Rat-atouille: A Mixed Method Study to Characterize Rodent Hunting and Consumption in the Context of Lassa Fever

The population of Sierra Leone is composed of more than 18 different ethnic groups, belonging to distinct language groups. While socio-political confederations may cut across linguistic lines with much ‘creolisation,’ the majority group is the Mende (32.2%), closely followed by the Temne (31.8%). A majority of the population (60%) is rural with a literacy rate of 43.3% (SSL 2007; CIA 2014) and just over half of the population lives below the poverty line (World Bank 2014).

The study was conducted between May and June 2014 (8 weeks) in Bo district in the Southern Province of Sierra Leone. In this district, the Mende are the dominant ethnic group (79%), followed by the Temne (7%), with Islam (72%) and Christianity (27%) the principle religions (SSL 2010). In rural areas, farming, fishing, and hunting serve as means of subsistence or to generate cash income.

The qualitative component of this study consisted of in-depth semi-structured interviews (IDI, n = 21), informal discussions, focus group discussions with ad hoc recruited participants (FGD, n = 4), and direct observations over the entire duration of the study time. Fourteen villages were purposively selected to include those of varying size, location, and distance from main transport axes. Between 1 and 3 discussions (IDI and/or FGD) were done in each village. Selection of villages was restricted according to travel distance for the study team (max = 40 km). Individuals were chosen purposefully to achieve representation from various groups (socio-economic status, religion, ethnic group, age, sex), those knowledgeable of the community (chief, teachers), and those known to engage in the behaviors of interest. These people were identified making use of the long-standing and continuous presence of our local research team in the area since 2010. Discussions were carried out in Mende, Krio, or English and facilitated by a translator using guidelines for IDI/FGD translations (Oxfam 2012). Prompts were adapted in an iterative process to inform new data collection and were supplemented with photographs of rodent species. The prompts relevant to this report covered food security, knowledge of rodents (e.g., vernacular names, habitats, morphology, and behavior), specific interactions with rodents (e.g., avoidance, hunting, preparation, and consumption), and knowledge of LF (e.g., transmission routes, symptoms, and prevention). Prompts were refined midway during the fieldwork as new themes emerged and included new topics on LF sensitization messages and attitudes toward our research. Interviews lasted on average for 1 h and were conversational and open-ended, treated as occasions for a mutual exchange of information rather than an opportunity to extract specific data. The research team devoted as much time as possible to informal interactions with the communities to establish trust. IDIs and FGDs were recorded and transcribed, and observations were documented with field notes and photographs. For analysis, transcripts were reviewed on a daily basis using a narrative analysis, focusing on the ways in which experiences of rodent interactions were relayed, their emotional content and temporal structure and thematic analysis, drawing out repeating motifs between the responses. A priori codes corresponding to biomedical risk factors for disease transmission (e.g., procedures for handling live and dead rodents) were developed prior to fieldwork. Because of the paucity of previous research, themes related to the wider socio-economic context of human–rodent interactions were generated using emergent codes. These were further discussed with the translators to verify that interpretive categories were correct. Text segments were then color-coded according to the categories of interest.

A cross-sectional questionnaire survey was carried out midway during the fieldwork by local staff based on findings from the qualitative component. Questions were in English with use of Mende terms and answer format was either single choice, multiple choice, or open-ended. Open-ended answers were used to determine any emerging themes that could inform the qualitative component. The quantitative survey contained 55 questions covering all forms of contact with rats (contact in homes, contact during hunting, butchering, and consumption) as well as food security and knowledge of LF. The survey was conducted with smartphones using OpenDataKit software (http://​opendatakit.​org), automatically collated on formhub (https://​formhub.​org) and exported as Microsoft Excel files.

The quantitative survey was carried out in nine villages selected by convenience as described above, with population size ranging from 500 to 1500. Because village populations varied in size, we initially intended selection with a constant sampling fraction (6%) to determine the sample size per village. However, this approach was abandoned midway through the survey when Ebola virus disease (EVD) was confirmed in the neighboring district, thus not all villages have the same proportional representation. Selection of individuals was carried out according to the WHO EPI Coverage Survey method (WHO 2008) due to the unavailability of a sampling frame at village level. A maximum of two individuals were surveyed per household, alternating between adult male, adult female (>18 years old), young male, and young female (<18 years old). Visits were done in the morning or evening as this is when most of the villagers are present and available.

In total, 524 subjects were recruited. Seven records were excluded because no village name was indicated, 57 because respondents lived in a major city (no comparative qualitative work was done in urban areas), 20 because respondents lived in other villages than the nine selected villages, and 11 because missing data on at least one outcome. The final sample size was thus 429. Sample size varies by question because skips were used to avoid asking redundant or irrelevant questions. Records with answers stating “unknown” or “don’t know” were not included in the analysis for that particular question. We provided a simple statistical description of the study participants from all nine villages. We calculated actual sampling fractions for each village, which we used as weights to account for differences in probability of selection. Accounting for the sampling design by Taylor linearization, we estimated proportions of subjects with respect to knowledge of Lassa fever, rats hunting, rats preparation, and rats consumption. We then carried out univariable and multivariable logistic regression models on each of the following outcomes: “hunted rats in the past 3 months,” “ever hunted rats,” “prepare rats at present,” “ate rats in the past 3 months,” and “ever eaten rats.” The explanatory variables used are “think that eating rats can cause disease (yes, no),” gender (female, male), age group (5–14, 15–24, 25–39, 40 years or above), educational level (none, primary, secondary or above, other), ethnicity (Mende, other), and religion (Muslim, Christian). No model selection approach was used. Adjusted Wald tests were used to assess whether there is evidence of association between explanatory variable and outcome. All analyses except the description of the study participants were performed with finite population correction using villages as strata, and were conducted with STATA 13. (StataCorp. 2013, TX: StataCorp LP).

