Trends and climatic drivers of leptospirosis in Malaysia, 2012–2022: spatiotemporal analysis

Introduction

Background

Leptospirosis is a globally distributed zoonotic infection caused by pathogenic spirochetes of the genus Leptospira and is recognised as one of the most widespread bacterial zoonoses worldwide (1-3). The disease affects both developed and developing countries, but it imposes a disproportionate burden on tropical and subtropical regions where environmental conditions favour pathogen survival (1). Global estimates suggest that leptospirosis contributes substantially to morbidity and mortality, with severe disease manifestations leading to significant case fatality rates in certain settings (1,4). Despite this burden, leptospirosis remains under-recognised and frequently under-reported, partly because its early clinical presentation is non-specific and resembles other common febrile illnesses, such as dengue, influenza, malaria, and viral hepatitis (1,4). This diagnostic overlap complicates timely detection and may contribute to misclassification within surveillance systems (1,4).

Transmission occurs primarily through direct or indirect contact with water, soil, or environmental surfaces contaminated with the urine of infected animals, particularly rodents, which serve as the principal reservoir hosts (2-4). Leptospira organisms are capable of surviving for prolonged periods in moist environments, especially under warm and humid conditions (2,5). Human infection typically occurs through abraded skin or mucosal membranes following exposure to contaminated environments (2). Occupational exposure in agriculture, livestock handling, sanitation work, and construction has been consistently identified as a risk factor (4). Recreational exposure during freshwater activities and contact with floodwaters have also been implicated in outbreaks (4,6). These transmission pathways underscore the strong ecological and environmental dimensions of leptospirosis epidemiology (2,6).

Climatic factors play a central role in shaping leptospirosis transmission dynamics. Environmental survival of Leptospira is enhanced in conditions of high rainfall and humidity, which maintain soil moisture and facilitate bacterial persistence in surface water (2,6). Heavy rainfall events may increase runoff and flooding, thereby dispersing contaminated soil and rodent urine into residential and agricultural areas (6,7). Flooding can displace rodent populations, alter human-environment interactions, and increase contact between susceptible individuals and contaminated water sources (6). Numerous studies have demonstrated temporal associations between rainfall accumulation and increased leptospirosis incidence in tropical regions (6,7). However, the magnitude and timing of these associations vary across geographic contexts, reflecting differences in infrastructure, drainage systems, land use patterns, and population behaviours (6).

Temperature also influences leptospirosis epidemiology, although findings have been heterogeneous (5,6). Warm temperatures generally favour bacterial survival and rodent reproduction, potentially increasing environmental contamination (5). Nevertheless, extreme heat may reduce bacterial persistence through desiccation or modify human behavioural patterns, resulting in varying epidemiological effects (6,7). Relative humidity further contributes to environmental persistence by maintaining soil moisture and reducing desiccation (2,5). These interacting climatic variables collectively shape the ecological niche of Leptospira, yet their relative contributions may differ across countries and ecological zones (6). Consequently, country-specific evaluations are required to clarify the climatic drivers of transmission (6,7).

In Southeast Asia, leptospirosis remains a significant public health concern, with recurrent outbreaks reported in association with monsoon seasons and flooding events (6,7). Rapid urbanisation, informal settlements with inadequate sanitation, and agricultural expansion have been identified as contextual factors that may amplify risks (6). Environmental disruption and land use transformation can alter rodent habitats and increase human exposure to contaminated environments (6). In addition, climate variability and extreme weather events are projected to intensify in many tropical regions, potentially influencing future disease patterns (6). These considerations highlight the importance of understanding how local climatic and environmental conditions interact to shape leptospirosis transmission.

Malaysia’s equatorial climate provides favourable conditions for the persistence and spread of Leptospira (6,8). The country experiences consistently high temperatures, substantial annual rainfall, and elevated humidity throughout the year (6). The Northeast monsoon, typically occurring between November and March, brings heavy rainfall to several states and contributes to periodic flooding (6,8). Such environmental conditions create sustained opportunities for environmental contamination and human exposure (6). National surveillance systems have reported leptospirosis as an endemic disease in Malaysia following its notification under the Prevention and Control of Infectious Diseases Act 1988 (Act 342) (8,9). Annual surveillance reports indicate fluctuating case numbers over time (8). Peaks in incidence have been observed during certain years, particularly following major flooding events (8,10), suggesting possible links with climatic variability and environmental changes (6,10).

The geographic distribution of leptospirosis in Malaysia is heterogeneous (8). States in East Malaysia, particularly Sarawak, have consistently reported a higher incidence compared with many states in Peninsular Malaysia (8). Variations in rainfall patterns, topography, population density, agricultural activities, and infrastructure may contribute to these differences (6,9). Sarawak experiences prolonged wet periods in some years, potentially sustaining environmental contamination (8). In contrast, some regions of Peninsular Malaysia exhibit more seasonal rainfall variability (6,8). These spatial differences point to a comprehensive national analysis to identify persistent high-risk areas and better understand geographic disparities in disease burden.

Rationale and knowledge gap

Demographic characteristics further contextualise leptospirosis patterns in Malaysia. Surveillance data have consistently shown a higher incidence among males and individuals in economically productive age groups (8). Occupational exposure in agriculture, plantation work, construction, and sanitation services likely contributes to this distribution (4,8). Rural communities engaged in farming or residing near water bodies may face increased exposure to contaminated soil and water (4,8). National data sometimes report ethnic distribution patterns (8), but these likely reflect geographic residence and occupational distribution, not inherent biological susceptibility (4). Understanding demographic distribution is therefore essential to interpret spatial patterns and design targeted preventive strategies.

