vs 2.7 ± 3.1 days; P<0.001). The main operative diagnosis was necrotising fasciitis (n=13, 76.5%) with confirmation by histopathology. The clinical characteristics of the 17 surgical patients are shown in Table 1. The mean duration from the time of initial presentation to the day of surgery was 5.5 ± 4.3 days. All 13 patients with necrotising fasciitis underwent emergency fasciotomy and debridement, and two required limb or digit amputation. The other four surgical patients without necrotising fasciitis only received local debridement with or without skin graft due to local tissue necrosis. Therefore, a smaller surgical wound and a shorter duration of hospitalisation were observed for these patients (Table 1). Nearly all surgical patients presented with local swelling and ecchymosis on the bite wound. Only one non-surgical patient presented with ecchymosis on a finger and was discharged from the ED 1 day later after four vials of antivenin were administered. The Figure shows the initial ecchymosis and necrosis of a N atra bite wound, the development of extensive tissue necrosis, and the postoperative wounds of a surgical patient (patient No. 9 in Table 1).

The demographic and clinical characteristics were compared between the surgical and non-surgical patients with N atra envenomation (Tables 2 and 3). Overall, the surgical patients received significantly higher doses of antivenin (9.2 ± 4.9 vials vs 3.8 ± 2.4 vials; P=0.002) and had significantly higher white blood cell counts (11.0 ± 3.7 x 10⁹ /L vs 8.2 ± 2.4 x 10⁹ /L; P=0.043). A higher respiratory rate was also evident in surgical patients (median [interquartile range]: 20 [20-21] vs 18 [16-18] breaths/min; P=0.015), but the incidence of missing data in both groups for this factor was high (Table 2). A significantly higher proportion of surgical patients received six or more vials of antivenin in total compared with non-surgical patients (82.4% vs 18.2%; P=0.001) [Table 3]. For local signs, symptoms and complications, a significantly higher proportion of surgical patients presented with local swelling (100% vs 72.7%; P=0.05), ecchymosis (82.4% vs 9.1%; P<0.001), necrosis (58.8% vs 0%, P=0.002), bulla formation (41.2% vs 0%; P=0.023), and necrotising fasciitis (76.5% vs 0%; P<0.001) [Table 3]. Age, season and site of snakebite, co-morbidity with diabetes, allergy to antivenin, and other systemic manifestations were not found to be significantly different between surgical and non-surgical patients. None of the patients with N atra envenomation presented with neurological symptoms. One patient with a small area of ecchymosis on the bite wound of his left hand did not receive surgical intervention, because the condition of the local wound improved and healed after administration of four vials of antivenin and intravenous antibiotics.

To determine clinical predictors of surgery, a multivariate logistic regression analysis was conducted for the significant variables derived from the univariate analysis. Necrotising fasciitis was not included in the multivariate analysis because it was a surgical finding and not an early sign that could be identified in the ED. The results showed that local ecchymosis (OR=34.36; 95% confidence interval [CI], 2.20-536.08; P=0.012) and a high total dose of antivenin (≥6 vials; OR=14.59; 95% CI, 1.10-192.72; P=0.042) were the most significant clinical predictors of surgery in patients with N atra envenomation.

To analyse the cause of necrotising fasciitis in surgical patients, the bacterial isolates identified from snakebite wounds were further analysed in surgical patients. The positive culture rate was 88.2% (n=15). More than one type of bacteria were isolated from the snakebite wound in 14 (82.4%) surgical patients. The isolated pathogens included aerobic Gram-positive and Gram-negative bacteria, as well as anaerobic bacteria. The most commonly identified pathogen was Morganella morganii (76.5%), followed by Enterococcus faecalis (58.8%) and Bacteroides fragilis (29.4%) [Table 4].

The susceptibility of the bacteria to common antibiotics was analysed (Table 5). All Gram-positive bacteria were susceptible to vancomycin and teicoplanin. All Gram-negative bacteria were susceptible to cefotaxime and amikacin. Cefmetazole, gentamicin, levofloxacin, and trimethoprim/sulfamethoxazole were also effective against the isolated Gram-negative bacteria. Nearly all anaerobic bacteria were susceptible to clindamycin and metronidazole (Table 5).

