Portal de Periódicos da CAPES

Contextual risk factors for regional distribution of Japanese encephalitis in the People’s Republic of China

2010; Wiley; Volume: 15; Issue: 8 Linguagem: Inglês

10.1111/j.1365-3156.2010.02563.x

1365-3156

Mingqin Cao, Zijian Feng, Junhui Zhang, Jiaqi Ma, Xiaosong Li,

Vector-borne infectious diseases

Objective To investigate the contextual risk factors for the regional distribution of Japanese encephalitis (JE) in the People's Republic of China to enhance the prevention and control of JE. Methods A multilevel Poisson regression model was used to analyse the association between the epidemic of JE and its contextual risk factors with an emphasis on the proportion of rice-planting area, the extent of pig rearing and the proportion of rural population. Results The highest risk of JE was observed in the southwestern and the central areas of P.R. China, characterized by high proportions of rice-planting area, rural population and extent of pig rearing. These contextual determinants seem to govern the risk of JE. Conclusion In P.R. China, an effective surveillance system should be established in the high-risk regions of JE; immunization coverage for the prevention of JE should be expanded in rural areas, and mosquito-control efforts should be made to enhance the prevention and control of JE. Facteurs de risque contextuels pour la distribution régionale de l'encéphalite japonaise dans la République Populaire de Chine Objectif: Investiguer les facteurs de risque contextuels pour la distribution régionale de l'encéphalite japonaise (EJ) en République Populaire de Chine, afin d'améliorer la prévention et la lutte contre l'EJ. Méthodes: Un modèle de régression de Poisson à multi niveaux a été utilisé pour analyser les associations entre l'épidémie de l'EJ et les facteurs de risque contextuels avec un accent sur la proportion des zones de culture de riz, l'étendue de l'élevage porcin et la proportion de la population rurale. Résultats: Le plus grand risque pour l'EJ a été observé dans le sud-ouest et les régions centrales de la République Populaire de Chine, caractérisées par de grandes étendues de superficie de culture de riz, une population rurale et l'étendue de l'élevage porcin. Ces déterminants contextuels semblent gouverner le risque pour l'EJ. Conclusion: En République Populaire de Chine, un système de surveillance efficace devrait être établi dans les régions à risque élevé pour l'EJ, la couverture vaccinale pour la prévention de l'EJ devrait être élargie dans les zones rurales et les efforts de lutte contre les moustiques devraient être effectués pour améliorer la prévention et la lutte contre l'EJ. Factores de riesgo contextuales para la distribución regional de la encefalitis japonesa en la República Popular de China Objetivo: Investigar los factores de riesgo contextuales para la distribución regional de la encefalitis Japonesa (EJ) en la República Popular de China, con el fin de mejorar la prevención y el control de la EJ. Métodos: Se utilizó un modelo multinivel de regresión de Poisson para analizar la asociación entre la epidemia de EJ y los factores de riesgo contextuales con un énfasis en la proporción de áreas para el cultivo de arroz, la extensión de cría de cerdos y la proporción de población rural. Resultados: El mayor riesgo de EJ se observó en el centro y sudoeste de la R.P. de China, con una alta proporción de áreas de plantaciones de arroz, población rural y amplias extensiones con cría de cerdos. Estos determinantes contextuales parecen ser los que gobiernan el riesgo de EJ. Conclusión: En la R.P. de China debería establecerse un sistema de vigilancia efectivo en las áreas de alto riesgo para EJ, expandir la cobertura de inmunización preventiva en las áreas rurales, y asegurar que los esfuerzos de control de mosquitos refuercen la prevención y el control de la EJ. Japanese encephalitis (JE) is a vector-borne viral disease with a high mortality and a high percentage of neuropsychiatric sequelae. The worldwide incidence of JE is 30 000–50 000 cases per year, and the estimated mortality is 10 000 per year (Tsai 2000). In the Asian and Pacific areas, JE is an endemic disease, mainly affecting children and young adults (Solomon 2006). JE virus is spread by marsh birds and intensified by pigs, mainly transmitted via the bite of infected Culex mosquito. Humans are dead-end hosts (Diagana et al. 2007). Many of the ecological, environmental, climatic and human behavioural factors are involved in the JE virus spread (Solomon 2006). Contextual determinants of JE include irrigated rice farming, pig rearing and the rural population. A primary vector of the JE virus is the mosquito Culex tritaeniorhynchus, which is often found in irrigated rice fields. Previous studies reviewed the effects of irrigated rice farming on the spread of JE on a global scale (Keiser et al. 2005). Large numbers of JE cases or JE outbreaks have been found in countries with extensive irrigated rice agriculture, such as Nepal, Thailand and Vietnam. Pig rearing is also a risk factor for JE, because an infected pig has serious and prolonged viraemia. Some studies also provided some data about rice irrigation and pork production in the JE-afflicted countries, which suggest that there is an association of rice planting and pig rearing with JE transmission (Erlanger et al. 2009). However, the literature is mainly descriptive, and hence, it is difficult to draw inference on the effects of rice planting and pig rearing on the incidence of JE. Most Chinese citizens live in the JE-prone areas, so a relatively high epidemic rate of JE has been found in the People's Republic of China (Solomon 2006). Vaccination against JE in children has been available since the 1980s, and, as a consequence, the JE incidence markedly decreased in the past decades. Unfortunately, each year, there are still 5000–10 000 JE cases reported. Endemicity of JE was reported in all provinces and regions of P.R. China, except Xinjiang, Qinghai and Xizang (Wang et al. 2009). However, the related studies have been rather limited, so that the regional distribution of JE and transmission patterns in P.R. China are still not fully understood (Luo et al. 1995; Wang et al. 2004; Li et al. 2006; Liu et al. 2006; Bi et al. 2007). In P.R. China, in 2005, the estimated total rice-planting area was estimated at 29.0 million ha, and approximately 22.1 million people lived in close proximity to irrigated rice areas and hence were potentially at risk of JE. There was an 87% increase in the pork production from 1990 to 2005 (Erlanger et al. 2009). We asked the question whether these contextual determinants can explain the distribution of JE across P.R. China. Because of the differences in the population at risk, the comparison of JE incidence rates cannot be directly made among the 31 provinces and regions of P.R. China. The standardized morbidity ratio (SMR), the ratio of the observed cases to the expected cases of JE within a geographic area, is a measure that puts the morbidity rates obtained from different regions on the same scale for a proper comparison. SMR is a maximum likelihood estimator (MLE) of the region-specific relative risk, which can be used for statistical inference (Lawson et al. 2003). The purpose of this study was to describe the regional variation of JE risk in the 31 provinces and regions across P.R. China. We compared SMR between the provinces and regions in 2004–2007 and assessed the relationship between the regional variation and three key risk factors, namely rice planting, pig rearing and proportion of rural population; all measured on the provincial level. Such knowledge can contribute to the design of intervention policies and healthcare resource allocation for prevention and control of JE. To our knowledge, we present the first study shedding light on the aforementioned risk factors for JE in P.R. China. The JE cases in the 31 provinces and regions of P.R. China were obtained from 2004 to 2007, facilitated by an Internet-based disease-reporting system (China Information System for Disease Control and Prevention; CISDCP), which was put in place in 2004 and was more sensitive and efficient than the previous case-reporting system (Wang et al. 2008). Data of the age-specific population and three contextual determinants (proportion of rice-planting area, pig-rearing number and proportion of rural population) were collected from National Statistics Almanacs for the years 2004–2007, which is published annually by the Nation Bureau of Statistics of China (http://www.stats.gov.cn/tjsj/ndsj). The proportion of rice-planting area is a ratio of the rice-planting areas to the total crop-planting areas. The extent of pig rearing is represented by the total number of the pigs on market in 1 year. The proportion of rural population is a ratio of the rural population to the total population of P.R. China in a given province or region. Standard morbidity is defined as the total JE incidence in P.R. China from 2004 to 2007, stratified by gender and age. The expected cases are the sum of the baseline group population multiplied by the corresponding standard group morbidity within a province or region. SMR is a ratio of the observed cases to the expected cases within the provinces or regions. The relative risk level of the JE incidence in those provinces or regions (SMR > 1) is higher than the average level in P.R. China and vice versa. Our multilevel model was fitted for an analysis on the JE data, because the repeatedly observed number of the JE cases is nested within the provinces or regions. Assuming the observed number of the JE cases within an area to follow a Poisson distribution because of a