Spatiotemporal epidemiology, environmental correlates, and demography of malaria in Tak Province, Thailand (2012–2015)

There is a goal to eliminate malaria across countries in the Greater Mekong Subregion (Cambodia, Yunnan Province in China, Lao People’s Democratic Republic (PDR), Myanmar, Thailand and Viet Nam) by 2030. In Thailand, malaria remains as a major public health priority with over 13 million people (19% of the total population) currently at risk [1], most of whom are situated along border areas with its neighbouring countries [2]. There is significant geographical heterogeneity in spatial distribution of disease incidence, which exemplifies the “border malaria” type [3], characterized by high transmission along international borders [3‐5].

Thailand has made significant progress towards its goal of eliminating malaria by 2024 [2]. Between 2012 and 2017, Thailand reduced its malaria cases by 67%, with an accelerated decrease of 39% during the period 2016 to 2017 [6] with scale-up of malaria elimination interventions. According to the 2018 World Malaria Report, the country had 11,440 reported confirmed cases in health facilities (which include 1075 cases at community level and 3023 cases from the private sector) and 11 reported deaths due to malaria in 2017. With evidence of emerging artemisinin drug resistance in Plasmodium falciparum malaria in the Thailand–Myanmar and Thailand–Cambodia–Lao PDR border areas [7‐12], including insecticide resistance [13], adding complexity to the elimination agenda in Thailand, it is increasingly important to understand the local epidemiology of malaria to guide targeted elimination efforts. Strengthening malaria surveillance has been highlighted by the World Health Organization as a core intervention to accelerate progress towards elimination [14, 15].

Tak Province, at the Thai–Myanmar border, is one of three high incidence areas in Thailand. It has historically recorded one of the highest incidences of malaria cases across the country [16]. The province reported the highest number of malaria cases nationwide in 2012 and 2013 [17], and ranked second only to Ubon Ratchathani Province the following 2 years [18]. In 2015, Tak accounted for 21% of total reported cases in Thailand [19]. Plasmodium vivax is the predominant species, reflective of the national situation [1]. The availability of detailed climate and forest cover data presents an opportunity to better understand the epidemiology of malaria in the area.

This study analysed routine malaria surveillance data from 2012 to 2015 to describe and identify possible factors driving the spatiotemporal trends of disease incidence.

From 2012 to 2015, Tak Province has experienced a substantial and consistent decline of both P. falciparum and P. vivax malaria with a decline in P. falciparum-to-P. vivax ratio over the same period. Around three quarters of the cases and areas with highest risk (indicated by API) were consistently in clusters of a few sub-districts in two out of the nine districts (Tha Song Yang and Umphang) at the northern and southern extremities of the province by the Thai–Myanmar border. There was some correlation with the percent decrease in forest cover and decline in API at sub-district level. This geographical pattern was similar for both species.

Although climate correlated reasonably well with API, there was no clear change in seasonal climate pattern over the period of the study. This suggests that change in climate was not responsible for the decrease in malaria. The relationship between trends in climate and malaria is not straightforward in that there is an optimal ecological niche of climate within which mosquito breeding and malaria transmission are more efficient [33]. In Thailand, there are seven Anopheles species known to transmit malaria, which have different environmental optima and geographic distributions and it is not clearly known to what degree they contribute to transmission in Tak Province [34]. Because the correlation was weak, the change in forest cover is unlikely to have been the main driver overall for the reduction in malaria. It is possible that small changes in forest cover may have a greater contribution in small foci where forest transmission is occurring but it was not possible to study this, as suitable detailed data was not available. More likely, it is a result of the Thai MOPH’s ongoing elimination efforts in the area with high coverage of long-lasting insecticidal nets, good access to diagnosis and treatment via malaria posts and health promoting hospitals, active case detection and village health volunteers existing with vector control and facilitating referral of symptomatic individuals [2]. However, it was not possible to assess this as suitable data on interventions were not available. This decline in cases mirrors that of what is seen elsewhere in Thailand, and in subsequent years, with overall strong progress towards the national goal of elimination [2]. Also unique to Tak Province is the presence of Shoklo Malaria Research Unit (SMRU) in Mae Sot District, which has been providing early diagnosis and treatment for malaria since 1986 [35]. SMRU operates at both sides of the Thai–Myanmar border, and reported 17,446 malaria cases between 2012 and 2015 (Additional file 1: Table S1 for annual reported cases by species) at sites in Tak Province with an 85.64% decline in reported malaria cases from 2012 to 2015. There was a similar decline in cases reported by the Thai MOPH at 75.78% in the same period. SMRU sites in Kayin State, Myanmar also saw a decline in malaria incidence, as brought about by recent gains in elimination efforts in Myanmar [36]. Due to migration of people between the two areas, the decline in reported cases on the Thai side may also be partly due to the decline on the other side of the border. Lack of functioning malaria posts in Tha Song Yang (2014) and Umphang (2015) Districts (source: SMRU unpublished reports) may have contributed to increased transmission in the areas identified as hotspots in this analysis.

