Introduction Typhoid fever is one of the leading causes of morbidity and mortality across the world [1].Typhoid is caused by a bacterium of the genus Salmonella. Salmonella infection in humans can be categorised into two broad types, that caused by low virulence serotypes of Salmonella enterica which cause food poisoning, and that caused by the high virulence serotypes Salmonella enterica typhi (S. typhi), that causes typhoid,and a group of serovars, known as S Paratyphi A, B and C, which cause Paratyphoid [2]. Humans are the only host of this latter group of pathogens. S. Typhi is a highly adapted human-specific pathogen [3], and the illness caused by these bacteria is a serious public health concern, particularly in developing countries [4]. A recent estimate found that 22 million new typhoid cases occur each year in the world with some 200,000 of these resulting in death [5], indicating that the global burden of this disease has increased steadily from a previous estimate of 16 million [6] however, case-fatality rates have decreased markedly [5]. The highest number of cases (>100 per 100,000 persons/year) and consequent fatalities are believed to occur in South Central and Southeast Asia [1]. Generally, typhoid is endemic in impoverished areas of the world where the provision of safe drinking water and sanitation is inadequate and the quality of life is poor. Although contaminated food [7]–[11] and water [9], [12]–[15] have been identified as the major risk factors for typhoid prevalence, a range of other factors have been reported in different endemic settings such as poor sanitation [16], close contact with typhoid cases or carriers [17], level of education, larger household size, closer location to water bodies [17], [18], flooding [19], personal hygiene [12], poor life style [20], and travelling to endemic areas [21]. In addition, climatic variables such as, rainfall, vapour pressure and temperature have an important effect on the transmission and distribution of typhoid infections in human populations [12], [22]. On the Indian subcontinent, Pakistan has the highest incidence (451.7 per 100,000 persons/year) of typhoid fever followed by India (214.2 per 100,000 persons/year) [23]. The mean age of those infected with typhoid is 15.5 years in India and 7.0 years in Pakistan. Bangladesh, located in South Asia, has a population that is mostly impoverished; thus, it is probable that typhoid incidence will be high. A population-based study reported that children and young adults had the highest age-specific rates of all enteric infection [24]. Typhoid disproportionately affects children, with the highest incidence rate being observed in children 0.05). Since previous population-based studies have mainly been conducted in urban locations in South Asia, some bias may have occurred, implying that the disease is largely confined to urban areas [16]. Urban areas in South Asia are rapidly growing compared to other parts of the world, and often characterized by inadequate provision of safe water and sanitation, hence the burden of this disease seems to be higher in urban places than its rural counterpart. This may also be introduced due to the fact that urban populations can, and do, seek medical help more often than rural populations, which could affect the number of cases that are recorded in these two locations. A distinct seasonal variation was found with almost half (45%) of the reported cases found to have occurred in the monsoon. This is contrasting to the finding of a prospective community-based study [4] but supports other results [18], [25]. Monthly distribution revealed that August to September had the highest cases while December to February showed relatively low cases. Environmental factors such as rainfall may have substantial influence to the occurrence of typhoid [12], [22] with increasing transmission of water borne pathogens during wet periods [62]. Because of heavy rainfall during the monsoon in South Asia, a peak of disease occurrence during July to October is not surprising as chances of surface water contamination is also high [18], particularly in densely populated areas like DMA. Although the case-fatality rate was relatively low during the study period, improvements to the water and sanitation infrastructures could reduce the risk of infection and fatality, hence reducing the disease burden. The spatial association between water bodies and the incidences of typhoid showed significant relationships. This finding suggests that people living closer to water bodies may have elevated risk of infection. This relationship has not been reported earlier, however, case-control studies in India [18] and Vietnam [17] revealed that residents close to water bodies, and who use surface water for drinking tend to have more typhoid risk. A similar observation was also reported for diarrhoea incidence [56]. The areas supporting our hypothesis of inverse relationship between typhoid occurrence and distance to waterbodies might explained by the fact that there is a higher faecal contamination load in rivers [63]. As surface and groundwater water quality get severely degraded due to increasing anthropogenic activities in DMA, this may have significant impact on the transmission and