Identifying the snake: First scoping review on practices of communities and healthcare providers confronted with snakebite across the world

Background Snakebite envenoming is a major global health problem that kills or disables half a million people in the world’s poorest countries. Biting snake identification is key to understanding snakebite eco-epidemiology and optimizing its clinical management. The role of snakebite victims and healthcare providers in biting snake identification has not been studied globally. Objective This scoping review aims to identify and characterize the practices in biting snake identification across the globe. Methods Epidemiological studies of snakebite in humans that provide information on biting snake identification were systematically searched in Web of Science and Pubmed from inception to 2nd February 2019. This search was further extended by snowball search, hand searching literature reviews, and using Google Scholar. Two independent reviewers screened publications and charted the data. Results We analysed 150 publications reporting 33,827 snakebite cases across 35 countries. On average 70% of victims/bystanders spotted the snake responsible for the bite and 38% captured/killed it and brought it to the healthcare facility. This practice occurred in 30 countries with both fast-moving, active-foraging as well as more secretive snake species. Methods for identifying biting snakes included snake body examination, victim/bystander biting snake description, interpretation of clinical features, and laboratory tests. In nine publications, a picture of the biting snake was taken and examined by snake experts. Snakes were identified at the species/genus level in only 18,065/33,827 (53%) snakebite cases. 106 misidentifications led to inadequate victim management. The 8,885 biting snakes captured and identified were from 149 species including 71 (48%) non-venomous species. Conclusion Snakebite victims and healthcare providers can play a central role in biting snake identification and novel approaches (e.g. photographing the snake, crowdsourcing) could help increase biting snake taxonomy collection to better understand snake ecology and snakebite epidemiology and ultimately improve snakebite management.


Introduction
An estimated 5.4 million snake bites occur globally every year. About half of these cause snakebite envenoming (SBE), killing 81,000-138,000 people and disabling 400,000 more in the poorest regions [1,2]. In May 2019, the World Health Organization (WHO) launched a road map to halve these deaths and disabilities by 2030, particularly focusing on the development of antivenoms and on their adequate distribution in the most affected countries [3,4]. For this, understanding what type of snakes and associated snake bites occur where is crucial, yet this depends first on the taxonomic identification (identification hereafter; i.e. family, genus and species) of these snakes [5][6][7]. At the clinical level, the identity of the biting snake (BSN) can help healthcare providers anticipate victims' syndromes and support decision making when treating the patient (i.e. whether or not to administer antivenom or which type of antivenom) [8,9]. This decision is important because not only are antivenoms effective against a limited number of venomous snakes but they have potentially lethal side effects such as fatal allergic reactions and should not be used to treat bites from non-venomous snakes. This is especially important in view of the scarcity and high costs of antivenom vials in many countries.
However, identifying the BSN is challenging due to the high diversity of snake species [10] in snakebite endemic countries (e.g. 310 snake species in India), and the limited herpetological knowledge of communities and healthcare providers confronted with snakebite. In rural areas, traditional healers are often the first to be consulted for snakebite at the community level [11] and they could play a role in BSN identification, yet this remains to be assessed. Different behaviours and practices exist across the world depending on the development of health systems, local culture and perception of snakes and snakebite, and snake diversity. There is no standardized protocol for identifying BSNs and recommended practices are often specific to certain regions. For instance, the WHO's Regional Office for South-East Asia (2016) [12] recommends experts identify the BSN based on a photo (i.e. taken with a mobile phone) or the animal's body, when killed by the victim or bystanders and brought to the health facility.
To our knowledge, no review has systematically explored the literature to identify and synthesize the nature and extent of information available on the practices and challenges in BSN identification across the world. The objective of this scoping review is to provide a global and comprehensive description of the diversity of behaviours and practices of communities and healthcare providers, and the implications of these, when confronted with snakebite and the need to identify BSNs. This review also assesses the capabilities of these communities and healthcare providers to identify the BSN, and the frequency and consequences of misidentifications. Finally, it addresses aspects related to the BSNs, including their diversity and behaviour, and how this may influence their identification.