Previous investigations in Malaysia have examined leptospirosis in specific districts or during outbreak periods (10). These studies have documented associations between rainfall, flooding, and case incidence (6,10). However, many analyses have restricted their scope to short timeframes or limited geographic areas. While informative, such studies do not permit evaluation of sustained national trends over extended periods (8). Moreover, rainfall has often been examined as a single exposure variable, with less attention given to the combined or independent roles of temperature, humidity, and flooding at a national scale (6,7). Comprehensive assessments integrating multiple climatic indicators over long-term periods remain comparatively limited in Malaysia.

Spatiotemporal perspectives are particularly important for environmentally mediated infectious diseases (6). Spatial clustering indicates areas where incidence is higher than expected under random distribution, suggesting local environmental or contextual drivers (6). Persistent hotspots may reflect stable ecological conditions that sustain transmission, whereas transient clusters may be associated with episodic events such as flooding (6,7). Temporal analysis enables identification of cyclical patterns, long-term increases or decreases, and anomalous peaks (6). Integrating spatial and temporal dimensions provides a more complex perspective on disease dynamics than either dimension alone.

Understanding long-term trends is especially relevant in the context of climate variability (6). Malaysia has experienced fluctuations in rainfall intensity and flooding frequency over the past decade (8). Extreme weather events may alter exposure pathways, infrastructure resilience, and rodent ecology (6,7). Examining an eleven-year period allows assessment of inter-annual variability beyond isolated outbreaks (8). Such longitudinal evaluation can help determine whether observed peaks represent episodic events or components of broader cyclical patterns.

The use of national surveillance data provides comprehensive geographic coverage and extended temporal scope (8). Surveillance systems capture reported cases across all states, enabling comparison of regional trends (8). However, interpretation of surveillance data must consider potential limitations, including under-reporting of mild cases and variations in diagnostic capacity across regions (4,8). Improvements in laboratory testing and electronic reporting may influence notification trends independently of true incidence (8). Nevertheless, when interpreted cautiously, surveillance data provides advantageous details about population-level patterns and environmental associations.

Given the continued burden of leptospirosis in Malaysia (8), the climatic and environmental conditions conducive to transmission (2,6), and the geographic heterogeneity observed across states (8), a comprehensive national assessment is warranted. Clarifying long-term trends and identifying persistent high-risk areas are critical for informing targeted public health interventions. Understanding associations between climatic variables and disease incidence can support climate-informed preparedness and guide allocation of surveillance and prevention resources (6,7).

By integrating long-term national surveillance data with climatic indicators, this investigation seeks to enhance understanding of how environmental variability shapes leptospirosis transmission dynamics in Malaysia. Evidence generated from this analysis may inform geographically targeted prevention strategies, strengthen surveillance systems, and support adaptive public health planning in the context of ongoing environmental and climatic change.

Objective

Accordingly, this study aimed to comprehensively examine leptospirosis in Malaysia between 2012 and 2022 by describing demographic characteristics, spatiotemporal patterns, and density at the national level. The study further sought to identify spatiotemporal clustering and hotspot areas. In addition, the study aimed to determine the association between selected climatic variables, including temperature, rainfall, humidity, and flooding, and yearly state-level leptospirosis incidence. It was hypothesised that leptospirosis in Malaysia demonstrates significant spatial and temporal variation, with identifiable clustering patterns and hotspots, and that climatic conditions are associated with variation in yearly state-level disease burden. We present this article in accordance with the RECORD reporting checklist (available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0038/rc).

Methods

Study design

This study employed a retrospective ecological design. The ecological unit of analysis was defined at the state level, with annual aggregated case counts being used to characterise temporal trends and geographic variation. This design enabled evaluation of population-level spatiotemporal patterns across diverse climatic, environmental, and socio-geographic contexts within Malaysia.

Setting and study population

The study included all confirmed and probable leptospirosis cases notified through Malaysia between 1 January 2012 and 31 December 2022. The analysis covered all states and federal territories in Malaysia, thereby ensuring comprehensive geographic representation across the country (Figure 1).

Figure 1 Map of the study area across 16 states and federal territories in Malaysia.

Universal sampling was applied, whereby all eligible cases reported within the study period were included. Cases were excluded if notifications were outside the study period, cases were outside the defined study setting, or records had missing or inconsistent data or duplicate records. Even though universal sampling was used in this study, a sample size calculation was conducted to determine whether the available number of cases would provide adequate statistical power for detecting associations between selected climatic exposures and leptospirosis incidence. The level of significance was set at P<0.05 with a 95% confidence interval, a margin of error of 5%, and a statistical power of 80%. Sample size was calculated using OpenEpi version 3.01 based on the Fleiss method (11). Effect size estimates were derived from previously published studies examining flooding, rainfall, temperature, and humidity as environmental risk factors.

The detailed parameters and corresponding sample size estimates are presented in Table S1. The largest estimated required sample size was 13,658. Allowing for a potential 20% reduction due to incomplete data, the minimum required sample size was 16,390. The total number of notified leptospirosis cases included in this study exceeded this threshold, indicating adequate statistical power for analysis.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study received ethical approval from the Medical Research and Ethics Committee (MREC), Ministry of Health Malaysia, with the ethical identification number NMRR ID-24-01790-LOK. Individual consent for this retrospective analysis was waived.

