Toxicon 59 (2012) 86–99

Contents lists available at SciVerse ScienceDirect

Toxicon journal homepage: www.elsevier.com/locate/toxicon

Review

Epidemiology of snakebites in Europe: A systematic review of the literature Jean-Philippe Chippaux a, b, * a b

Institut de Recherche pour le Développement, UMR 216, Mother and Child Facing Tropical Diseases, 08 BP 841 Cotonou, Bénin Université Paris Descartes, Sorbonne Paris Cité, Faculté de Pharmacie, France

a r t i c l e i n f o

a b s t r a c t

Article history: Received 16 May 2011 Received in revised form 7 October 2011 Accepted 11 October 2011 Available online 25 October 2011

Snakebites are rare medical emergency cases in Europe but may sometimes be severe and lead to complications. A better knowledge of snakebite epidemiology may help health authorities to better understand therapeutic requirements, especially concerning antivenoms, and thus improve treatment of snakebite. An extensive literature search for studies and articles published between 1970 and 2010 was performed. Both indexed and nonindexed articles were examined, the analysis of which took into account the heterogeneity between the studies and weighted the studies according to size of the study population covered. Most of the articles involved hospitalized patients who represented more than 90% of snakebites. Incidence, mortality and population at risk were estimated after stratification into three regions (northern, central and southern Europe) based both on viper species distribution and climatic characteristics. There was no significant variation in incidence from the north to the south of Europe. In the whole of Europe, including European Russia and Turkey, the annual number of snakebite cases was estimated at 7992 [CI 95% ¼ 6860–9178] bites, out of which approximately 15% were considered severe (grade 3). These bites usually occurred between May and September, with a more dispersed distribution in southern Europe. The average number of deaths per annum was 4 [0.7–7.7]. Children and male victims are more affected, contrary to what one would expect given their respective proportion in the entire population. Both upper and lower limb bites were recorded at an equal frequency while the bites in other parts of the body were very rare. Immunotherapy was prescribed in one out of three snakebites in Europe, with a very high geographical variability, in spite of excellent tolerance, at least considering highly-purified immunoglobulin fragments. Snakebites are uncommon in Europe but can cause life-threatening envenomation. Fragments of highly-purified immunoglobulins are now very well tolerated and dramatically reduce both severity and mortality of snakebites when used in treatment. Ó 2011 Elsevier Ltd. All rights reserved.

Keywords: Snakebite Envenomings Epidemiology Antivenom Europe

1. Introduction Snakebites are relatively rare medical emergency cases in Europe, but their evolution may sometimes present haematological, cardio-vascular and surgical complications. After most of Vipera bites, local swelling may be extensive and involve the trunk. Regarding Vipera berus, a species * Corresponding author: Institut de Recherche pour le Développement, UMR 216, Mother and Child Facing Tropical Diseases, 08 BP 841 Cotonou, Bénin, France. E-mail address: [email protected]. 0041-0101/$ – see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.toxicon.2011.10.008

widespread in Northern and Eastern Europe, it seems that the most common systemic symptoms are gastrointestinal upset with recurrent vomiting, circulatory instability, hypotension and haematological disturbances. Phospholipase A2 neurotoxins have been isolated from the venom of Vipera ammodytes (Luksic et al., 2006) and from venom of some populations of Vipera aspis (Ferquel et al., 2007) or V. berus (Ramazanova et al., 2008; Westerström et al., 2010). Few reversible neuromuscular paralysis have been reported from the literature mainly after V. ammodytes or V. aspis bites. On the contrary, bites by sub-species of Vipera ursinii seem to

