Japanese encephalitis (JE) is a serious neurologic disease and a major health concern in many Asian countries, including Cambodia. It is caused by Japanese encephalitis virus (JEV), a mosquito-borne flavivirus that infects several vertebrate species, although mainly pigs, horses, and humans develop clinical symptoms [1
]. The major mosquito vectors prefer to breed in rice fields, and pigs are considered important amplifying hosts [2
]. Infection in humans may cause encephalitis and death, or neuropsychiatric sequela among survivors [3
]. In endemic areas, such as Cambodia, JE is considered a childhood disease, as most adults have acquired active immunity through repeated exposure to JEV [1
]. The disease is generally considered rural, and proximity to rice fields and pigs are associated with increased risk of transmission.
In 2018, Cambodia had a population of 16.2 million people, of which 12.4 million resided in rural areas [4
]. Rice cultivation and pig farming are common practices, especially in the countryside [5
]. Many families are smallholders that keep their pigs close to their homes, with less than 20% of the farms counting as large scale [5
]. Therefore, a major part of the population is at risk of infection. Cases of clinical JE occur year-round in children under 15 years of age. In 2007, the incidence was estimated to be 11.1 per 100,000 children [6
]. The circulation of JEV is high, as most pigs seroconvert before 6 months of age [7
]. Apart from being a threat to human health, JE manifests in reproductive disorders in infected sows and boars [8
] and may have a negative impact on families’ livelihoods, as pig farming generates both income and food.
The main purpose of this study was to determine the seroprevalence of JEV antibodies in smallholder pigs in rural parts of Kampong Thom, Preah Vihear, Ratanakiri, and Stung Treng provinces, northeastern Cambodia. Although JE is considered endemic, the circulation of JEV among pigs has previously only been studied in the southern part of the country. Another purpose was to identify possible associations between serologic status and animal characteristics, pig management, and reproductive disorders, and to investigate the smallholders’ knowledge of mosquito-borne diseases and use of preventive measures. In 2016, vaccination against JE was incorporated into the national childhood immunization program [9
]. This study explores how well this program has succeeded in reaching children in these areas.
In total, 139 households were included in the study, of which 39 were located in Kampong Thom, 34 in Preah Vihear, 35 in Ratanakiri, and 31 in Stung Treng. Of the respondents answering the questionnaire, 75.4% were female and 24.6% were male. The mean age was 41 years, ranging from 18 to 74. Most had finished primary school (42.6%), but 34.6% had no education, and only one respondent had attained a college or university degree. The share of uneducated respondents was highest in Kampong Thom (48.7%) and lowest in Stung Treng (12.9%). The mean number of pigs kept by the households was 4.2, ranging from 1 to 20. No pigs were kept in buildings that provided protection against mosquitoes.
In total, 242 pigs were sampled, of which 63 came from Kampong Thom, 51 from Preah Vihear, 63 from Ratanakiri, and 65 from Stung Treng. The sampled pigs were between 3 months and 5 years of age, with the mean age being 9 months. The majority (60.9%) were female. In Preah Vihear and Stung Treng, over 90% of the sampled pigs were indigenous breeds compared with Kampong Thom and Ratanakiri, where around half of the pigs were indigenous breeds. The rest of the sampled pigs were mainly exotic breeds and very few were crossbreeds.
Of the collected samples, most generated valid test results. However, in Preah Vihear, only eight samples could be interpreted since the OD values of the positive and negative test controls were similar on one of the ELISA plates. In total, 185 (93.0%) of the 199 samples that generated valid test results were positive (Table 1
). All pigs over 6 months of age had detectable JEV antibodies compared with 89.1% of pigs under 6 months of age. However, age was not significantly correlated with serologic status (p
= 0.087). There was a significant difference in seroprevalence between the provinces (p
< 0.001). The seroprevalence was highest in Stung Treng (98.5%) and lowest in Preah Vihear (62.5%) (Table 1
and Figure 1
). The mean age of seropositive pigs was 8.6 months (95% confidence interval (CI) 7.2–10.1) compared with 4 months in seronegative pigs (95% CI 3.5–4.5). Kampong Thom had the highest share of tested pigs over 6 months of age, while Stung Treng had the lowest (Table 2
). All pigs that were crossbreeds tested positive, as did most pigs of commercial (96.8%) and indigenous (91.0%) breeds. Type of breed was not significantly associated with seropositivity (p
Households with at least one seropositive pig were considered “seropositive households.” Of the households that let their pigs roam freely during the dry season, the rainy season, or both seasons, 92.3% had at least one pig that were seropositive compared with 94.1% of the households that always kept their pigs confined or tethered. Around 91% of households that were in sight of rice fields were seropositive compared with 97.4% of other households. Neither housing system (p = 0.668) nor proximity to rice fields (p = 0.213) were significantly associated with serologic status of households.