Data sets and sources

Case definition

Case classification followed the national case definition as outlined in the Guidelines for the Diagnosis, Management, Prevention and Control of Leptospirosis in Malaysia [2011] (12). A confirmed case was defined as a clinically compatible case with laboratory confirmation, including a microscopic agglutination test (MAT) titre of ≥1:400 for a single serum specimen or a fourfold rise in titre for paired sera; positive polymerase chain reaction within 10 days of disease onset; positive culture for pathogenic Leptospira from blood within 7 days of onset or urine after day 10; demonstration of Leptospira in tissue using immunohistochemical staining; or positive results from two different rapid diagnostic tests in settings with limited laboratory capacity. A probable case was defined as a clinically compatible illness with positive enzyme-linked immunosorbent assay (ELISA) or other rapid diagnostic tests. Clinical presentation typically included acute febrile illness with exposure to potentially contaminated water or environments, accompanied by symptoms such as myalgia, headache, conjunctival suffusion, jaundice, renal impairment, haemorrhagic manifestations, or gastrointestinal symptoms.

The case definition and reporting framework remained consistent from 2012 to 2022. Although laboratory capacity for MAT testing expanded from July 2013, with an increase in the number of Ministry of Health laboratories capable of confirmatory testing and the establishment of zonal testing centres, diagnostic criteria and notification requirements were unchanged during the study period. Both confirmed and probable cases were therefore included to reflect national incidence reporting practices applicable to the study years.

Leptospirosis surveillance data

Leptospirosis cases were obtained from the Ministry of Health Malaysia national electronic communicable disease notification system. The extracted dataset comprised all registered confirmed and probable leptospirosis cases reported nationwide between 1 January 2012 and 31 December 2022. The dataset included demographic characteristics, notification dates, and the recorded address of occurrence for each case. The address of the occurrence was converted into geographic coordinates through a geocoding process using the Awesome Table add-on in Google Sheets. The resulting latitude and longitude coordinates were subsequently overlaid onto Malaysia’s administrative state boundary basemap to determine the corresponding state for each case. This procedure ensured consistent state-level classification for subsequent ecological and spatial analyses.

Investigators were granted access to a de-identified dataset extracted from the system for research purposes. Access was limited to variables required for analysis and did not include any personal identifiers. Data cleaning procedures were conducted prior to analysis, including removal of duplicate notifications, verification of date variables, standardisation of state identifiers, validation of geographic coordinates, assessment of missing values, and consistency checks across demographic fields. Potential misclassification arising from under-reporting or differential diagnostic capacity across states was acknowledged as an inherent limitation of surveillance-based data.

Population data

District-level population data were derived from the Gridded Population of the World, Version 4.11 (GPWv4.11), Population Density Year Grid 5 Year Estimates dataset developed by the Socioeconomic Data and Applications Center (SEDAC) (13). This dataset provides population density estimates in gridded raster format at a regular spatial resolution for benchmark years, including 2010, 2015, and 2020. To obtain annual population estimates for the study period from 2012 to 2022, intermediate yearly population surfaces were estimated using an exponential growth model (Table S2) based on the observed growth rates between 2010 and 2015 and 2015 and 2020 (14-17).

All raster layers were aligned to a common coordinate reference system and clipped to Malaysia’s administrative state and district boundaries. District-level population counts were calculated by summing all grid cell population values within each district boundary for each year. The resulting dataset comprised annual district-level population counts for the period 2012 to 2022, which were subsequently used as denominators for cumulative incidence and mortality rate calculations.

In addition, state-level population data were obtained from the Department of Statistics Malaysia (DOSM), which provides official annual population estimates based on national census data and intercensal projections. These data were used to complement the district-level estimates and to ensure consistency with nationally reported population figures. These state-level population counts were then incorporated into the statistical models as an offset term (log population), allowing the estimation of incidence rate ratios (IRRs) by accounting for differences in population size across states and years. This approach ensures that the model estimates reflect rates rather than absolute counts, thereby improving the interpretability of the associations between climatic factors and leptospirosis incidence.

Climatic data

Climatic data were obtained from multiple sources and were originally available at daily resolution before being averaged to annual state-level values for analysis. Flooding was derived from satellite remote sensing using the Moderate Resolution Imaging Spectroradiometer aboard the Terra satellite, specifically the Surface Reflectance 8 Day L3 Global 500 m SIN Grid product (MOD09A1) (18). Daily flood extent was identified using the Modified Normalised Difference Water Index (MNDWI) calculated from surface reflectance band 4 (green) and band 7 (infrared), which represent ground-level reflectance in the absence of atmospheric scattering or absorption. Pixels with MNDWI values greater than or equal to zero were classified as water (19,20). State-specific flood extent was calculated by overlaying classified water pixels onto Malaysia’s administrative state boundary basemap. The percentage of flooded area within each state was calculated as the percentage of water-classified pixels relative to the total number of pixels within the state boundary for each day. Missing daily observations were assessed and managed prior to the averaging process to ensure completeness and consistency. These daily state-level flood percentages were then averaged across each year to generate annual flood estimates.

Daily rainfall data were obtained from monitoring stations under the Department of Drainage and Irrigation Malaysia (DID). Meanwhile, daily average temperature and daily average relative humidity data were obtained from monitoring stations under the Department of Environment Malaysia (DOE). Missing daily observations were assessed and managed prior to the averaging process. Station-level daily measurements were first assigned to their respective states based on geographic location. For each state, daily values from all monitoring stations within that state were averaged to produce state-level daily estimates. These daily state-level rainfall, temperature, and humidity values were then averaged across each year to obtain annual rainfall, annual temperature, and annual relative humidity values for each state. All climatic variables were subsequently linked to the leptospirosis dataset at the ecological level using state and year as matching identifiers. No individual-level linkage was performed, ensuring consistency with the ecological design of the study.