very low incidence (per 100 000), a two-level multilevel Poisson regression model was built to analyse the relationship between the JE development and the contextual risk factors. The low-level units are the repeatedly observed data within the 31 provinces and regions, and the high-level units are the data of the corresponding provinces and regions. Of note, the proportion of rice-planting area, the extent of pig rearing and the proportion of rural population were abbreviated as "PRPA", "EPR" and "PRP", respectively. was called an offset in the model. β0, a constant, was a logarithm value of the average relative risk when all of the explanatory factors were the baseline values (coded zero). β1, β2 and β3 were regression coefficients. β1 represented a logarithm value of the relative risk ratio when the proportion of rice-planting area changed by one measure unit, while the other explanatory factors (extent of pig rearing and proportion of rural population) were fixed. If β1 was greater than zero, was >1 (i.e. relative risk ratio >1), the explanatory variable (proportion of rice-planting area) was a risk factor that increased the risk of the JE development, and vice versa. The contents of β2 and β3 were similar to the content of β1. , a random effect of the jth province, was a normal distribution with a mean of 0 and a variance of . If was considered statistically significant, which meant a different effect in the different provinces, the risk of the JE development was different between provinces or regions after three explanatory variables were adjusted. β0, β1, β2 and β3 were considered as fixed effects, and was called a random effect in the model. For the model stability and the comparison of the different measures, each explanatory variable was divided into quartiles. The two-level multilevel Poisson model was fitted in MLwiN (Rasbash et al. 2005), using Markov chain Monte Carlo (MCMC) methods for the Bayesian estimates (Browne 2003) rather than the traditional maximum likelihood methods to obtain more accurate values of the random effect (Goldstein 2003). The relative risk ratios were calculated by exponentiating the fixed coefficients. The 95% confidence intervals (95% CI) were taken from the quantiles of the MCMC distribution. A total of 22 334 JE cases were reported in P.R. China from 2004 to 2007. There were 13 707 male and 8627 female cases. Under 5-year-old children accounted for 46%, 5- to 15-year-old school children for 42% and those >15 years old for 12%. Most JE cases occurred between June and October, with a close relationship between seasons to JE across P.R. China. The JE cases were reported in all the provinces and regions except Qinghai, Xinjiang and Xizang (Table 1). The incidence rate ranged from 0.01 per 100 000 to 4 per 100 000 in the different provinces and regions. High incidence rates of JE (>0.5 per 100 000) were reported in Guizhou, Chongqing, Sichuan, Yunnan, Guangxi and Shaanxi. Those provinces or regions were mainly located in the southwestern or the central areas of P.R. China. The Spearman correlation coefficients (i.e. a non-parametric method of bivariate relationship) between SMR of JE and the proportion of rice-planting area, the extent of pig rearing and the proportion of rural population were 0.34 (P < 0.001), 0.56 (P < 0.001) and 0.40 (P < 0.001), respectively. We found positive correlations between SMR of JE and the three risk factors. Given the nested data structure, the two-level multilevel Poisson regression model was built to show the relationship between JE and three contextual determinants investigated. The results of the two-level multilevel Poisson regression analysis are summarized in Table 2, showing that the three risk factors were all statistically significant. A higher proportion of rice-planting area and a higher proportion of rural population were predictive factors for higher risk of JE, and a large extent of pig rearing predicted a higher risk of JE, but a declining trend was observed with an increasing pig-rearing number. The risks of JE changed during the years 2004–2007. was statistically significant, which meant regional variation of JE risk across P.R. China. The expansion of the JE virus-endemic area depends on irrigated rice farming and pig rearing (Oya & Kurane 2007). A higher proportion of rice-planting area predicted a higher risk of JE in P.R. China. Although the JE vectors are able to breed in many habitats containing water, one of their major preferred larval habitats are rice fields. High densities of JE vector were reported in rice fields after the rainy season when there is plenty of water and temperatures are high, facilitating larval grow in large numbers (Sunish & Reuben 2001; Van den Hurk et al. 2009). A higher proportion of rice-planting area predicted a higher proportion of man-made breeding sites and a greater