Japanese encephalitis (JE) virus is an arthropod borne viral zoonosis that is endemic throughout eastern, south-eastern and southern Asian countries [1], [2]. JE virus (JEV) infection can cause irreversible damage to the central nervous system of humans, who serve as incidental ‘dead end’ hosts because they do not produce sufficient viremia to infect mosquitos [3]. Approximately, 60% of the world's population lives in JE endemic regions [4] and a 2011 review estimated that the annual incidence was 1.8/100,000 and 5.4/100,000 for children 0–14 years old in 24 JE endemic countries [5]. The majority of human infections are asymptomatic, and only a small ratio (1∶25 to 1∶1,000) develop clinical infections [6]. However, the prognosis for people who develop encephalitis is poor: approximately 25% die, and 30% to 60% of survivors suffer from neurological sequelae [7], [8].

JEV transmission is complex, involving numerous vertebrate and mosquito species, and is poorly understood in Bangladesh. Studies from other Asian countries show that Culex species are the primary vectors driving transmission; Aedes mosquitoes are also competent vectors but likely play only a minor role [9], [10]. Host species for the virus include ardeid wading birds and some domestic animals. In particular, pigs appear to play a major role in transmission cycles due to large-scale viral amplification and relatively high viral titers [11], promoting onward transmission [12], [13]. Several other domestic animal and bird species including cattle, goats, dog, ducks and chickens also become infected, but because they produce low level of viremia for a brief time, they are unlikely to play a significant role in transmission [14]–[17].

Hospital based acute meningoencephalitis surveillance in Bangladesh began in 2003 and identified that JEV infection was responsible for 6% of all encephalitis at surveillance hospitals [18]. Further work to characterize the burden of JE in 2009 estimated that the incidence was highest in the northwest part of the country, with 2.7 cases per 100,000 population per year [19], which is similar to its incidence in other JE endemic countries before the introduction of JE vaccine into national immunization programs [20], [21].

Although Bangladesh is a predominantly Muslim country, pigs are raised by some ethnic minority communities and in nomadic herds [22]. Understanding the distribution of pigs and the extent to which they get infected with JEV will help to define their role in the JEV transmission cycle. The objectives of this study were to describe the transmission dynamics of JEV among pigs in the northwestern part of Bangladesh, where human incidence is high, [19] and to estimate the potential impact of vaccinating pigs on JE incidence in the pig population.

We identified 11,364 pigs (Sus scrofa) in the study area with a mean of 1.5 pigs per km²; the majority (61%, n = 6963) were over 12 months of age and 50% were female. Most (88%, n = 9977) of the pigs were raised in backyards (median per backyard: 1 pig, range: 1–25) with the remainder (12%; n = 1387) raised in 28 nomadic pig herds (median pigs per herd: 39, range: 6–145) (Table 2). Of the 28 herds, we mapped grazing routes of 9 herds. The pig herding routes had an average length of 108 km per year (range: 30–329 km per year). We also visited 20 (52%) of the 111 unions (administrative cluster of villages) where we did not receive any information on pig raising through pig raiser's social informational network and could only identify an additional five pigs, suggesting our approach was able to capture the vast majority of pigs.

We identified 20 locations in our study area where pork and live pigs were sold. Pork was sold either daily or weekly in a pig slaughterer's backyard, which in most cases (80%) was near an established live pig market. In addition, there was one pig fair in Naogaon District held once a year during December-January. All nomadic pig herders participated in that fair to sell their pigs to backyard pig-raisers and other pig herders. Commercial pig dealers who sold pigs and pork at the markets throughout the country also participated in this fair. Other than local pig trade between the backyard farmers, the pig-raisers received the majority of their piglets from the herders. Backyard raisers mostly bought 2–3 months old piglets, which they raised for about a year or two before slaughtering for their consumption or for money. On special occasions, such as weddings and spiritual ceremonies, the backyard raiser's preferred to buy older pigs for immediate slaughter. The pig traders who sold the pigs in markets for slaughter usually bought the 5–6 month old pigs from the herders, and supplied the pork markets across the country.

We tested 312 pigs for JEV antibodies and 30% had evidence of previous infection (Table 2). Pig-raising households were primarily located in the central part of our study area (Figure 1). We failed to reject the null hypothesis of no spatial heterogeneity in the location of seropositive pigs (p-value of 0.3, Figures S1 and S2 in text S1), suggesting that the extent to which pigs were infected did not differ across the study region.

Evidence of past JEV infection increased consistently with age, supporting a constant average force of infection in the pig population. Approximately 50% of pigs had evidence of past infection by 3 years of age (Figure 2). A force of infection of 0.20 per year best fitted the age distribution of seropositive cases (1/8 likelihood interval of 0.16–0.25, Figure S3 in text S1) indicating that 1 in 5 susceptible pigs got infected each year.