Statistical analysis

Missing data

A total of 51,580 leptospirosis case records were initially extracted for the study period. Of these, nine records (0.02%) had missing age information, and seven records (0.01%) had missing ethnicity information. Records with incomplete demographic data were excluded prior to analysis.

Because the MOD09A1 product provides surface reflectance data at 8-day composite intervals rather than continuous daily observations, flood extent was available only at discrete time points. To align flood estimates with other climatic variables and to ensure consistent temporal comparability across states, intermediate values between composite periods were estimated. Missing flood values were calculated separately for each state using linear interpolation between adjacent composite observations through the zoo package in R software version 4.3.3. Interpolation was applied to flood percentage, flooded pixels, and total pixels to preserve the internal consistency of derived measures. No extrapolation beyond the range of observed satellite composites was performed.

For daily rainfall across 1,179 monitoring stations under the DID Malaysia, the median percentage of missing values per station was 3.41% [interquartile range (IQR), 0.45–17.67%] over the study period. For daily average temperature and daily average relative humidity across monitoring stations under the DOE Malaysia, missingness varied by concession phase due to a change in the concession company responsible for station operations and data management. During 2012–2017, across 54 stations, the median percentage of missing values was 16.63% (IQR, 8.29–42.23%) for temperature and 17.07% (IQR, 7.15–41.72%) for relative humidity. During 2018–2022, across 66 stations, the corresponding median percentages were 3.98% (IQR, 3.88–4.29%) for temperature and 3.90% (IQR, 3.88–4.12%) for relative humidity. The pattern of missing data was examined using the naniar package in R software to perform Little’s test for Missing Completely at Random (MCAR) (21). The test was statistically significant, indicating that the data were not missing completely at random. A logistic regression model was subsequently fitted using the base stats package, with missingness specified as the dependent variable and state, station coordinates, year, and month included as predictors. Missingness was significantly associated with observed temporal and spatial characteristics, suggesting that the probability of missing values depended on measured factors rather than unobserved levels. Based on these findings, the missing data showed a Missing at Random (MAR) pattern. To address missing values, multiple imputation using chained equations was implemented at the station level using the mice package in R software (22). The imputation model included state, station coordinates, year, and month as auxiliary predictors to account for spatial and temporal variation. Predictive mean matching was applied to preserve the observed distribution and ensure plausible imputed values. Five imputed datasets were generated to reflect imputation uncertainty. Because the subsequent analysis involved automated model selection, which could produce different model structures across the imputed datasets and could not be directly combined using Rubin’s rules, the imputed values were averaged across the five completed datasets for each observation to create a single pooled dataset. The effect of this approach was assessed through a sensitivity analysis comparing estimates from the averaged pooled dataset model with estimates pooled across the five imputed models using Rubin’s rules for the four primary univariable models. The two approaches produced closely comparable coefficients and IRRs, with identical in statistical significance conclusions (Tables S3,S4).

Descriptive analysis

Continuous variables were assessed for distribution using graphical methods, including histograms with density curves and Q-Q plots (Figures S1,S2). Categorical variables were summarised using proportions and presented as percentage bar charts (Figures S3-S5). Leptospirosis cases were summarised by year of notification. Age, as a continuous variable, was described using the median and IQR. Subsequently, age was categorised in accordance with the Malaysian population age structure, namely <15 years (young age), 15 to 64 years (working age), and ≥65 years (old age) (23). Categorical variables, including age group, sex, ethnicity, and state, were presented as frequencies with percentages. The yearly distribution of leptospirosis cases in Malaysia was plotted according to the date of notification.

Case records containing longitude and latitude were converted into simple feature objects with a geographic coordinate reference system (WGS84). State boundary shapefiles were validated and standardised prior to mapping to ensure geometric consistency. Spatial point distribution maps were generated to visually represent the geographic distribution of leptospirosis cases across Malaysia. Each plotted point corresponded to the recorded coordinates of an individual case. State polygons were overlaid as background layers to contextualise the spatial distribution. In addition to the overall national map, a series of yearly faceted maps was produced to illustrate temporal changes in spatial patterns from 2012 to 2022. Maps were produced using the ggplot2 package, with spatial layers handled through the sf package (24).

Cluster analysis

Kernel density estimates (KDE) are one of the density-based approaches under the first-order effect in spatial point pattern analysis (25). KDE refers to the probability density function used to estimate the spatial intensity of leptospirosis (Table S2). All case coordinates were projected to a planar coordinate reference system (EPSG:3857) prior to analysis. Malaysia’s administrative boundary was defined as the observation window for KDE estimation. Year-specific KDE surfaces were generated by subsetting cases according to year. In this study, the bandwidth smoothing parameter was fixed at 10 kilometres to provide appropriate spatial smoothing at the national level. Duplicate coordinate points were removed prior to KDE estimation. Kernel intensity was calculated using the spatstat package in R software and initially expressed per square metre before being converted to per square kilometre (26). The KDE outputs were converted into raster format and visualised as yearly spatial intensity maps. The sequential kernel maps were plotted using the tmap and raster packages in R software (24).

Another method of spatial point pattern analysis under the distance-based approach was performed to examine spatial spread using the Nearest Neighbour Index (NNI) (27). NNI analysis was conducted on yearly subsets of leptospirosis case locations from 2012 to 2022. For each year, spatial point data were extracted and analysed if the number of cases was at least five. The convex hull of the yearly point distribution was used to define the observation window. The NNI value represents the ratio of the observed mean nearest neighbour distance to the expected mean distance under complete spatial randomness (Table S2). An NNI value close to 1 indicates a random spatial pattern, a value less than 1 suggests clustering, and a value greater than 1 indicates spatial dispersion (27). Year-specific NNI values were computed using the spatialEco package in R software, and results for the period 2012 to 2022 were tabulated for interpretation (28). A P value less than 0.05 was considered statistically significant.