In our study, skin change of ecchymosis on the bite wound was a good clinical predictor of surgery for N atra envenomation. The majority of N atra venom is cytotoxic, not haemorrhagic. The cardiotoxin and phospholipase A2 in N atra venom are direct cytotoxic polypeptides and cause degradation of cell membranes. They induce cell death by activating calcium-dependent proteases, and inhibit mitochondrial respiration. Hyaluronidase in N atra venom destroys interstitial constituents and precipitates the spreading of venom.11 A histopathological study of N atra bite wounds demonstrated thrombotic and fibrinoid deposits in superficial and deep dermal vessels, and leukocytoclastic vasculitis.12 Hence, both the cytotoxic and ischaemic effects of N atra venom may lead to blood extravasation from the destroyed subcutaneous vessels or capillaries and result in the characteristic ecchymosis on the bite wound. This finding may be a potentially important clinical sign of irreversible subcutaneous tissue necrosis due to development of tissue ischaemia.3 If management at this stage is inadequate, tissue destruction may progress to involve the fascia rapidly and extensively with ultimate development of necrotising fasciitis.13 In our patients, extensive tissue destruction beyond the original bite site was evident once necrotising fasciitis developed. Further study is required to verify whether early surgical intervention can prevent the development of necrotising fasciitis, reduce the size of surgical wound, or shorten the length of hospital stay. Nonetheless, surgical assessment may be needed in patients with N atra bite who present with local ecchymosis on the bite wound.

Traditionally, immediate injection of antivenin to neutralise N atra venom was the only efficient management.14 A study using an enzyme-linked immunosorbent assay to detect the amount of N atra venom revealed that two to eight vials of antivenin are sufficient to eliminate systemic circulating venom if presentation is early.6 The efficacy of systemically administrated antivenin to diminish local tissue destruction is still controversial, however, and needs further study.3 In an animal study, the cytotoxic venom of N atra was shown to bind with high affinity to tissues leading to high levels of local tissue destruction.15 This finding may explain the difficulties associated with neutralisation of local venom toxicity, especially in cases of delayed presentation. Thus, the adequate dose of antivenin for preventing advanced tissue destruction remains unknown. In our study, nearly all patients presented within 1 hour following envenoming. Intravenous injection of antivenin was administered as soon as clinically possible following identification of cobra envenoming. Interestingly, the use of higher doses of antivenin in patients with N atra envenomation did not decrease surgical rates even in cases of early presentation. More than half of the patients underwent surgery and the majority were diagnosed with necrotising fasciitis. Surgical intervention appears to be crucial for the management of N atra envenomation. Hence, the identification of clinical predictors of surgical need and sufficient evidence to support surgeons’ decisions to carry out early surgical intervention are important issues in N atra management.

High bacterial isolation rates and the growth of mixed spectrums of bacteria from bite wounds indicate bacterial infection (which may be another cause of necrotising fasciitis in N atra envenomation), bacterial colonisation, or both. Morganella morganii and Enterococcus species were the most common pathogens cultured from N atra bite wounds in this study. This finding is consistent with the bacterial cultures taken from oral swabs of N atra in Hong Kong.16 Similar results were also described in a previous study in western Taiwan.17 Hence, the use of adequate antibiotics is important in N atra envenomation management. In accordance with the results of our tests of the antibiotic susceptibility of the isolated bacteria, treatment with glycopeptide antibiotics (vancomycin or teicoplanin) combined with a third-generation cephalosporin (cefotaxime) with or without anti-anaerobic antibiotics (clindamycin or metronidazole) is recommended.

There are several limitations in our study. First, this was a retrospective chart review comparative study. Non-uniform description of symptoms and signs documented by different providers may have influenced the validity of the statistics. Second, the small sample size may limit the statistical power in the multivariate analysis. Third, there are no definitive guidelines for the management of venomous snakebites in Taiwan, and various treatment strategies were employed; this may have influenced the final outcome. A large-scale prospective study is warranted to verify the risk factors we have identified to provide more accurate data for early risk stratification, treatment, and management of these patients.