distribution of typhoid. In addition, low income inhabitants in the study area frequently use surface water for cooking, bathing etc. As a result, contamination of these water bodies may have substantial impact on the disease dynamics in the communities. As S.Typhi bacteria can survive in water for days [64], contaminated surface water such as sewage, freshwater and groundwater would act as etiological agents of typhoid [65]. Inspection of the t-value and parameter estimate maps of typhoid infection and distance to water bodies further corroborates the spatial association of these two variables (Figure 4a&b). We found that mostly communities living close to the rivers Buriganga, Turag, and Balu had an elevated risk of typhoid infection compared with communities in other locations. These three rivers have been found to have extreme pollution loads throughout the year, measured in terms of coliform counts and other physio-chemical parameters [66]–[71], hence the assumption of an increase in the disease burden is warranted. Also, risk factor investigations for typhoid have shown that all source of drinking water, including pipe water, tube wells and surface water are perpetually highly contaminated in the study area [8], [25], and therefore increases the chance of water borne infection among people living in that area. The transmission dynamics of typhoid in relation to water quality, therefore remains a very promising area for further investigation. It is important to note that we have used major water bodies to regress against dependent variable which is in coarse resolution. Using a finer resolution water bodies map may provide further detail as people in the study area depend on small waterbodies such as ponds for their domestic and bathing purposes. The global autocorrelation analysis using the Moran's I demonstrated that the spatial distribution of typhoid was clustered for all years (2005–2009) (Table 4), signifying that the disease is not uniformly or randomly distributed over DMA. This information can guide public health professionals in their search for possible interventions. An interesting distribution pattern was observed in the typhoid incidence map (Figure 5), namely, that typhoid infections reported in the mahalla's were often located close to water bodies such as river network, lakes and ponds. One may conclude from this distribution that people closer to water bodies are more likely to be affected by typhoid fever because of huge pollution loads of surface water bodies, and the spatial regression analysis carried out in this study also supports this finding. The LISA map (Figure 6) indicated that significant spatial clustering of census tracts with regard to typhoid endemicity in DMA. Our result suggests that empirical Bayesian-smoothed typhoid rates were spatially dependent for the years 2005–2009. This study identified 3 multi-centred and five single-centred clusters. These spatial cluster maps can be used as an initial step in the development of disease risk prediction map since neighbouring spatial units tend to share similar environments and are often connected by the spread of communicable disease [72]. Typhoid incidences in the study area have been reported to be correlated with socio-economic, environmental and sanitation factors [8], [25]. Therefore, an integrated study considering socio-economic, environmental and other relevant factors would greatly benefit public health community in deeper understanding of the dynamics and transmission of typhoid risk in DMA or elsewhere. Since rapid urbanization and food habits tend to alter the prevalence of typhoid [2], this study underscore the necessity of the implementation of sustained safe water and sanitation associated with rapid urban expansion in DMA. The temporal analysis of the relationship between typhoid cases and hydro-meteorological factors revealed that the number of reported cases was amplified by increases in temperature, rainfall and river levels (Figure 8). While the seasonal distribution that we found in this study was similar to the distributions reported in earlier studies, one study by Lin et. al. [73] reported a contradictory finding for the association between river levels and typhoid incidences in Vietnam. Vapour pressure, temperature and precipitation have elsewhere been found to have significant associations with enteric diseases [12], [22], [74], which substantiates the result of this study. Our statistical model further stipulates that increase in rainfall and temperature lead to the higher typhoid cases in the study area. Since flooding is pervasive during the monsoon in DMA, increases in rainfall during the rainy season pollute the surface water which may have caused higher incidences of typhoid [75]. In addition, tube wells that are also flooded during the monsoon may be another source of infection due to contamination with faecal organisms [76], [77]. This study suggests that safe water supply remains a key issue in developing strategies for controlling typhoid infection in DMA. 10.1371/journal.pntd.0001998.g008 Figure 8 Spatial clusters (hotspots) of typhoid in DMA during 2005–2009. See Figure S3 for high resolution version. Our study is not without limitations. First of all, the disease data