The incidence of leptospirosis in Malaysia was calculated at the district level by dividing the number of notified leptospirosis cases in each district by the corresponding estimated district population for the same year. Case counts were first aggregated by district and year of notification. District population estimates were then merged with case counts to ensure that all districts and years were retained in the analysis. Districts with no reported cases were assigned a value of zero to avoid exclusion from rate calculations. The cumulative incidence was expressed per 100,000 population to enable comparison across districts and over time. Mortality rates were calculated using a similar procedure. Deaths were identified from case records based on the reported patient status and aggregated by district and year. These death counts were merged with the corresponding population denominators, and mortality rates were expressed per 100,000 population. District-level mortality rates were produced to summarise spatial and temporal variations in leptospirosis-related deaths. The formulas used for cumulative incidence and mortality rate were as follows:

Cumulativeincidence=NofnewleptospirosiscasesduringtheyearPopulationatriskduringthesameyear×105[1]

Mortalityrate=NofdeathsduetoleptospirosisduringtheyearTotalpopulationduringthesameyear×105[2]

In this study, Global Moran’s I was used to assess spatial autocorrelation of leptospirosis incidence at the district level across Malaysia. Global Moran’s I reflects the first law of geography, where “Everything is associated with others, and close things are more related compared with distant things”. The statistic measures whether similar values of disease incidence occur near one another across the study area (29). District-level cumulative incidence values were first joined with district boundary shapefiles using the sf package in R software. For each year from 2012 to 2022, spatial coordinates were derived from polygon centroids to represent the geographic location of each district. A spatial weights matrix was then constructed using the spdep package based on the k-nearest neighbour (KNN) approach, where each district was connected to its eight nearest neighbouring districts (k=8). Row standardised spatial weights were applied to ensure comparability across districts. Global Moran’s I statistics were subsequently calculated for each year using the annual district-level incidence values and the corresponding spatial weights matrix. The significance of spatial autocorrelation was evaluated using the Moran’s I statistical test, producing both the Moran’s I value and the associated P value for each year. Interpretation of Moran’s I values followed conventional criteria. Values approaching +1 indicate positive spatial autocorrelation, where districts with high or low incidence tend to cluster geographically. Values close to 0 indicate a random spatial pattern with no spatial dependence. Negative values approaching −1 indicate spatial dispersion, where districts with high incidence tend to be surrounded by districts with low incidence (30).

The Moran’s I statistic has properties to determine the presence of spatial autocorrelation of disease incidence over the whole research area. However, it has limitations in identifying clustered patterns at the local level. Therefore, Local Indicators of Spatial Association (LISA) were applied to identify spatial clusters of leptospirosis incidence at the district level. LISA decomposes the global Moran’s I statistic into contributions from individual spatial units, allowing identification of districts that contribute significantly to spatial clustering. The analysis used the same spatial weights matrix derived from the KNN structure with eight neighbouring districts, ensuring consistency with the global spatial autocorrelation analysis. For each year from 2012 to 2022, Local Moran’s I values were computed based on district-level incidence rates. District-level incidence values were centred by subtracting the mean incidence of the study area to allow comparison between local and global patterns. The results were classified into four standard LISA cluster categories based on the relationship between a district’s incidence value and the average incidence of its neighbouring districts. Districts with high incidence surrounded by districts with similarly high incidence were classified as high-high (HH) clusters, representing potential hotspots. Districts with low incidence surrounded by districts with similarly low incidence were classified as low-low (LL) clusters, representing cold spots. High-low (HL) clusters represent districts with high incidence surrounded by neighbouring districts with low incidence, while low-high (LH) clusters represent districts with low incidence surrounded by districts with high incidence. Statistical significance of the Local Moran’s I statistic was assessed using a significance threshold of P<0.05. Districts that did not meet the significance threshold were classified as non-significant and not considered part of a spatial cluster (31). Year-specific LISA maps were produced to illustrate changes in spatial clustering of leptospirosis incidence across Malaysia from 2012 to 2022.

After identifying the clustering pattern, further analysis was conducted to detect districts with significantly high concentrations of leptospirosis incidence, commonly referred to as ‘high-risk hotspots’. To determine the statistical significance of these hotspots, the Getis-Ord Gi* statistic was applied. This method evaluates the degree of spatial clustering by comparing the local sum of a variable within a neighbourhood to the overall sum of that variable across the entire study area. The analysis used the same spatial weights matrix derived from the KNN structure with eight neighbouring districts that was previously constructed for the spatial autocorrelation analyses. For each year from 2012 to 2022, the Gi* statistic was computed using district-level incidence values. The resulting Gi* values were expressed as z-scores, representing the degree to which a district and its neighbouring districts exhibit unusually high or low incidence values relative to the overall study area. Interpretation of the Gi* statistic followed standard spatial statistical criteria. Positive z-scores indicate clustering of high values, suggesting the presence of a hotspot where districts with high incidence are surrounded by districts with similarly high incidence. Conversely, negative z-scores indicate clustering of low values, representing cold spots where districts with low incidence are surrounded by districts with similarly low incidence. Values close to zero indicate the absence of statistically significant spatial clustering (32). Annual maps were generated to show the spatial evolution of leptospirosis incidence hotspots across Malaysia from 2012 to 2022.