Eligibility criteria
Original epidemiological studies and clinical case report/series reporting snakebite cases in humans caused by wild snakes and including information on their taxonomic identification were considered. Studies from all geographical areas and publication dates were eligible.

Search strategy
Web of Science and Pubmed were searched from inception to February 2, 2019 using the following key words: snakebite, snake bite, snake envenoming, snake envenomation, case, victim, event, patient, biting snake, culprit, offending snake, species, identif � , misidentif � , unidentif � , identity, and mistaking (see S1 Table for full search strategy). In addition to the database searches, we used extensive secondary search techniques, such as snowball search, hand searching literature reviews, and performing key word searches in Google Scholar. This secondary search was conducted in English, French, and Spanish by three of the authors (IB, SBM, AMD). With PubMed, we accessed primarily publications in the field of medicine and life sciences, while with Web of Science and Google Scholar we covered most scientific fields.

Publication selection
Searched publications were merged using citation software EndNote X7 and duplicates were removed. IB, an expert in the domain of snakebite, screened all the titles and abstracts and excluded those publications that did not match eligibility criteria. Eligible publications were read by IB and SBM, and their relevance was further assessed, particularly focusing on information pertaining to the BSN (i.e. number of snakes identified and methods used). A final set of publications was produced by discussion and consensus between IB and SBM. No quality assessment of the publications selected was made.
Charting the data IB and SBM charted the data from selected publications independently. Their results were compared and discrepancies resolved through discussion and consensus. The variables extracted were: publication identifiers (authors, journal, year of publication, language); study characteristics (study design, country, setting, sample size); number of BSNs identified; number of victims that saw the BSN; number of BSNs captured or killed; taxonomic granularity of BSN identification (family, genus, or species); the way the BSNs were identified. We cross-referenced all common and scientific names of snakes reported in the publications with the Reptile Database [10] and kept track of changes in taxonomy where relevant. One of us (AMD) searched the literature for information on the activity time (diurnal/nocturnal/both), foraging strategy (active/passive/both), and general habitat (terrestrial, aquatic, arboreal, fossorial) of each species. Some species lacked species-specific information and their behaviour was extrapolated from those of congeners. We also gathered data on the number of snake species per country from the Reptile Database [10] and coded these as medically-important venomous snakes (MIVS) or not following the WHO. We used the pcor.test function in package ppcor (v. 1.1) in R (v. 3.5.1) to estimate partial correlation coefficients among the number of BSN species per country, the total number of snake species in that country, the total number of snake bites reported in publications from that country, and the MIVS status of the snake species.

Selection and characteristics of publications
A total of 467 unique publications resulted from the initial search, and 150 of these were included in the review (Fig 1, S2 Table). These publications covered 35 countries on all continents (Table 1) and were mostly retrospective or prospective hospital-based epidemiological studies (76%), conducted in Asia (50%), published after 2000 (77%) in English (94%). In total, 33,827 snakebite cases were reported in these studies and the median number of cases per study was 90-91 (range 1-3,411).

Spotting the BSN
Out of the total, twenty-three publications (15%) across 13 countries mentioned the number of victims that saw the BSN. In total, 2,723 victims out of 3,865 (70.5%) have seen the BSN (range 51.9% to 93.2% across studies) ( Table 2). Factors associated with the circumstances/ context of the bite were often described for those cases where the snake was not spotted. This includes, for example, bites that occurred at night, assumed to be caused by nocturnal snakes (e.g. Trimeresurus sp., Bungarus sp.) (e.g. Gabon, Hong Kong, Nepal, Saudi Arabia, Sri Lanka) [16][17][18][19][20]

BSN captured/killed and brought to health facilities
A total of 114 (76%) publications in 30 countries described snakebite cases where victims or bystanders captured the BSN and brought it to the health facility (S2 Table). In 78 (68%) of these 114 publications and 24 countries, the BSN was killed. Globally, the BSN was captured/ killed in 9,671/25,188 (38%) snakebite cases, but this practice varied among countries (Table 3). Cultural perceptions affected this behaviour. In Nepal, an Indian cobra (Naja naja) was not killed by the victim "in fear of revenge by its partner" [20]. In Sri Lanka "there may be a reluctance to capture or kill the animal because of fear or superstition" [44], and a snakebite victim "refrained from catching the snake due to religious ethics" [27]. In South Africa (northern Natal), killed BSNs were often not brought to hospitals because "most snakes are incinerated immediately after having been killed for purposes of protection from the bones of the snake, which are believed to be dangerous even after death" [45].
In four publications, victims were bitten while attempting to kill a snake. This was the case for 53% and 100% of patients bitten on the fingers in two Australian studies [29, 30] and caused the death of an 80-year-old traditional healer in Cameroon [53] and a worker in Hong Kong [17].