The multivariable logistic model included the variables identified as associated with pig level seropositivity at p
< 0.2. Results showed that increasing age and the province Stung Treng increased the odds ratio compared to Ratanakiri, while Preah Vihear had a lower odds ratio (Table 3
2.2. Reproductive Disorders
There were no reports of abortions in any of the sampled sows during the past year. Only one sow was known to have had both stillborn and mummified fetuses, two to have had stillborn fetuses, and one to have had mummified fetuses. Test results were only valid for one of these sows, which was positive for JEV. However, when respondents were asked if any of their sows had aborted during the past year, 24.4% answered ‘Yes.’ Stillborn or mummified fetuses were observed in 18.5% and 10.1% of households, while 9.3% and 11.2% had had weak or shaking piglets. There was no significant association between reported abortions and serologic status of households (p
= 0.336, Table 4
). However, none of the seronegative households had experienced any abortions. Among the seropositive households, 22.9% had experienced abortions. Furthermore, no seronegative households had observed any stillborn or mummified fetuses, whereas 31.5% of the seropositive households had observed either (Table 5
). However, these differences were not significant (p
= 0.519). Weak or shaking piglets had not been observed in seronegative households but were reported by 13.0% of the seropositive households. Similar to abortions, stillbirths, and mummified fetuses, there was no significant correlation between either weak (p
= 1.000) or shaking (p
= 1.000) piglets and serologic status of households despite the absence of such piglets in seronegative households.
2.3. Disease Knowledge and Prevention
Of all 139 respondents, 98 (70.5%) reported to have heard of Japanese encephalitis (Table 6
), whereas 21 (21.4%) claimed that they knew what it is but only six could deliver an explanation by describing clinical signs, such as fever, headache, diarrhea, and salivation. Of those who reported to have heard of JE, 93 (94.9%) knew that humans can become infected, but only seven (7.1%) knew that pigs are susceptible. Only six (6.5%) of those who knew that people could become infected claimed that they knew how. Three delivered an explanation, although only one correctly explained it as being transmitted by mosquitos. Of the respondents who knew that pigs could become infected, only one claimed to know how but gave no explanation. Four respondents had somebody in their family who had had JE, and four knew people outside of the family who had had it. One respondent reported that there had been one JE-caused death in the village. There were no significant relationships between disease knowledge (respondents who had heard of JE) and sex (p
= 0.661) or education level (p
= 0.316) of the respondent.
When respondents that had heard of JE were asked if their family was vaccinated, all replied but one. Of these, eight (8.2%) reported that the whole family was vaccinated against JE, whereas in 85 (87.6%) of the households, only the children were vaccinated against JE. In total, 93 (95.9%) households had vaccinated children. In all households (84) that answered the question of who had paid for the vaccination, respondents reported that it had been funded by the government. Households that were not vaccinated were asked to explain why not. Of these, one respondent explained that they did not know that people can become infected with JEV. Another respondent answered that they were afraid of vaccine side effects. There was no significant association between vaccinated households (any household with vaccinated family members) and province (p
= 0.149). However, there was a significant difference between the provinces (p
= 0.002) for households where all family members were vaccinated. In Preah Vihear and Stung Treng, the whole family was vaccinated in 27.3% and 6.9% of the households, and in Kampong Thom and Ratanakiri, there were no such households (Figure 2
Of 137 respondents, 134 (97.8%) reported to have heard of diseases being transmitted to people through mosquito bites. All 139 households used some sort of mosquito protection: 96.4% used bed nets, 56.8% used insect repellents, and 5.8% used covering clothes. In total, 21 (15.1%) households protected their pigs in some way. Of these households, 47.6% used insect repellents, 14.3% used smoke, one (4.8%) used mosquito nets, and one used light. The remaining households did not specify how they protected their pigs. There was no significant association between protection of pigs from mosquito bites and serologic status of households (p = 1.000).