Regression analysis

To examine the association between climatic factors and leptospirosis incidence, a generalised linear mixed model (GLMM) was used. The response variable was the yearly number of leptospirosis cases. The proportion of zero counts in the outcome variable was examined and found to be 1.14%, indicating that only a small fraction of observations contained zero cases. Because the outcome variable represented count data with potential overdispersion (Figure S6) and without substantial zero inflation, a negative binomial distribution with a log link function was specified (33,34). Models were fitted using the glmmTMB package in R software. An offset term, defined as the logarithm of the population, was included to account for differences in population size across state year observations. The general form of the negative binomial GLMM used in this study can be expressed as:

Yit∼NB(μit,k)[3]

Log(μit)=β0+β1Tempit+β2Rainit+β3Humit+β4Floodit+log(Popit)+ui+vt[4]

where Y_it represents the number of leptospirosis cases in state i during year t, μ_it is the expected number of cases, k is the dispersion parameter of the negative binomial distribution, β₀ is the intercept, β₁ to β₄ represent regression coefficients for the climatic predictors (temperature, rainfall, humidity, and flooding), log(Pop_it) is the offset term for population, u_i represents the random effect for state, and v_t represents the random effect for year.

Random effects for year and state were included to account for temporal and spatial heterogeneity in the data. This modelling structure allowed variation in leptospirosis incidence across different years and geographic regions while estimating the fixed effects of climatic variables. These climatic variables are listed and described in Table S5. Prior to modelling, rainfall, temperature, humidity, and flooding were standardised by subtracting the mean and dividing by the standard deviation to ensure comparability of effect estimates. Scaled variables were generated using z-score transformation.

Variable and model selection

Initially, separate univariable GLMMs were fitted for each climatic variable to examine their individual associations with leptospirosis incidence. Each model included a single climatic factor, the log population offset, and random effects for year and state. Before fitting the multivariable models, pairwise correlations among the climatic variables were assessed using Pearson’s correlation coefficients. An absolute correlation coefficient of |r| ≥0.70 was considered indicative of strong correlation and potential multicollinearity. Where two variables exceeded this threshold, only one was selected for inclusion in the multivariable model based on substantive relevance and the minimum Akaike information criterion (AIC) value of the univariable model.

Subsequently, a multivariable model including all climatic variables was constructed. Model selection was performed using the MuMIn package through the dredge function, which generates the 2^N model table with subsets of fixed effect terms, where N is the number of candidate variables. This procedure evaluated all possible combinations of the climatic predictors included in the multivariable model. To retain a multivariable framework, only models containing at least two explanatory variables were considered for selection. Candidate models were ranked using the AIC. Although the model with the lowest AIC was identified, models with a ΔAIC ≤2 relative to the lowest-AIC model were considered to have comparable empirical support. The final model was selected from among these candidate models based on the substantive relevance of the included climatic variables. Effect estimates from the final selected model were reported as IRR with corresponding 95% confidence intervals, obtained by exponentiating the regression coefficients.

Model diagnostics

Residual dependence was also assessed to evaluate whether temporal and spatial structure remained in the model residuals. Pearson residuals were extracted from each fitted model. Temporal dependence was examined by calculating the correlation between residuals and their one-year lag within each state. Spatial dependence was assessed using Moran’s I statistic based on the mean state-level Pearson residuals. Spatial weights were constructed using a four-nearest-neighbour approach based on state centroids derived from the state shapefile after projection to EPSG:3857, with row-standardised weights applied. These residual checks were used to assess whether the fitted models had adequately accounted for remaining temporal and spatial dependence (Table S6). Residual dependence indicated moderate temporal autocorrelation across all fitted models, while Global Moran’s I statistics were non-significant, suggesting no evidence of remaining spatial autocorrelation in the model residuals.

Results

A total of 51,564 leptospirosis cases were reported in Malaysia between 2012 and 2022 after excluding 16 records with incomplete demographic information from the initial dataset. The median age of cases remained relatively consistent across the study period, ranging from 29 to 33 years, with an overall median age of 30 years. The majority of infections occurred among individuals aged 15 to 64 years, accounting for 76.4% (n=39,379) of cases. Children aged below 15 years represented 16.7% (n=8,596), while older adults aged 65 years and above accounted for 7.0% (n=3,589). Males accounted for the majority of reported cases throughout the study period, comprising 68.1% (n=35,107) of infections, whereas 31.9% (n=16,457) occurred among females. With respect to ethnicity, Malay Bumiputera represented the largest proportion of cases (62.6%, n=32,294), followed by Other Bumiputera (17.1%, n=8,827) and Others (non-citizen) (8.4%, n=4,328). Smaller proportions were observed among Chinese (6.2%, n=3,191) and Indian (4.7%, n=2,433) populations, while others (citizen) groups accounted for 1.0% (n=491). Geographically, cases were distributed across all Malaysian states. The highest cumulative number of cases was reported in Selangor (17.9%, n=9,207), followed by Kelantan (14.3%, n=7,370) and Sarawak (14.2%, n=7,323). Other states with notable numbers of cases included Kedah (8.2%, n=4,254), Terengganu (7.6%, n=3,923), and Perak (7.4%, n=3,812). In contrast, fewer cases were reported in Perlis (0.3%, n=159), Labuan (0.3%, n=172), and Putrajaya (0.1%, n=74). Across the study period, the annual number of reported cases (Table S7 and Figure 2) increased from 3,691 cases in 2012 and reached a peak between 2014 and 2016, with 7,680 cases in 2014, 8,286 cases in 2015, and 5,286 cases in 2016. The number of cases then decreased to 4,344 in 2017, followed by a secondary increase to 5,094 in 2018 and 5,177 in 2019. They subsequently declined to 1,737 in 2021 before increasing modestly to 2,875 in 2022.