BSN identification methods
From the methods section of selected publications, we extracted information on the way the BSNs were identified when victims reached a healthcare facility (S2 Table). One or a combination of the following methods were used: i) examining the captured BSN brought to the healthcare facility (110 publications). The person who identified the BSN was not always reported. In 43 publications, and particularly in Brazil (n = 10), Australia (n = 7), Sri Lanka (n = 7), and India (n = 6), the identification was made by a snake expert. This occurred particularly for case report/case series to confirm unusual snakebite events [85,87,94] and for randomized controlled trials of antivenoms or epidemiological studies focusing on a particular snake species. The latter involves patients bitten by a specific snake species and therefore giving precise taxonomic attribution to the biting snake is key.
ii) verbal description of the BSN by victims/bystanders (52 publications). They described the colour (e.g. brown or black) or size (e.g. large) of the BSN. In some studies, photographs or preserved specimens in curated collections of local material were shown to victims or bystanders to assist in the identification. Several publications mentioned that a majority of patients were unable to recognize snake species based on photographs (e.g. [35,45,66,110] 76,111]. In Brazil, health care providers identified the BSN at the genus level (Bothrops/Crotalus/Lachesis/Micrurus) based on patients' signs and symptoms [102,112,113]. In these studies, most patients were effectively treated, but two deaths occurred possibly caused by the use of non-specific antivenom according to the authors [113]. The practice of bringing the BSN to the health facility is uncommon in northeastern Brazil compared to other Brazilian regions [104,113]. Two studies focusing on krait bites also took into account bite circumstances (i.e. bite at night while victims were asleep). iv) laboratory methods (23 publications). Immunoassays (EIA, ELISA) that detect venom antigens of some snake species in victims' blood were used in Australia, Bangladesh, Brazil, Ecuador, Myanmar, Sri Lanka, and Thailand mainly to retrospectively identify BSNs and for research purposes. A pilot study explored the use of molecular tools to identify snakes in Nepal [56]. v) examining a picture of the BSN (9 publications). This practice was reported in the US (n = 3), Morocco (n = 2), Malaysia (n = 1), Australia (n = 1), Colombia (n = 1) and Laos (n = 1). The picture of the BSN was taken by the victim/bystanders, some with a mobile phone, or by the medical staff when the dead snake was brought to the hospital. The picture was then sent to a herpetologist (Laos, USA) or a Poison Control Center (Colombia, Malaysia, Morocco, USA). BSNs were identified at the species or genus level in 18,065 out of 33,827 snakebite cases (53%) (S2 Table).

Capability of victims and healthcare providers to identify snakes
Victims/bystanders. In nine studies, victims claimed they could identify the BSN and reported its common or local name (Table 4). In a Bangladeshi hospital, most snakebite victims were too distressed to describe the BSN and often misidentified it even when it was brought to the health facility [55]. Victims may not know the name of the snake [24] or the diversity of vernacular names given to a single snake species and/or taxonomic synonymies generate confusion for doctors (e.g. in the Brazilian Amazon [103,112,114]). In South Africa the word 'mamba' is often used by the local population to mean 'snake' of any species [50]. A survey of 150 inhabitants in southern Nepal showed that respondents were generally unable to identify different snake species [20]. In many parts of the world, authors considered snake identification by victims too unreliable to assist in routine snakebite management or in clinical snakebite research [24, 28, 66,110].
Healthcare providers. The identity and credentials of the person doing the identification of BSN brought to health facility was often not reported. This limits our assessment of healthcare provider capability in identifying snakes. In three studies in India, Sri Lanka, and Thailand, experts systematically re-examined dead snakes initially identified by hospital personnel and misidentifications were reported in respectively 17/44, 51/860, and 27/1631 cases [44,117,118]. In the US, pictures of the BSN were sent to a Poison Control Center and identification by a snake expert was compared to that of healthcare providers. Healthcare accuracy for copperhead (Agkistrodon contortrix) and cottonmouth (A. piscivorus) identifications was respectively 68% and 74% [119]. Many authors highlighted the lack of training of healthcare providers in identifying biting species including in Brazil, India, Nepal, Singapore, South Africa, and Thailand [34,45,56,97,113,117,118,120,121].