The apparent prevalence of JEV antibodies observed in this study was very high, as all pigs over 6 months of age and 89.1% of pigs from 3 to 6 months of age tested positive with the ELISA. The high seroprevalence, as well as the slight difference between these age groups, was in correspondence with results of other studies on JEV infection in pigs in Cambodia. Duong et al. [7
] found that 95.2% of pigs between 6 months and 1 year of age were positive compared with 87.3% of pigs between 4 and 6 months of age. Cappelle et al. [10
] and di Francesco et al. [11
] found that more than 98% of the monitored pigs seroconverted before 6 months of age. The reason for the lower prevalence in younger pigs is most likely explained by the weaning of maternal antibodies around 3 months of age followed by the varying, although relatively high infection rate in susceptible pigs, as observed by Cappelle et al. [10
] and di Francesco et al. [11
]. A recent study in Vietnam found that the seroprevalence among pigs between 4 and 5 months of age was even lower; 53–74% [12
]. However, the seroprevalence among adult pigs in Vietnam has been found to be 100% [13
]. The ELISA test used in the present study does not distinguish between actively produced and passively acquired JEV antibodies. Therefore, it cannot be assumed that all positive pigs under 6 months of age have been infected with JEV, as some may be positive because they still are protected by maternal antibodies. As reported by Scherer et al. [14
], most pigs lose their passive immunity between 4 and 6 months of age, but for some, maternal antibodies are still detectable at 7 months of age.
There was a significant difference in seroprevalence between the provinces included in this study. However, for Preah Vihear, where the prevalence (62.5%) was remarkably lower than in the other provinces, 43 of the samples could not be interpreted and were thus excluded from the data analyses. Only eight samples remained, of which five were positive. The low number of included samples is therefore not representative of the whole province. However, Duong et al. [7
] found that the prevalence varied between the eight Cambodian provinces included in their study. In three of the studied provinces, all pigs between 2 months and 1 year of age were positive compared with 52.5% in Kampong Cham and 60% in Takeo. Because of the large sample size and the big difference between provinces, a lower prevalence in Preah Vihear cannot be ruled out.
There were no significant associations between reproductive disorders and serologic status of households. However, abortions, stillbirths, mummified fetuses, and weak or shaking piglets had not been observed in any of the seronegative households during the past year. Reproductive disorders had not been observed in most of the seropositive households either, but the total absence in all seronegative households is still an interesting finding. However, as all pigs over 6 months of age were seropositive, the presence of seronegative pigs in seronegative households is not an indication of the serological status in adult, sexually mature pigs. Moreover, the fact that pigs are seropositive does not provide any information about when they became infected. Thus, it is possible that pigs do not become infected with JEV until after they reach sexual maturity. If infection occurs before 60 to 70 days of gestation in gilts or sows, then the virus can cause reproductive disorders [8
]. However, given the endemic situation in the country, most pigs probably become infected before reaching sexual maturity, and it is unlikely to find an association between serologic status and reproductive performance [15
]. It is also possible that the study was underpowered to detect small differences with this high prevalence. The households included in the present study were asked if they had observed any reproductive disorders during the past year, and their responses may have been influenced by recall bias given the long time period. The ongoing outbreak of African Swine Fever in Cambodia and many other Asian countries in 2019 may also have affected farmer responses, since the provincial official veterinarian was always present during the interviews, and farmers might have been cautious to mention if their pigs had showed any signs of disease.
Almost all respondents had heard of mosquito-borne diseases, and over two-thirds had heard of JE. However, only one person knew that people become infected with JEV through mosquito bites. This indicates that there is a scope to improve knowledge about the transmission of JEV, as well as other vector-borne diseases in these rural provinces. Although the knowledge of JEV transmission appeared to be low, all households used some type of mosquito protection. Almost all households used bed nets and more than half used insect repellent, although the frequency of application is unknown. Because almost all respondents were aware of that mosquito bites can cause disease, they were likely motivated to protect themselves, although practical inconvenience or high costs might influence the frequency of application.