Figure 2 Trend of leptospirosis cases in Malaysia between 2012 and 2022. Purple bars represent the monthly number of reported cases. The blue line represents the LOESS trend, and the grey shaded area indicates the corresponding 95% confidence interval around the smoothed trend. LOESS, locally estimated scatterplot smoothing.

The overall distribution of leptospirosis cases in Malaysia between 2012 and 2022 (Figure S7) showed that cases were reported across districts throughout the country, indicating widespread occurrence in both Peninsular Malaysia and East Malaysia during the study period. Several districts in Selangor, Kelantan, and Sarawak reported comparatively higher numbers of cases than other areas. The yearly distribution of leptospirosis cases (Figure S8) further demonstrated temporal variation in the geographic distribution of cases across districts. A noticeable increase in reported cases was observed during the 2014 to 2016 period, particularly across several districts in Peninsular Malaysia. In the subsequent years, the number of reported cases varied across districts, although cases continued to be reported throughout the country, and by 2022, leptospirosis cases remained geographically distributed across districts in both Peninsular Malaysia and East Malaysia.

The spatial density of leptospirosis cases across Malaysia between 2012 and 2022 is presented in Figure 3. The maps illustrate spatial variation in the density of reported cases across districts throughout the study period. Higher density values were observed in several districts across Peninsular Malaysia, particularly along the western and eastern regions, while only selected districts in Sarawak exhibited similarly high-density values. Overall, high density districts were more widespread in Peninsular Malaysia than in East Malaysia. Temporal variation in spatial density was also evident across the study years. Higher density values were observed during the 2014 to 2016 period, corresponding with the years in which larger numbers of cases were reported. In the subsequent years, the spatial density generally decreased in several districts, although areas with higher density remained present across different regions of the country. By 2022, spatial density values continued to vary across districts in both Peninsular Malaysia and East Malaysia.

Figure 3 Spatial density of leptospirosis cases in Malaysia between 2012 and 2022. KDE, kernel density estimate.

The spatial pattern of leptospirosis cases demonstrated a clustered distribution throughout the study period, with NNI values ranging from 0.19 to 0.28 (Table 1). An NNI value smaller than 1 indicates that the observed distribution of cases is more clustered than would be expected under a random spatial pattern. The observed mean distances between cases were substantially smaller than the expected mean distances under spatial randomness, further supporting the presence of spatial clustering. The strongest clustering was observed during the 2014 and 2015 periods, where the NNI values were 0.19 (P<0.001), coinciding with the years that recorded higher numbers of leptospirosis cases. For instance, in 2015, the observed mean distance between neighbouring cases was 1,085.75 metres, compared with an expected mean distance of 5,825.47 metres under a random spatial distribution. In contrast, relatively higher NNI values were observed in 2021 (NNI =0.28) and 2020 (NNI =0.25), suggesting comparatively weaker clustering during these years, although the spatial pattern remained clustered. Overall, the P values were <0.001 across all years, indicating that the clustering pattern of leptospirosis cases was statistically significant and consistently present throughout the study period.

The cumulative incidence was calculated and used for Moran’s I, LISA, and Getis-Ord Gi* analyses. The result is presented in Table S8 and Figure S9. Several districts consistently recorded a higher incidence compared with surrounding areas. In Peninsular Malaysia, relatively elevated cumulative incidence was observed in districts within Kedah, Kelantan, Perak, Negeri Sembilan, and Terengganu. Districts with the highest average cumulative incidence included Jelebu (104.03 per 1000,000), Gua Musang (99.75 per 100,000), Jeli (81.17 per 100,000), Kuala Krai (77.71 per 100,000), Dungun (77.30 per 100,000), Setiu (71.98 per 100,000), Sik (66.81 per 100,000), Batang Padang (57.96 per 100,000), and Pasir Mas (55.62 per 100,000). In East Malaysia, higher cumulative incidence was particularly concentrated in several districts in Sarawak, most prominently Kapit, which recorded the highest average incidence (525.98 per 100,000), followed by Marudi (86.44 per 100,000), Kanowit (75.84 per 100,000), Song (66.40 per 100,000), Julau (61.14 per 100,000), and Betong (57.34 per 100,000). Temporal patterns were also evident, with more districts showing higher cumulative incidence during the peak period between 2014 and 2016. These results provide an overview of districts with higher leptospirosis risk. The mortality rate was also calculated and is shown in Table S9 and Figure S10, further supporting the distribution of disease burden observed in the cumulative incidence patterns. Higher average mortality rates were observed in districts such as Kuala Krai (1.50 per 100,000), Hulu Perak (1.32 per 100,000), Rembau (1.17 per 100,000), Jelebu (1.13 per 100,000), and Gua Musang (1.08 per 100,000) in Peninsular Malaysia. In East Malaysia, mortality burden was again prominent in several districts in Sarawak, particularly Kapit (5.31 per 100,000), followed by Marudi (2.69 per 100,000), Maradong (2.46 per 100,000), Betong (1.82 per 100,000), Lawas (1.73 per 100,000), Song (1.67 per 100,000), Limbang (1.66 per 100,000), and Simunjan (1.63 per 100,000), while Kuala Penyu (2.15 per 100,000) recorded the highest average mortality in Sabah.