Of the six common venomous snakes in eastern Taiwan, bites by N atra most frequently lead to surgical intervention. Severe tissue necrosis and necrotising fasciitis were the main findings during surgery. Patients who present with ecchymosis on the bite wound or who require higher doses of antivenin may have a higher probability of surgical intervention. In addition to early and adequate antivenin treatment, combined broad-spectrum antibiotics and surgical intervention may be needed in the management of N atra snakebites.

This work was supported by Buddhist Tzu Chi General Hospital Grants TCRD103-53 (to the first author).

All authors have disclosed no conflicts of interest.

1. Liau MY, Huang RJ. Toxoids and antivenoms of venomous snakes in Taiwan. Toxin Rev 1997;16:163-75.

2. Hung DZ. Taiwan’s venomous snakebite: epidemiological, evolution and geographic differences. Trans R Soc Trop Med Hyg 2004;98:96-101. Crossref

3. Wong OF, Lam TS, Fung HT, Choy CH. Five-year experience with Chinese cobra (naja atra)–related injuries in two acute hospitals in Hong Kong. Hong Kong Med J 2010;16:36-43.

4. Li S, Wang J, Zhang X, et al. Proteomic characterization of two snake venoms: Naja naja atra and Agkistrodon halys. Biochem J 2004;384:119-27. Crossref

5. Watt G, Padre L, Tuazon L, Theakston RD, Laughlin L. Bites by the Philippine cobra (Naja naja philippinensis): prominent neurotoxicity with minimal local signs. Am J Trop Med Hyg 1988;39:306-11.

6. Hung DZ, Liau MY, Lin-Shiau SY. The clinical significance of venom detection in patients of cobra snakebite. Toxicon 2003;41:409-15. Crossref

7. Wang W, Chen QF, Yin RX, et al. Clinical features and treatment experience: A review of 292 Chinese cobra snakebites. Environ Toxicol Pharmacol 2014;37:648-55. Crossref

8. Huang LW, Wang JD, Huang JA, Hu SY, Wang LM, Tsan YT. Wound infections secondary to snakebite in central Taiwan. J Venom Anim Toxins Incl Trop Dis 2012;18:272-6. Crossref

9. Hung DZ, Wu ML, Deng JF, Lin-Shiau SY. Russell’s viper snakebite in Taiwan: differences from other Asian countries. Toxicon 2002;40:1291-8. Crossref

10. Chen YW, Chen MH, Chen YC, et al. Differences in clinical profiles of patients with Protobothrops mucrosquamatus and Viridovipera stejnegeri envenoming in Taiwan. Am J Trop Med Hyg 2009;80:28-32.

11. Harris JB. Myotoxic phospholipases A2 and the regeneration of skeletal muscles. Toxicon 2003;42:933-45. Crossref

12. Pongprasit P, Mitrakul C, Noppakun N. Histopathology and microbiological study of cobra bite wounds. J Med Assoc Thai 1988;71:475-80.

13. Gozal D, Ziser A, Shupak A, Ariel A, Melamed Y. Necrotizing fasciitis. Arch Surg 1986;121:233-5. Crossref

14. Russell FE. Snake venom immunology: historical and practical considerations. Toxin Rev 1988;7:1-82. Crossref

15. Guo MP, Wang QC, Liu GF. Pharmacokinetics of cytotoxin from Chinese cobra (Naja naja atra) venom. Toxicon 1993;31:339-43. Crossref

16. Lam KK, Crow P, Ng KH, et al. A cross-sectional survey of snake oral bacterial flora from Hong Kong, SAR, China. Emerg Med J 2011;28:107-14. Crossref

17. Chen CM, Wu KG, Chen CJ, Wang CM. Bacterial infection in association with snakebite: a 10-year experience in a northern Taiwan medical center. J Microbiol Immunol Infect 2011;44:456-60. Crossref