Only one household used mosquito nets to protect their pigs, while ten used insect repellents and three used smoke. Households that did not protect their pigs were not asked why, although the lack of knowledge about that pigs can become infected with JEV through mosquito bites, as was found in this study, might be an explanation. Because JEV infection is usually subclinical in pigs [8
], it is understandable if pig farmers are not aware of the infection risk. The lack of knowledge regarding the JEV infection route to both pigs and humans indicates that there is a need for more education on JEV and strategies to suppress virus circulation and avoid infection. This may be achieved through information campaigns and/or elementary school education. The effects of mosquito protection for pigs would need to be evaluated in the ecological setting it is intended for, since there are many factors that influence the intensity of JEV transmission, such as precipitation [16
], mosquito breeding grounds [17
] and amplifying birds [2
], and there may be other preventive measures that are more cost-effective. In this study, there was no observed difference in serologic status between the households that protected their pigs and the other households regarding serologic status.
Households that had heard of JE (which represented two-thirds of all households) were asked whether they were vaccinated against JE or not. The majority of these had vaccinated children, and more than 90% of the vaccinations were funded by the government. There was no significant difference between the provinces regarding the prevalence of households with vaccinated children, and the vaccination coverage appeared to be high for children in all provinces. The incorporation of JEV into the national childhood immunization program in 2016 [9
] has likely been successful in reaching children of rural families in these areas.
One proposed strategy to reduce the transmission of JEV is to vaccinate pigs. However, there are many reasons for why this strategy would not be applicable to the Cambodian context. First, which is a general liability, the turnover rate in pig production is high, and vaccination would need to be performed frequently and would require high costs in terms of both management and investment [8
]. Second, because most adult sows in Cambodia are likely to be immune to JEV, as observed in this study and by Duong et al. [7
], piglets acquire passive immunity through colostrum intake. The presence of maternal antibodies could interfere with vaccination, making it ineffective [8
]. Also, there is a relatively short window between the waning of maternal antibodies at 2 to 3 months of age and the development of active immunity through natural infection with JEV, as most pigs in Cambodia start producing antibodies in response to infection before 6 months of age [10
]. Third, vaccination of pigs does not eliminate the risk of human infection, since JEV can also be transmitted from birds [2
]. Many households in Cambodia keep chickens or ducks [11
], which have potential to infect mosquito vectors with JEV [20
]. Moreover, wild aquatic birds such as herons and egrets—the natural maintenance reservoir for JEV [2
]—are present in Cambodia.
Another proposed control strategy is to reduce the number of mosquito vectors by reducing the number of larvae in rice fields. Keiser et al. [17
] compared chemical and biological intervention strategies and concluded that one of the most sustainable and worthwhile strategies is alternate wet and dry irrigation (AWDI). However, this strategy can only be performed in settings where irrigation water can be managed. The irrigation infrastructure in Cambodia has grown gradually since the 1990s but remains significantly underdeveloped [21
]. In 2010, approximately 24% of the rice land was estimated to be irrigated. Most rice fields are rain fed, with or without supplementary irrigation. Drainage systems are poorly developed, and floods are frequent during the rainy season. Because of limited irrigation and drainage possibilities, AWDI may not be very feasible in Cambodia. If the water supply and drainage of rice fields could be controlled, then mosquito vectors could potentially be reduced.
A possible source of false test results in this study is the unknown specificity of the ELISA test. It is designed to detect antibodies against the envelope protein of West Nile virus, but cross-reacts with antibodies against other flaviviruses, which exhibit the same type of protein. The test has previously been successfully used to detect JEV antibodies in sera from pigs in Vietnam [22
]. In that study, 105 of 108 ELISA-positive samples were confirmed to be truly positive for JEV with virus neutralization test, which is a highly specific serological assay used to distinguish between antibodies against different flaviviruses. Except for JEV, the only flaviviruses present in Cambodia that can infect and induce antibody production in pigs are the Dengue and Zika viruses [8
]. However, infection with these viruses has only been observed in experimentally infected pigs [8
], and the circulation of Zika virus is low in Cambodia [23
]. Thus, it is unlikely that the ELISA would have cross-reacted with antibodies against any other flaviviruses than JEV. Because there are no known sensitivity or specificity values, the predictive value of a positive test (PVPT) could not be calculated. However, since the PVPT correlates with prevalence and the apparent prevalence was very high, most of the positive rest results are likely to be true positive.