J.-P. Chippaux / Toxicon 59 (2012) 86–99

induce less severe envenomation, usually restricted to local symptoms (Orsini et al., 1998; Krecsák et al., 2011). However this species is rare and lives in limited areas involving mainly herpetologists and requiring local symptomatic treatment. If we exclude Malpolon monspessulanus, an opistoglyphous Colubridae, which does not seem to be responsible for a significant number of bites (Pozio, 1988; de Haro, 2003; Pommier and de Haro, 2007) and which can be found throughout the Mediterranean region, the remaining poisonous European snake species belong to the sub-family of Viperinae. In spite of the remarkable resemblance between the different species belonging to this sub-family, there are significant differences in venom composition, even between individuals of each species. These differences explain the differential severity of symptoms and antivenom responses observed in patients and different regions (Guiavarch et al., 2011). The treatment of snakebites is still problematic, despite the availability of effective and well tolerated antivenom. In the past, the use of antivenom was often recommended but later strongly discouraged because of concerns over anaphylactic shock and horse protein allergies (Chippaux and Goyffon, 1998). Since the commercialisation of highly-purified immunoglobulin fragments, immunotherapy has been demonstrated to significantly reduce the risk of snakebite-related complications and duration of hospitalisation (Stahel et al., 1985; Harry et al., 1999; Persson, 2001; Karlson-Stiber et al., 2006; Chippaux, 2010). This study has two objectives: first, to present the incidence and main characteristics of accidental snakebites in Europe, particularly their severity, and secondly, to report on the current antivenom use modalities. Illegitimate snakebites, occurring when a snake, be it local or exotic, is voluntarily handled, are not taken into account in this study. The risk of illegitimate snakebite is increasing in the whole of Europe and requires special care and treatment (Chippaux, 1982; Schaper et al., 2009). 2. Material and methods A systematic review of articles related to snakebite in Europe was carried out. Searches were performed in bibliographical databases, especially MedlineÒ and InistÒCNRS, using the following key words: Europ* AND [snake* OR viper* OR adder] AND [envenom* OR antiven*]. Articles from 1980 through December 8th, 2010 were included. Initially, the articles were sorted by title and key words. Only epidemiology or clinical articles, as well as those on snakebite healthcare in Europe, were selected. In addition, non-indexed documents referenced in the indexed articles were collected to complete the meta-analysis database whatever the document language: French, Italian, Spanish, Portuguese, all understood by the investigator, and German, Polish, Norwegian, Turkish, Croatian, with the help of translator softwares (http://translate.google.fr and SYSTRAN version 6Ò, San Diego, USA). The studies describing fewer than 5 cases or focused on specific clinical complications were excluded from the systematic review. Similarly, articles that did not specify the criteria for the collection of data (locality, modality, period and/or duration of the study, criteria for the selection of

87

victims, particularly with respect to diagnosis and seriousness of poisoning) were not included. Similarly, clinical articles on individual or peculiar complications were excluded. In situations where the total study population was not specified in the article, population figures for the study period were gathered by referring to official national documents or United Nations documents (www.un.org/esa/population) and consulting the Wikipedia online encyclopaedia. Information on mortality was gathered by analysing literature, consulting the VipersGarden (Schweiger, 2010) website and referring to the Epidemiological Centre on the Medical causes of Death (CEPIDC, 2010). The latter gathers data from the National Institute of Health and Medical Research (Institut National de Santé et de la Recherche Médicale -INSERM) for metropolitan France, with the code E905.0 up to 1999, and then X20 in effect from 2000. The distribution of European poisonous snake species was taken from information on The Reptile Database website (http:// reptile-database.reptarium.cz/) and from zoological literature. The data were analysed using meta-analysis techniques with the aid of Comprehensive Meta Analysis v. 2$2$050 software (BiostatÒ, Englewood, NJ, USA). A latitudinal geographical stratification was used based on the climate of the regions and the distribution of viper species. North of the 50th parallel (Northern Zone), the only species found is V. berus. Between 45 and 50 longitude, the two major European species, V. berus and V. aspis, are sympatric in various areas. Finally, in the area south of the 45th parallel, which corresponds to a region composed of mostly Mediterranean fauna, several snake species are present (Fig. 1). The French studies are distributed in Northern, Central and Southern Europe based on the appropriate biotopes. To evaluate the heterogeneity between studies and groups of studies, the I2 Higgins test was used (I2 < 25% ¼ low heterogeneity, 25% < I2 < 50% ¼ acceptable heterogeneity, I2 > 50% ¼ strong heterogeneity). All tests were conducted according to a random effect model (i.e., an open model in which effects are not predicted) that indicates some heterogeneity of the studies that occur by chance or may be linked to explicable but undefined variations. This approach makes it possible to estimate the variables by taking into account the heterogeneity of results and the weight of each study according to number and type of population under review (Borenstein et al., 2009). Comparisons of frequencies and averages were studied using Epi-Info 6.04dÒ (CDC Atlanta, GA, USA) and tendency curves with ExcelÒ 2007 (Microsoft Corp., Redmond, WA, USA). The comparisons were carried out using the c2 test, corrected (Yates) in case of low numbers, when the variables were normally distributed, and by using the Kruskall– Wallis test (c2KW) when the variable distribution was not Gaussian. The confidence interval used was 95%, and the significance level was 5%. Finally, MOOSE guidelines for systematic reviews of observational studies have been checked and the compliance with the standard verified. 3. Results Examinations of MedlineÒ and InistÒ-CNRS provided 3260 and 455 articles, respectively, published between