A strength of this study is that it combined detailed data from all sources reporting to the MOPH in Thailand together [37, 38], with population data from the same area. This allowed examination of which groups of people had the highest rates of malaria in each major species. Similar to other studies in the region [39‐41] and elsewhere [42], around two-thirds were male, particularly in young adults, possibly due to increased occupational risk and forest exposure in this group. More than half the cases were of Myanmar nationality. Data were not available to indicate whether they were short-term or long-term migrants, but this study mapped cases by place of residence where this information was available (i.e. which sub-district a patient confirmed to have malaria was reported to be from). When cases were mapped by nationality, those of Myanmar nationality were clustered in a few sub-districts in the centre of the province along the country border.

The study had several important limitations. Although the data was of high quality, there was some missing information, e.g. 2.40% of cases had unknown malaria species, 13.88% of patients had no age recorded, 4.08% had no nationality recorded, and 1.30% had missing sub-district codes (Additional file 1: Table S2 for yearly percentages, and 4-year average). The completeness of data for each field did not consistently increase over the period of the study. The study utilized data obtained mostly from passive case detection with a few from active case detection (Additional file 1: Table S3 for an annual breakdown of confirmed malaria cases by detection method), but it was not clear which applied to each case. There are other sources of surveillance data that are not fully captured in the MOPH database, including those from the military and non-governmental organizations [43]. PCR is not used for routine surveillance in Thailand and there was no available data from the government on asymptomatic infections, which could render asymptomatic cases to be undetected, as some research studies have revealed in both high transmission [44, 45] and low transmission [46, 47] settings at the Thai–Myanmar border. Data were not included from SMRU, and these additional case data could be substantial enough to affect the spatial and temporal patterns of disease found in this study. However, it is not clear to what degree cases recorded by SMRU overlap with the dataset used in this study due to, e.g. presentation of cases to more than one organization within the same malaria episode, and there is no system to effectively integrate the two databases accounting for this potential overlap. Perhaps more importantly, many of these cases are short-term migrants who are likely to have been infected in Myanmar, there being a porous border between the two countries in this area. Including cases from SMRU should thus have had little impact on the results as the analysis was based on place of residence being in Tak Province. Other data that were not available for the analysis were numbers of people tested, or detailed data on interventions including bed net distribution or treatment, which could have been used to confirm the downward trend in cases being due to successful elimination efforts. Finally, the data used to map forest cover had limitations in that there was no information on forest gain from 2013 to 2015 (stable forest gain was assumed during this period), only forest loss [26].

It would be informative to explore further about possible causes of the spatial pattern of disease. For example, delays in seeking treatment [48], importation or spread through human movement [49, 50], travel into the forest [51], locations of particular occupational risk groups, or the role of undiagnosed cases [52] and parasite genetics [53]. This would require additional surveys to collect this information. Another possible future analysis would be to use satellite remote sensing data as a higher resolution source of information on climate to examine the variation across space. More detailed mapping of the distribution of cases to assess the degree of clustering would also be helpful but this would require village level data on place of residence, which was not available for this study. This is particularly important as further progress is made towards elimination requiring targeting of these clusters to reach zero. It would also be informative to include data from neighbouring areas of Myanmar [54] to assess risk at either side of the border. Given the cross-border nature of the study setting [55], and how this could affect malaria transmission, there is an opportunity to advocate for a regional data system that would allow for a combined analysis of trends for both Tak Province in Thailand and Kayin State in Myanmar.

There has been a large decline in reported clinical malaria from 2012 to 2015 in Tak Province. Climate and forest correlated with malaria incidence rates but did not account for this decrease. Ongoing elimination interventions on one or both sides of the border are more likely to have been the cause but it was not possible to assess this due to a lack of suitable data. Two main hot spot areas were identified along the Thai–Myanmar border that should be studied in more detail so they can be targeted for elimination activities.