increase in the vectors and the human-vector contacts (Diagana et al. 2007). Pig rearing is common in the countries prone to JE epidemic (Solomon 2006). Introduction of pig farming has introduced the JE epidemic in some regions of Asia. As expected, a greater number of pigs predicted a higher risk of JE. But why did a decreasing trend occur with an increased number of pigs in P.R. China? First, close contact of pigs with humans occurs in only those pigs that are reared in open or unroofed pigpens, which are close enough to the human dwelling and hence governs the spread of the JE virus. Second, a growing number of pigs are mostly reared in modern pig farms separate from human housing (Nitatpattana et al. 2008). So, the number of pigs plays an interesting role in the converse relationship with the relative risk of JE. Scattered pig-rearing farms are the main patterns in the central and the western P.R. China, while the modern pig-rearing farms are the main patterns in the eastern P.R. China (Chen et al. 2008). Thus, the rising number of pigs reared in the central and the western regions of P.R. China may play an important role in JE transmission. Most JE cases were found in rural areas, especially in the rice-planting and the pig-rearing areas. The proportion of the rural population can grossly reflect the condition of the economic development in the provinces and regions across P.R. China (Zhu et al. 2008). The less-developed region with a greater proportion of the rural population was often short of funds for JE prevention and control. We found that the incidence of JE decreased from 2005 to 2007 but increased in 2006, comparable to the first year of investigation (2004). In the longer term, the incidence of JE in P.R. China showed a gradual declining trend with some fluctuation (Wang et al. 2009). The random effect between the provinces and regions () was statistically significant, when the fixed effects of the three risk factors and the years were adjusted (Table 2), which interpreted other potential factors not involved in the study result on variation of the JE distribution such as vaccination (Liu et al. 2006), climate (Bi et al. 2003, 2007) and financial support. In the north-western regions, there is less rice planting and less pig rearing. Moreover, temperatures are lower, which might explain why fewer JE cases were reported there. The difference of immunization coverage and socio-economic level may contribute to the different distributions of JE (Liu et al. 2006). Some areas characterized by rapid economic development, such as south-eastern coastal provinces, provided free JE vaccines for immunization campaigns. But in the less-developed provinces, such as south-western and central provinces, the relatively low coverage of JE immunization led to a higher risk of JE. Our study has at least two limitations. First, it is only an ecological study on a provincial scale in P.R. China, so it was difficult to interpret the JE risk for an individual near the irrigated rice-fields or the pig-rearing places. Second, only the data of the 4 years were collected, so it was difficult to show the time-series trend of JE. There are three main routes for prevention and control of JE: control of the vector, avoidance of human exposure and immunization by vaccination (Mackenzie et al. 2004). Mosquito control can reduce the JE transmission, but it is difficult to put it in practice. Avoidance of human exposure to an infected mosquito is only a short-term solution. So, immunization is the most important and efficient method for a long-term protection. Vaccination against JE is a key preventive measure for the susceptible population. Vaccination against JE was integrated into National Expanded Program of Immunization in 2007 (http://www.chinanip.org.cn). The Chinese government has given a priority policy and a priority funds for the prevention and control of JE. Regional variation of JE risk was observed across P.R. China, which was quantitatively associated with proportion of rice-planting area, extent of pig rearing and proportion of rural population. An effective surveillance system should be established in high-risk provinces and regions, especially where the rural residents live in close proximity to irrigated rice fields and pig-rearing places. Immunization coverage for JE should be expanded, especially in rural areas, and mosquito control should be continued to further reduce the incidence of JE. We are grateful to Professor Yang Min for valuable advice on the analysis of the data and the discussion as well as the helpful review of our manuscript. We thank two anonymous peer reviewers for carefully revising our manuscript. This work was financially supported by National Natural Science Foundation of China (No. 30571618) and National Special Foundation for Health Research (No. 200802133).