88

J.-P. Chippaux / Toxicon 59 (2012) 86–99

Fig. 1. Distribution of European viper species.

J.-P. Chippaux / Toxicon 59 (2012) 86–99

1980 and 2010 containing epidemiological or clinical data on European snakebites. On the basis of article titles and abstracts, 99 articles were selected. The article references helped to add 104 non-indexed articles, theses, chapters of books and international congress communications to the data for the meta-study. After the removal of already published data (6 articles), irrelevant articles, such as those considered to be incomplete or those that did not fit the criteria (118 articles), studies lacking enough methodological precisions (11 articles) and 3 which were absent from all the consulted libraries, 62 references remained (Fig. 1; Table 1). The resulting data set helped, a) to estimate the incidence, mortality and severity of envenoming, b) to identify populations that are predisposed to snakebites and c) to identify the frequency of antivenom use and any undesirable effects of antivenom administration. Most studies reported series of indoor patients and few others published by poison control centers might have mentioned phone calls which came from snakebite victims who did not consult or treated at home. As a result, some patients without envenomation were taken into account although they were outdoor patients while all patients with symptoms of envenomation were hospitalized; the latter represent a high majority of cases reported here. Subsequently, all snakebites are here considered for the calculus of incidence, while envenomations are recorded for the calculus of the morbidity. The heterogeneity of the studies was remarkable in regards to reported incidences of snakebites (I2 ¼ 93%). Neither the geographical stratification nor the quest for significantly different studies helped exclude the studies primarily responsible for the high heterogeneity. However, for all the other epidemiological variables, the heterogeneity seemed to be within acceptable limits. Hence, the studies from the literature sources were used in evaluating snakebite epidemiology. The annual incidence of snakebites was 1.06 [0.97–1.15] per 100,000 inhabitants without significant differences between the geographical zones being observed (c2KW ¼ 0.17; P ¼ 0.92; d.d.l. ¼ 2; Fig. 2): 1.03 [0.84–1.21] in the north, 1.02 [0.88–1.17] in the central Europe, and 1.10 [0.96–1.25] in southern Europe. There was no significant correlation between the incidence of snakebites and population density, but a nonsignificant downward trend correlating to population density was observed (Fig. 3). Few studies report on specific seasonal incidence (Pozio, 1988; Luksi c et al., 2006; Reading et al., 1995; Béar, 1990; CAP Lille, 2001; Petite, 2005; Curi c et al., 2009; de Haro et al., 2009; Casaretli and Ozkan, 2010). The present meta-analysis revealed that bites usually occurred from March to October and reached their peak between May and August, with lower figures reported in northern and central Europe than in southern Europe (Fig. 4). The annual number of bites in Europe, including western Russia and Turkey, on average, was given as 7992 [6860–9178], most of them (more than 90%) being hospitalised. Half of the bites occurred in central Europe, one third occurred in southern Europe and 15% occurred in northern Europe. In France, data on the detailed causes of mortality were available on the CEPIDC website, but the data only covered 1979 and later years. Between 1980 and 2008, 36 deaths (19