BSN misidentifications
The BSNs were misidentified in 106 snakebite cases in Australia (n = 3), Brazil (n = 10), Hong Kong (n = 1), India (n = 6), Malaysia (n = 1), Sri Lanka (n = 54), and Thailand (n = 31). The consequences for the victims are detailed in Table 5 according to three scenarios. Most frequently, a venomous species was confused with another venomous species or a non or mildly venomous (opisthoglyphous) species was misidentified as a venomous snake.

Diversity of captured/killed BSN
A total of 8,885 BSNs from 149 species in 12 families were captured and identified. Of these, 6,750 BSNs were identified to species, 2,082 to genus, and 53 as "non-venomous". In total, 7,628 individual snakes of 69 species were MIVS, 1,205 individuals of 71 species were non-venomous, and 52 individuals of 9 species were potentially dangerous but understudied species (species that lack specific data on clinical symptoms or venom toxicity but which are closely related to dangerous snakes and possess fangs and venom glands (see [126]) (  Table), as well as one amphisbaenid lizard (Amphisbaena mertensii) [101] and one serpentine amphibian (caecilian [118]) (not included in snake totals above). Of these 149 snake species, 32 were diurnal, 76 were nocturnal, and 41 were or could be active throughout the day, depending on the season. A total of 84 species were active foragers, 45 were ambush predators, 14 used both strategies, and 6 were unknown. General habitat use of 26 species was arboreal, 7 fossorial, 22 aquatic or semi-aquatic, and 94 were primarily terrestrial, with 4 of these partially aquatic, 5 partially arboreal, and 7 partially fossorial (S4 Table). A substantial number of snakes-1,859 (21%) individuals of 22 species-were reported in the literature using only the common name. Of those reported using the scientific name, 20 species (92 individuals) had been moved to another genus since the time of publication, 10 species (241 individuals) had experienced other taxonomic changes (splits or lumps) (e.g. Echis ocellatus in Cameroon has been renamed E. romani [127]), and the names of 3 species (6 individuals) contained minor misspellings as reported (S3 Table).

