In recent years, snakebite has emerged as a significant tropical health concern, with global implications. The World Health Organization (WHO, 2023) estimates an alarming 1.8 to 2.7 million cases of snakebite envenoming annually, resulting in an estimated 81,410 to 137,880 fatalities worldwide. In 2019, an estimated 63,400 people (95% uncertainty interval 38,900–78,600) died globally from snakebites (GBD 2019 Snakebite Envenomation Collaborators, 2022). This corresponds to an age-standardized mortality rate (ASMR) of 0.8 deaths (0.5–1.0) per 100,000 population. This issue is particularly acute in India, a tropical and subtropical nation with a substantial rural population, where approximately 45,900 snakebite-related deaths occur annually (Mohapatra et al., 2011). A more recent study by Suraweera et al. (2020) revises the annual snakebite death toll in India to a staggering 58,000. In response to the snakebite crisis, in 2019, the WHO launched a strategic initiative, aimed at preventing and controlling snakebite, setting an ambitious target to reduce deaths and cases of serious disability by half by 2030 compared to the 2015 baseline (World Health Organization, 2019).

Researchers have extensively investigated the environmental factors linked to snakebites, such as humidity, precipitation, and temperature (see, e.g., Bravo-Vega et al., 2022, Ferreira et al., 2020, Phillips et al., 2019). While several studies have tried to associate snakebite occurrences with rainy seasons, there is significant variability in how precipitation impacts snakebites. For instance, Angarita-Gerlein et al. (2017) found a significant correlation between precipitation and snakebite incidence in 49.36% of municipalities in Colombia. Similarly, Phillips et al. (2019) noted a decrease in snakebite incidence during droughts and an increase following precipitation in their study of snakebites in California from 1997 to 2017. These findings differ from popular press reports of increased snakebites during drought conditions (Clarke, 2014, Scauzillo, 2014).

The reasons for the rise in snakebites in regions like San Diego and Southern California are debated among experts (SnakeProtection, 2024). Some attribute it to drought conditions, which reduce prey animals for snakes, leading vipers to move into areas with more human activity. Wildfires worsen this situation by disrupting snake habitats, resulting in more snake bites among humans and dogs. However, other experts argue that snakes are adapted to drought conditions and that the increase in bites is due to seasonal changes. As daylight hours increase, rattlesnakes become more active, searching for food and mates. Moreover, the higher number of people hiking in spring and summer increases the chances of encounters with snakes.

Recent findings by Bravo-Vega et al. (2022) highlight the heterogeneous effects of rainfall on snakebite incidence. They found that the relationship between rainfall and snakebite incidence varied across regions. Rainfall was found to drive snakebite incidence only in areas with distinct dry seasons, where rainfall becomes a limiting resource, while temperature did not appear to modulate snakebite incidence.

Our hypothesis is that the inconsistent effects of rain (temperature) on snakebite incidence are due to spatial variability. In some areas, precipitation (temperature) may be protective, while in others, it may increase risk. This could depend on factors such as the species of snake or the type of agriculture prevalent in the region. To test this hypothesis, we employ a case-crossover design model, a well-established method for assessing the health impact of short-term exposures (see, for instance, Konstantinoudis et al., 2022, Chen et al., 2016). This model contrasts the exposure on the date of each snakebite case with the exposure on one or more ‘control days’, effectively accounting for individual-level covariates (e.g., age, sex) that remain constant over the brief period spanning the case and its controls. To accommodate spatial variability in the effects of rain (temperature) on snakebites, we extend the classical case-crossover model by proposing a spatial case-crossover model. Traditional case-crossover models typically assume constant effects across space, which may obscure important regional differences in exposure-response relationships. In contrast, our proposed spatial case-crossover model introduces spatially varying coefficients, modeled using a Matérn covariance structure, to allow the exposure effects (rainfall or temperature) to vary smoothly across geographical locations. This extension enables us to model and quantify spatial heterogeneity in environmental effects, which is critical in a country like India where ecological, climatic, and socio-economic conditions vary substantially across regions. By borrowing strength across space, this framework not only improves estimation precision but also enhances the interpretability of spatial patterns in epidemiological risks. We validate this model through a simulation study. Additionally, we consider various exposure metrics, including precipitation (temperature) on the same day as the snakebite (lag0), one and two days prior (lag1 and lag2), the maximum precipitation (temperature) over the previous three days (3-day max), and cumulative precipitation (temperature) over the preceding week (weekly total). Our study utilized mortality data from India’s Million Death Study (MDS), conducted from 2001 to 2013 (Jha et al., 2005), with further details provided in Section 2.1.

The subsequent sections of this paper are organized as follows: Section 2 presents a description of the data; Section 3 details the statistical methodology; Section 4 reports the study’s findings, and Section 5 presents the key conclusions of our study and offers additional discussion on potential limitations and future directions.