89

men and 17 women; 9 children below age 15, and 27 adults) caused by poisonous bites from local snakes were recorded, representing an annual mortality of 0.0021 per 100,000 inhabitants (CEPIDC, 2010). In other European countries, mortality documentation was more difficult to find, but the annual mortality rates given by local studies were similar. The heterogeneity of the studies concerning annual mortality variable was acceptable (I2 ¼ 26%), except for the rates recorded in southern Europe (I2 ¼ 59%). Average annual mortality was 0.0006 [0.0002–0.001] per 100,000 inhabitants, which was not significant according to the geographical partition (c2KW ¼ 2.52; P ¼ 0.28; d.d.l. ¼ 2; Fig. 5). In northern and central Europe, mortality was 0.0004 [0–0.001], and mortality was 0.0008 [0.0003–0.001] in the south. On average, 4 [0.7–7.7] deaths occurred annually in Europe. The frequency of asymptomatic bites and severe envenoming were obtained from studies with low heterogeneity (I2 ¼ 0% and 29% respectively). The prevalence of asymptomatic bites (11 [6–21] %) showed a decreasing gradient (c2KW ¼ 4.31; P ¼ 0.12; d.d.l. ¼ 2) from northern to southern Europe: 20 [7–45] % northern, 11 [6–22] % central and 7 [3–17] % southern. However, the annual prevalence of grade III poisoning according to Audebert et al. (1992; Table 2) was 13 [10–16] % (Fig. 6). This prevalence was the same throughout Europe (c2KW ¼ 0.3; P ¼ 0.86; d.d.l. ¼ 2) and was believed to concern 1032 [700–1530] patients. Annual incidence for grades II and III poisoning (3970 [2885–5050] patients for the whole of Europe) was estimated by a small number of highly heterogeneous (I2 ¼ 95%) studies. The incidence for grade II and III poisoning is important as it corresponds to the number of patients to whom immunotherapy is recommended, but these data should be considered with precaution due to their high heterogeneity. Studies that made it possible to determine age and sex showed acceptable, though limited heterogeneity (I2 ¼ 46% and 49% respectively). Children below age 15 and male victims represented 27 [16–43] % and 62 [54–69] % of patients, respectively. In addition, the geographical gradient of snakebite prevalence of children below 15 years demonstrated a remarkable downward trend from northern to southern Europe (c2KW ¼ 8.8; P ¼ 0.01; d.d.l. ¼ 2): 38 [29–49] % northern, 29 [21–39] % central, and 17 [12–23] % southern (Fig. 7). There was a decreasing, but not significant, gradient in male-snakebite prevalence from north to south (c2KW ¼ 4.45; P ¼ 0.11; d.d.l. ¼ 2), respectively: 66 [56–75] % northern, 65 [58–71] % central, and 56 [50–62] % southern. When the severity of envenoming is considered, no difference was noticed between children and adults (Pozio, 1988; Luksi c et al., 2006; Karlson-Stiber et al., 2006; Petite, 2005; de Haro et al., 2009; Grönlund et al., 2003)., apart from a French study (Harry et al., 1999) in which 23 children did not show signs of grade III poisoning, while the frequency observed in 137 adults was 24% (c2 ¼ 5.3; P ¼ 0.02). Similarly, no difference in severity was observed between male and female victims. The location of the bite was documented in moderately heterogeneous studies (I2 ¼ 46%), except in central Europe where I2 ¼ 63%. The upper limb was affected in 49 [41–57]

90

J.-P. Chippaux / Toxicon 59 (2012) 86–99

Table 1 Studies used for epidemiological assessment of snakebite in Europe (I ¼ incidence; M ¼ mortality; S ¼ severity; R ¼ population at risk; G ¼ pregnancy; L ¼ bite site; P ¼ season of the bite; T ¼ treatment and/or adverse reactions). Countries

Years

Covered population (100000)

Number of cases

Data provided

References

Andorra, Bosnia Croatia Croatia Denmark, Sweden Europe Europe Europe Finland France France France France France France France France France France France France France France France France France France Germany Germany Greece Herzegovina Hungary Italy Italy Norway Norway Poland Poland Poland Portugal Russia Spain Spain Spain Spain Spain Sweden Sweden

1984–1988 1983–2006 1980–1995 1982–2002 1900–1947

0.5 45 8.5 50 57

20 389 439 542 5031

I, I, I, I, I,

Gonzalez, 1991 Curi c et al., 2009 Radoni c et al., 1997 Luksi c et al., 2006 Marquart, 1951