Visceral leishmaniasis (VL) is a parasitic disease that is endemic in more than 62 countries, with an estimated 1 300 000 new patients each year worldwide [1]. VL has been one of important public health problems in the People’s Republic of China. Prior to the initiation of a national control program in 1951, there were about 530 000 VL cases in China in 1951 [2,3,4].
The first confirmed case of VL in China was reported in 1907 [5]. In the following years, cases have been reported in 16 provinces north of the Yangtze River, which were recognized as endemic areas [6]. The disease was especially rampant in eastern China and also prevalent in Shaanxi, Gansu, and Xinjiang in western China; sporadic cases occurred in Beijing, Liaoning, Ningxia, Qinghai, Inner Mongolia, Hubei, Shanxi, and the southern part of Sichuan. According to surveys conducted in 1951 [3, 7, 8], about 530 000 cases were reported and the average incidence in different provinces was 10–50 per 10 000 people. Following great efforts of the national control program, the disease has been largely brought under control in the eastern regions of the country after the 1960s [6, 9, 10]. VL cases decreased from more than 530 000 originally distributed in more than 660 counties to several hundred cases concentrated in less than 30 counties in the new century.
However, transmission was not interrupted in the mountainous and desert regions, where sporadic cases continue to present. Currently, VL is still prevalent or sporadically distributed in six western provinces and autonomous regions, including Xinjiang, Gansu, Sichuan, Shaanxi, Shanxi, and Inner Mongolia. Four species of sandfly (Ph. Chinensis, Ph. Longiductus, Ph. Wui and Ph. Alexandri), which serve as staple vectors for the transmission of VL in China, are distributed in different geographic areas, contributing to the different ecological features and transmission patterns in different areas [3, 11]. The most sensitive population threatened by VL in western China is children. The main parasite species, Leishmania infantum, which mainly affects children, is distributed in some remote areas of the regions, and new cases are reported every year. Among these endemic regions, three provinces/autonomous regions, i.e., Xinjiang, Gansu, and Sichuan, still face serious health problems due to VL transmission, and infants are the main population under threat of infection [7]. In China, there are two major protozoan parasites, i.e., Leishmania donovani and L. infantum, cause VL. They are transmitted by three species of sandfly, i.e., Phlebotomus longiductus, Phlebotomus wui, and Phlebotomus chinensis [3, 11]. Two epidemiological types of VL, i.e., the anthroponotic type and the zoonotic type, have been classified based on the ecosystem and source of infection in China [12].
The anthroponotic type, caused by L. donovani, currently is endemic in the oases of the plains of Kashi prefecture, Xinjiang Uygur Autonomous Region. Most cases infected with L. donovani occur in young people and children. The transmission cycle is from human to human, and no animal host has been found yet. Its vector is a peridomestic species, P. longiductus, which is widely distributed in Xinjiang. During history, the control strategies carried out in endemic areas for the anthroponotic type of VL, which was also widespread in eastern China in the 1950s, emphasized on mass screening for patients, in combination with vector control. A large number of infected patients were found and received medical treatment in the 1950s. At the same time, the density and natural infection rate of sandflies gradually decreased by indoor or outdoor use of insecticides. As a result, the prevalence of the anthroponotic type of VL was effectively controlled by the end of the 1950s in China.
The zoonotic type, caused by L. infantum, involves an animal host as the principal source of infection [7, 13]. This type has been divided into two subtypes, the mountainous subtype and the desert subtype, based on the ecosystem and epidemiological characteristics, i.e., geographical and landscape characteristics, vector (sandfly) species and their ecology, and source of infection [14, 15]. Contrarily to anthroponotic VL, spraying of insecticides had was little effect on vector control because of the exophilic behavior of the sandfly of zoonotic endemic areas. In mountain subtype endemic areas, killing infected dogs and prohibition of raising domestic dogs, combined with spraying non-infected dogs with deltamethrin, markedly interrupted the transmission of mountain subtype VL. The various control strategies for different types of VL made great contributions to the control of VL, and by the end of the 1950s, it was almost eliminated in China.
Unfortunately, there was no effective way to control desert subtype VL due to the unknown source of infection and wild habitats of the vector; spraying of insecticide showed little effect [12]. People who live in houses of poor quality usually lack individual protection and consciousness, increasing the risk of being bitten by sandflies and becoming infected. Moreover, it should be considered that most cases reported after year 2000 were of this type of VL. The most prominent endemic region of desert subtype VL is the oases of the plains in Kashi Prefecture, Xinjiang Uygur Autonomous Region. Three spatio-temporal clusters with respect to the distribution of VL-infected families during 1990–2005 were identified by scan statistical analysis [16, 17]: (i) Boshikelam, Haohan township of Kashi city and Awati town in Shufu county, (ii) reclaimed farmland of Bachu county andand (iii) Yinwusitang town in Shufu county and Yangdaman town in Shule county.
Besides Xinjiang, VL is still endemic in some mountainous areas in north Sichuan Province and south Gansu Province. Infected dogs and patients with L. donovani are major infection sources. Although the disease has been under control since the end of the 1950s, it has gradually reemerged since 1972 [7, 18, 19] in north Sichuan and south Gansu. After that, control strategies, including treatment of patients, health education, dog registration and management, and sandfly control in combination with sanitary policies of cities or towns, were implemented, and the numbers of reported cases in those counties decreased significantly. However, VL cases still occur sporadically.
During our study period (2004–2018), more than half of cases reported in China occurred in the Xinjiang Uygur Autonomous Region, and effective control measures should be taken there. Moreover, there were two outbreaks in Jiashi county of Xinjiang (a desert subtype of the zoonotic type endemic area), during 2008–2009 and 2014–2015, respectively [20,21,22], with the incidence rate more than 20 times as high as the average annual incidence. It is still unclear whether some kinds of animals transmit the pathogen there. To elucidate the reasons behind the outbreaks, researchers have explored several factors, including land cover [23], meteorological factors [20], and socio-economic status [24], suggesting some possible associations between outbreak and air temperature, relative humidity, land use, and socio-economic status. In addition, several studies [6,7,8,9, 11,12,13] on VL epidemiology have been performed in China, including a retrospective review of notified VL cases between 2005 and 2010 based on the passive surveillance data, a description of epidemiological features of VL in China during 2004–2012 [12, 13], and a review of phlebotomine sandflies transmitting VL and their geographical distribution in China [11, 14]. All those studies have shown that cluster VL transmission patterns have been observed in some areas of western China in different periods of time, indicating that intervention strategies should focus their attention on high-risk transmission areas.
The objective of this study is (i) to explore the spatial and temporal characteristics of VL in Sichuan Province, Gansu Province and Xinjiang Uygur Autonomous Region in China from 2004 to 2018 and (ii) to identify the risk areas for VL transmission, which could provide evidence-based information for effective control efforts. We performed an investigation in a spatio-temporal manner using an additive–multiplicative decomposition of the conditional intensity function [25].