To describe the dynamics of JEV transmission in pigs we built a compartmental model and fitted it to the force of infection estimate. Using the model, we estimated that the basic reproductive ratio (R₀) among pigs was 1.2. Therefore, on average, each infected pig would transmit to 1.2 other pigs (via mosquitoes) in a completely susceptible population. We subsequently explored the impact of vaccinating pigs. We found that vaccinating only half of all pigs each year, a potentially feasible goal from a vaccination campaign, would result in an 82% reduction in incidence, assuming that the vaccine efficacy was 95% and that 5% of pig infections result from external introductions; we would expect a small marginal benefit if we were able to vaccinate 75% of pigs (89% reduction in annual incidence) (Figure 3). In the base model, vaccinating only a quarter of pigs would result in a 61% reduction in annual incidence. We conducted sensitivity analysis that varied both the rate of external introductions and the vaccination coverage (Figure 3C). We found that even if 10% of pig infections were externally introduced, vaccinating 50% of the pigs would still result in a 72% reduction in annual incidence after five years. In a second sensitivity analyses we assumed 50% vaccine coverage, but assumed a lower vaccine efficacy of 50% (compared to 95% in the base model). Under these assumptions, vaccinating half the pigs would still result in a 54% reduction in JE incidence (compared to 82% in the base model).

Characterizing incidence patterns in pigs is crucial to understanding both the dynamics of JEV transmission in pig populations and the potential impact of intervention efforts. Here we conducted an intensive pig census across an area that has experienced high levels of JEV infection in humans [19]. We identified over ten thousand pigs with at least one pig found in approximately half of the unions in the study area. Evidence of past JEV infection in the pigs was found across the study area with no spatial heterogeneity in the location of seropositive pigs. There was a clear increase in seropositivity by age, characteristic of endemic diseases and we estimated that 20% of susceptible pigs get infected each year. We are not aware of previously published estimates of the basic reproductive number or the force of infection of Japanese encephalitis in pig populations.

Our model shows that by vaccinating only half (or 4,000) of the susceptible pigs each year, the incidence among pigs would be reduced by >70%, even if up to 10% of pig infections were from other animal reservoirs, such as wading birds. In other parts of Asia, pig immunization against JEV was discontinued because of logistical difficulties in keeping large herds with high annual turnovers vaccinated [29]. However, in this setting, there were fewer pigs than on large commercial pig raising operations and pigs lived for one year, on average, meaning that vaccination once per year could be sufficient to provide protection. These factors might make pig vaccination more logistically feasible in these communities. Pig-raisers in this part of Bangladesh are marginalized populations and economically disadvantaged [22] and JEV infection among sows can result in abortions and deaths of newborn pigs [30]. Although there are no published reports of using JE pig vaccination as an effective intervention to improve poverty among pig raisers, the small reduction in pig miscarriages and stillbirths from vaccination could translate into meaningful reduction in economic losses for this economically disadvantaged group. Pig raising communities are typically wary of outside intervention, and currently receive very little in the way of veterinary services provided by government-owned livestock clinics in the country [22]. Therefore, interventions to vaccine pigs could be more difficult to implement than in the setting of large commercial farms where previous vaccination efforts were attempted. Nonetheless, a pig vaccination strategy might be feasible and acceptable, particularly if it would provide some benefit to pig raisers. Pig-raising communities maintain strong relationships with one another, including through the pig trade. These close networks may provide an opportunity to efficiently access pigs and disseminate communication messages for a vaccination campaign, such as during the yearly pig-marketing event in Naogaon.

The findings from this study are subject to some important limitations. First, the number of pigs in this area may be underreported by our census if large pig-raising communities were not identified through our snowball approach. However, this appears unlikely as efforts to identify pigs in unions where no pig-raisers were reported through our snowball approach yielded very few additional pigs. Second, it is possible that we have overestimated the seroprevalence of pigs in our study sample due to cross-reactivity of antibodies to other flaviviruses [16], although there is no evidence that these viruses are endemic in pig populations in Bangladesh. A study of human encephalitis patients residing in this area found no evidence of infection with West Nile Virus (ES Gurley, personal communication), a possible viruses that often produce cross-reactions in the region [16]. Finally, we do not know the proportion of pig infections that originate from other host species. However, sensitivity analyses that varied this assumption widely still resulted in substantial reduction in pig infections from vaccination.

The high level of seropositivity in pigs in the study area suggests they may be a key reservoir for the virus and could contribute importantly to JE risk in humans. If pig infections do increase human risk substantially, reducing incidence in pigs could also result in reduced incidence in humans. All countries that have effectively reduced the burden of JE in human populations have done so through human vaccination campaigns [31]. However, there are currently no plans to introduce JE vaccine into the expanded program for immunization in Bangladesh. Given the poor prognosis for humans with JE disease and the lack of human vaccine in this area, vaccinating pigs may represent an opportunity to reduce disease risk in humans. For example, one study from Assam, India [32], which borders Bangladesh in the north, showed that human risk of JEV infection was reduced by 66% in areas where insecticide treated mosquito nets were used to prevent infections in pigs. Further studies are required to estimate the contribution of pigs in determining JE risk in humans in this area and to understand if preventing pig JEV infections could also confer a human health benefit, in the absence of human vaccination.