Discussion
Based on a final selection of 150 publications from across the world, this first scoping review on BSN identification practices of communities and healthcare providers confronted with snakebite shows that: (I) BSN identification is important for snakebite epidemiology and clinical management yet there is a diversity of practices depending on cultural, ecological and healthcare contexts and there are no official standards (II) the majority of victims see the BSN but they are unreliable in their identification (III) the practice of capturing or killing the BSN occurs in all continents, especially in Asian countries with a diversity of snakes, and (IV) healthcare providers struggle to identify a BSN presented in the health facility and misidentifications occur.
Snakebite victims/bystanders in many snakebite endemic countries are aware of the importance of BSN identification and can play an important role. We found that on average 70% of victims/bystanders spotted the BSN and 38% managed to capture or kill it and brought it to the health facility for identification. Although this is a dangerous practice (i.e. due to the risk of secondary bites [17,29,30,53,[128][129][130][131][132][133]), it occurs worldwide but particularly in Asia (e.g. Table 6. Diversity and abundance of BSNs captured/killed and identified.
BSN identification could also complement syndromic approach to snakebite [36, 44, 76], particularly in those cases where symptoms caused by venoms of different snake species overlap (e.g. Russell's viper causes paralytic signs suggesting elapid neurotoxicity in Sri Lanka) [12].
Snake species diversity and the fact that bites from some "non-venomous" snakes can cause signs of envenoming (e.g. Clelia plumbea and Philodryas olfersii in Brazil) [64,94,124,126] complicate snakebite clinical management and have led to snake misidentifications. Snake identification can be challenging for communities and healthcare providers (i.e. diversity of snakes, mimicry). In two villages in rural Tanzania, most of the respondents to a survey could not precisely differentiate between venomous and non-venomous snakes [149]. In a cross-sectional survey of 119 healthcare providers in Laos, 86 participants (72.3%) had inadequate knowledge of snake identification [150]. Although healthcare providers could be trained to recognize locally prevalent snake species (e.g. at the Damak Snakebite Treatment Center in Nepal [151]), reinforcing the collaboration between communities at risk, healthcare providers and snake experts could be a more effective approach to increase the number of BSNs correctly identified.
With this review, we retrospectively collated all the species names of the 8,885 BSN that were captured and identified and we have built a first extensive global list of snakes having bitten humans. This list includes 69 MIVS and 71 non-venomous snake species, as well as nine that are potentially dangerous but understudied species [126]. MIVS species are already listed by the WHO (2010) [152]. This review extends this list to non-venomous snakes, although there is significant bias in which non-venomous snakes are brought to health facilities and the number of non-venomous BSN species was strongly correlated with the number of snake bites, suggesting that many more species of non-venomous BSNs exist and a limit has not been reached within epidemiological data. Nevertheless, this confirms that many snake bites globally are caused by non-venomous snakes, although victims may not always realize this immediately or at all.

Limitations of this study
Many publications had to be excluded from the review because the way the BSN identification was done was unclear or not reported even though the BSN genus or species was mentioned. The list of BSN species we built is limited to the publications we retrieved and included in the review. It could have been extended, had more publications met the eligibility criteria. We recommend that future epidemiological studies on snakebite clearly describe the method(s) used to identify the BSN, including the credentials of the person who did the identification and their confidence in the identification. We used key words related to 'identification' to specifically gather snakebite publications describing snake identification, although we may have missed relevant publications that do not mention these keywords or that are published in other languages (e.g. Russian, Chinese). Information on snake identification was reported in an inconsistent and fragmented manner. We managed this problem by involving two authors in data extraction and comparing data collected until a consensus was reached. Some publications, particularly prospective studies, applied specific inclusion criteria (e.g. a specific snake species, dead snake brought to hospital). These were excluded from analyses where they were sources of bias (e.g. calculation of proportion of captured snakes). The selected publications are mainly hospital-based studies with very few community-based studies. We missed the behaviour of snakebite victims who did not seek treatment at hospitals because of asymptomatic bite or use of traditional healers. Snake taxonomy is constantly changing and an average of 30 new species per year have been described since the year 2000. Although we were always able to definitively decide which species/genera were meant, rare or newly-described species may be missed by all identification methodologies, and taxonomic instability further complicates an already-challenging situation [153,154]. Finally, we cannot account for situations where snake misidentification was never discovered and incorrect names have been published, which seems likely in a subset of cases.

Conclusion
This global scoping review showed that BSN identification in snakebite endemic countries includes a diversity of methods and practices: capturing/killing the BSN and examination of its body, description of the BSN by victim/bystanders, interpretation of clinical features, laboratory tests, and photographing the BSN. The capacity of snakebite victims, bystanders and healthcare providers to spot and identify the BSN is context-specific and depends on circumstances of the bite, the local snake diversity, and their own knowledge of local snakes. BSN misidentifications occur and lead to inappropriate management of the victims. The influence of cultural perceptions about snakes and role of traditional healers in snake identification are largely unexplored in the literature and urges for further research. Victims/bystanders managed to capture a diversity of BSNs, including fast-moving nervous snakes. This is dangerous and not recommended, and photographing the snake could be an alternative option [12]. We provided the first evidence-based list of venomous and non-venomous snake species involved in bites to humans. This list could be further extended by implementing snake identification as part of the clinical practice. Such a systematic collection of the taxonomy of BSNs at the global level is of considerable interest to better understand snake ecology and snakebite epidemiology and ultimately improve SBE management.
Supporting information S1