The study was approved by the ethics committee of the Government of Sierra Leone, Charité Berlin, and the Royal Veterinary College. An informed consent form was read out in English, Mende, or Krio to each participant, and consent was obtained from all individual participants included in this study. At the end of every visit, villagers were given a specific opportunity to ask questions about LF.

During discussions, nearly all informants were able to correctly identify and name individual species from photographs. Species are distinguished, and sometimes named, according to their physical characteristics (color, markings, hairiness, size, shape, smell), their behavior (diet, nocturnal or diurnal), or the location where they are found (house, village, bush, swamp). Shrews (Crocidura spp.) and small- to medium-sized rodents (such as Lemniscomys striatus, Lophuromys sikapusi, Mus musculus, Nannomys spp., Rattus spp., Praomys spp., Mastomys spp.) are collectively termed “rats” in English, “arata” in Krio, and “nyini” in Mende. In our study, we use the same categorization when referring to the word “rat.” Larger species of rodents, such as the cane rat (Thryonomys swinderianus) and the Gambian pouched rat (Cricetomys gambianus) do not fit into this category. Individual species are referred to using their vernacular name in Mende (Table 2). Nevertheless, Mastomys spp. (M. erythroleucus and M. natalensis) and Praomys spp. (P. rostratus and P. tullbergi) are morphologically very similar, especially when observed at dusk, the peak activity time for both species (Duplantier and Granjon 1992), hence these species are not distinguished and together are termed “vorley.” Overall, species can also be grouped into “bush rats” (e.g., Lemniscomys striatus, Lophuromys sikapusi, Mastomys spp., Praomys spp.) or “town/village rats” (Mastomys spp., Rattus rattus, Mus musculus), although these are flexible categories that vary according to where the animal is mostly seen. M. natalensis is considered both a bush rat and town rat as it is confused with either M. erythroleucus (a bush rat), Praomys spp. (a bush rat), or Rattus rattus (a town rat). Our ongoing ecological studies indicate that M. natalensis and R. rattus share the commensal habitat in rural villages around Bo (mean ratio of M. natalensis to other commensal rodents: 60%, 222/373, range 25–84%, unpublished data).

Most informants had previously heard of LF and considered it a serious and fatal disease, expressing familiarity with the symptoms and the special burial practices required for deceased cases (such as the use of body bags and not touching the deceased during burials). For the quantitative survey, less than half of respondents associated LF with animals (38.3%, 173/429, Table 3) and during discussions none of the informants knew the exact LF reservoir but frequently mentioned that shrews in particular could transmit LF.

The reasons for this belief are multifactorial: shrews, whether caught in the town or in the bush, are deemed “different from other rats,” in terms of their behavior (aggressiveness), diet (carnivorous), and morphology (musk gland, elongated snout), an understanding that may have been compounded by errors in the delivery and/or comprehension of previous sensitization messages citing shrews as the reservoir for LASV. Similarly, town rats also have a “repellent” morphology (poor coat condition, maggots, on skin) and are seen to live in unhealthy places that are in close proximity to humans (such as in toilets, cemeteries, and garbage dumps), which tends to associate this category of rat with endemic diseases such as LF, malaria, yellow fever, typhoid fever, cholera, and EVD.

The term “hunting” is used here to describe any form of catching or trapping of rodents, from subsistence activities to child’s play, but excludes other specific rodent control methods such as rodenticides and cats.

Traps are built specifically to catch rats. “Torley” (Fig. 1a) and “kongoumie” (Fig. 1b) are trigger traps that ensnare a rat when it touches the bait. Torley is the most commonly used and easy to construct, and can be carried in bundles of up to 50 pieces. Gbushie (Fig. 1c) is a heavy clay structure that crushes the prey when it touches the bait. Traps are used mostly during the dry season because this is when rats are believed by respondents to be most active. In some villages, setting traps in the bush was described as a child’s activity, but in other villages “even the elders” use them. Rat hunting can also be done by “brushing,” an agricultural practice that refers to clearing land with machetes and that can involve multiple people (from 2 to 40 individuals) who “brush” in a circular pattern to kill rats with machetes as they are flushed out from the grass. Traps and brushing are used to kill rats for pest control or for food. A similar method for catching rats is to surround houses, kitchens, or farmhouses when they are being dismantled or repaired, which reportedly resulted in catches of 20–45 rats at a time. Brushing is a predominantly male activity, although women and children will also participate if present, or occasionally organize their own brushing. Hunting of rats by women tends to be opportunistic, for example if rats are encountered in granaries. Unlike meat from larger wildlife species, rat meat is not hunted for commercial reasons and is rarely found for sale. Direct contact with rats and their fluids can occur during any of the described forms of hunting.