that were acquired from hospitals may have underestimated or overestimated the typhoid records. Because the data were historical records and documented from the record room of each hospital, we had no valid method to ascertain repeated hospitalizations of an individual patient. In addition, hospital-based surveillance may underestimate actual infected population because only people in a severely weakened state tend to get admitted for treatment. Secondly, we only consider 11 major health service providers, the majority of which were public hospitals. The study could be improved by including data from private clinics where most of the affluent members of the population seek health services. Thirdly, we also could not separate cases into typhoid and paratyphoid groups. Isolation of these two types would allow us to estimate the disease dynamics and identify the most prevalent disease in DMA. Fourthly, the use of two or more methods to identify clustering is suggested as different analytical methods may recognize different underlying spatial patterns in the same dataset [78]. In this study, only one clustering method was used. Therefore, a future study should employ other spatial analytical technique to validate the result. Despite the limitations above, the major strength of this study is the derivation of the first fine-scale regional map of the spatial distribution of typhoid and its epidemiology in Bangladesh. Conclusions Using multi-temporal typhoid data and spatial analytical methods, this study explored the epidemiology and spatial patterns of typhoid infection in DMA of Bangladesh. Epidemiological characteristics showed that the disease disproportionately affects the male population and certain age groups. We did not notice any significance on the occurrence of typhoid between urban and rural areas. Seasonal analysis showed that the risk of typhoid infection is high during monsoon. Temporal distribution suggested that the disease is increasing with time which underscores the importance of prevention. Cluster maps that have developed in this study would help planners to assess spatial risk for typhoid incidences in DMA or elsewhere, and to derive appropriate health policy. The findings of this study could contribute to the understanding of spatial variability of the burden of disease at the community level and may be useful in making decisions about vaccination. Local public health officials can use the information to identify the areas having higher disease occurrences and prepare for targeted interventions. For example, children can be targeted for immunization as other measures such as improvement of water supply and sanitation require what would be a huge investment for a resource-poor country. In this study, spatial and environmental factors were used to identify possible causal factor for typhoid incidences. In addition to these factors, other variables such as population density can be used to examine the factors that are most responsible at the local level. To prevent the spread of typhoid, awareness program should be initiated for the people who rely on nearby water bodies for drinking and domestic purposes. Because of recurrent flooding in the study area in the monsoon season, infected debris could have been another source of disease transmission that would increase the risk of acquiring the disease. Therefore, typhoid prevention can be addressed through both short- and long-term measures. As a short-term measure, people should be informed through a targeted campaign program of the dangers of using unboiled surface water during the monsoon. Medium-term measures could include the improvement of drainage facilities to minimize runoff of human waste into water bodies and long-term measures may be the development of a strong surveillance system to identify both cases and carriers. Finally, an efficient vaccination program can be undertaken for age-specific population at risk, though vaccines are not an alternative to safe water and good hygiene practices [79]. Supporting Information Figure S1 Spatial regression between typhoid incidence (per 100,000 people) and distance to water bodies. A) Shows spatial distribution of the t-value, B) shows the parameter estimates. High resolution version of Figure 5. (TIF) Click here for additional data file. Figure S2 Spatial variation in the occurrence of typhoid infection. This shows the raw annual incidence rate(A) and EB-smoothed incidence rates (B) from 2005 to 2009 in census districts in DMA: High resolution version of Figure 6. (TIF) Click here for additional data file. Figure S3 Spatial clusters (hotspots) of typhoid in DMA during 2005–2009. (TIF) Click here for additional data file. Figure S4 Sensitivity analysis. Percent change (and 95% CIs) in the number of typhoid cases for (A) river level (per 0.1 m increase above the threshold), (B) rainfall (per 10 mm increase below threshold) and (C) temperature (per 1°C increase) with each number of harmonics and indicator variable of month (M). Presented results are from final models adjusted for seasonal variation (8 harmonics), inter-annual variations, and public holidays. (TIF) Click here for additional data file. Checklist S1 STROBE checklist (DOC) Click here for additional data file.