The qualitative and quantitative surveys showed that hunting activities are often practiced (Table 3). Many informants (qualitative) and respondents (quantitative) declared having hunted over their lifetime (95%, 18/19 and 42.4%, 186/429) and in the past 3 months (47%, 8/17 and 11.4%, 54/429). The quantitative survey also indicates that Mastomys spp. and Praomys spp. are the most commonly caught species (Fig. 2). Further, more than two-thirds of rat hunters handled live rats during hunting (69.0%, 125/186) and about one-third reported having been in contact with urine or having been bitten (32.2%, 61/186 and 28.0%, 53/186, respectively, Table 3).

Rats are singed over a fire to remove the hair, eviscerated, and sometimes butchered (Fig. 1d–g). Organs such as kidneys, liver, and heart may be kept. Rats are then smoked, grilled, and/or stewed. The meat, including bones, is eaten on its own or added to pulse dishes. No basic hygiene measures were observed during the study period despite contact with blood being reported frequently in the quantitative survey (99.6%, 188/189, Table 3). During our direct observations, nobody washed their hands, in contrast to the responses given in the quantitative survey (67%, 138/189, Table 3). Both women and men of all ages are involved in preparing rats, and these are either eaten alone or shared with friends and family depending on the size and location of the catch.

All species of rodents are eaten except for shrews (irrespective of where they are caught) and town rats (Fig. 3). Some Muslims reported not eating L. striatus because they interpreted the striped coat as divine writing. With very rare exceptions, adults reported never eating town rats or shrews. Those that did justified it because of hunger or as an act of defiance toward (public health) authorities. When interviewed without the presence of an adult relative, children were more readily to admit eating both bush and town rats. Parents acknowledged that their children probably hunted and ate rats in hiding, and that there was little that they could do to stop them.

Preparing and consuming rats was carried out without ceremony or occasion. A commonly cited reason for consumption was its supplementary value: rat meat provides a “very important source of protein” and is necessary for maintaining a “balanced diet”, satisfying a “want of meat”, born “out of poverty.” It is important to note that this desire for nutritional security was expressed not in terms of quantity but for a diverse quality of food sources. All informants bar two (one adult, one child) stressed that they would not go hungry if rat was no longer available. In contrast to bland foods such as rice or beans, rat meat was overwhelmingly described as ”a sweet (tasty) meat” that makes a “very good dish,” so much so that it was popularized in a song by singer Amie Kallon in the 1970s describing rat meat as “sweeter” than cow meat. About half of the informants stressed that they would continue eating rat meat even if cow meat was available and affordable. Those that expressed a preference for beef emphasized the comparatively fewer bones and more meat in cows, though some added that large rodents, which combine the benefits of both rat and cow meat, are preferable to both.

However, questions of consumption preference were ultimately trumped by pragmatism: it was considered wasteful to throw away bush rats killed for pest control. Many informants explained that while other sources of protein are available (larger wildlife species, fish, chicken, frozen fish) rat meat is free, easy, and fast to catch. Informants no longer consuming rats reported eating less meat or spending more money to buy alternatives. Among those who no longer ate rats (or said they didn’t), fear from disease, in particular LF, was by far the single most important reason. Linguistic responses by those who confirmed rat consumption included words or expressions such as “tempted,” “trying not to,” and “nearly stopped,” suggesting that whatever the reasons for eating rats, they outweighed the fear generated by LF.

Many informants talked freely and casually about rat consumption, but some would contradict themselves, modify their statements, change tense, or cut the conversation short. Other informants opened up as trust was established (“I don’t want to lie,” “I will be honest”). In addition, the proportion of behaviors reported during IDIs and FGDs contrasted with the proportion reported during the quantitative survey (Tables 3 and 4). Rat consumption was more often reported during the IDIs and FDGs.

In the multivariate analysis, gender was the only variable that was consistently and significantly associated with a history of hunting rats in the past 3 months and over lifetime, preparation of rats, and consumption of rats in the past 3 months and over lifetime (Table 5). All other variables (age, education, religion, and ethnic group) were not significantly or consistently associated with these behaviors.

In response to the clear discomfort that the topic of rat consumption generated for some of our informants, interview prompts were refined to explore the reasons for that reticence. A recurrent theme was the fear of talking about rat consumption to strangers. Informants explained that they were afraid to acknowledge this practice because they had been advised against eating rats through sensitization messages from health care workers and through the radio. The survey team aroused suspicion, as they were identified as government workers, taking notes on electronic devices to send to “higher authorities.” It was thought that if authorities knew about who ate rats, they might prevent those persons from accessing health care services, take them away for testing, or even inject them with LASV. Other reasons were also mentioned, such as fear of blackmail (from the research team), and being excluded from potential benefits that the study might bring. However, there was no sense of shame in admitting to eat bush rats, in contrast to town rats and shrews.