The spatial autocorrelation of leptospirosis cases across districts in Malaysia was examined using the Global Moran’s I statistic, as presented in Table 1. The Moran’s I values ranged from 0.023 to 0.210 over the study period, indicating varying degrees of spatial autocorrelation in the incidence of leptospirosis cases. Most years demonstrated significant positive spatial autocorrelation. The highest Moran’s I value was observed in 2013 (Moran’s I =0.209, P<0.001), followed by 2015 (Moran’s I =0.193, P<0.001) and 2014 (Moran’s I =0.178, P<0.001), indicating stronger spatial clustering during these years. Positive spatial autocorrelation remained significant in subsequent years, although the magnitude of the Moran’s I values was comparatively lower. In contrast, 2012 showed a non-significant Moran’s I value (Moran’s I =0.023, P=0.13), suggesting that the spatial distribution of leptospirosis incidence during that year did not significantly deviate from a random pattern. Nevertheless, significant spatial autocorrelation was observed in most of the remaining years, including 2021 (Moran’s I =0.049, P=0.03) and 2022 (Moran’s I =0.159, P<0.001).

The LISA analysis of leptospirosis incidence at the district level is presented in Figure 4. The analysis identified statistically significant spatial clusters classified as HH, LL, LH, and HL clusters across different years during the study period. HH clusters, representing districts with high incidence surrounded by neighbouring districts with similarly high incidence, were observed in several years. In 2012, a cluster was identified in Belaga (Sarawak). In 2013, clusters were detected in Kecil Lojing (Kelantan), Lipis (Pahang), and several districts in Sarawak, including Julau, Kanowit, Lubok Antu, Pakan, and Sarikei. During 2014, clustering expanded to districts in Kelantan (Gua Musang, Jeli, Kuala Krai, Machang, and Pasir Mas); Perak (Hulu Perak); Sabah (Beaufort, Kuala Penyu, and Tenom); and Sarawak (Belaga, Bintulu, Marudi, and Sebauh). In 2015, clusters remained concentrated in Kelantan districts, including Bachok, Gua Musang, Jeli, Kuala Krai, Machang, Pasir Mas, Pasir Puteh, Tanah Merah, and Tumpat, as well as Hulu Perak (Perak) and Besut (Terengganu). From 2016 onwards, clusters were predominantly identified in Sarawak districts such as Selangau, Song, Kapit, Bukit Mabong, Lubok Antu, and Tatau, with additional clusters in Hulu Terengganu (Terengganu) and Kuala Pilah (Negeri Sembilan) in certain years. In 2022, clusters were again observed in Kelantan (Gua Musang, Jeli, and Kuala Krai); Sarawak (Belaga, Bukit Mabong, Song, and Tatau); and Hulu Terengganu (Terengganu).

Figure 4 Map of LISA analysis of leptospirosis incidence in Malaysia at the district level. LISA, local indicators of spatial association.

LL clusters, indicating districts with low incidence surrounded by neighbouring districts with similarly low incidence, were observed in 2013. These clusters were located in several districts in Johor (Batu Pahat, Johor Bahru, Kluang, Kota Tinggi, Kulai, Mersing, and Pontian); Perak (Bandar Baharu and Kerian); and Pulau Pinang (Barat Daya, Seberang Perai Selatan, Seberang Perai Tengah, Seberang Perai Utara, and Timur Laut). LH clusters, representing districts with relatively low incidence surrounded by neighbouring districts with higher incidence, were observed in multiple years. In 2012, LH clusters were identified in Rembau (Negeri Sembilan) and several Sarawak districts, including Bintulu, Bukit Mabong, Sebauh, Selangau, Song, and Tatau. In 2013, clusters were detected in Betong, Bukit Mabong, Saratok, Sebauh, and Sibu (Sarawak). In 2014, LH clusters occurred in Bachok and Tanah Merah (Kelantan); Labuan and Nabawan (Sabah); several Sarawak districts, including Bukit Mabong, Lawas, Limbang, Miri, and Subis; and Hulu Terengganu (Terengganu). Similar LH clusters persisted across several Sarawak districts between 2015 and 2020, while additional LH clusters were observed in 2021 in Jempol and Seremban (Negeri Sembilan), Bentong, Bera, Maran and Temerloh (Pahang), and Ulu Langat (Selangor). In 2022, LH clusters were detected in Pasir Mas (Kelantan), Lipis (Pahang), Hulu Perak (Perak), and Sebauh and Selangau (Sarawak). Conversely, HL clusters, representing districts with higher incidence surrounded by neighbouring districts with lower incidence, were observed in several years. In 2013, HL clusters occurred in Song and Subis (Sarawak). In 2014, clusters were detected in Pendang (Kedah) and Kota Bharu and Pasir Puteh (Kelantan). In 2015, HL clusters were identified in Kota Bharu (Kelantan), Selangau and Song (Sarawak), and Hulu Terengganu (Terengganu). Additional HL clusters were observed in Bukit Mabong (Sarawak) in 2019 and Bintulu (Sarawak) in 2022.

The Getis-Ord Gi* analysis further supported the spatial clustering patterns identified in the LISA analysis (Figure S11). Districts previously identified as HH clusters in the LISA analysis, particularly in several districts in Sarawak and parts of Kelantan and Terengganu, consistently exhibited high Gi* statistic values, indicating significant hot spots of leptospirosis incidence across multiple years between 2012 and 2022. These hot spots were most evident during the peak transmission period between 2014 and 2016 and remained detectable in several districts in subsequent years, including 2022. In contrast, districts classified as LL clusters in the LISA analysis, particularly in parts of Johor, Perak, and Pulau Pinang, generally corresponded with lower Gi* values, indicating cold spot areas with relatively lower leptospirosis incidence.

To identify climatic factors associated with leptospirosis incidence at the state level, univariable analyses were first conducted for the scaled climatic variables (Table S10). Among the examined variables, temperature showed a statistically significant association with leptospirosis incidence (IRR =0.87, 95% CI: 0.77–0.98, P=0.02), while rainfall (IRR =0.91, 95% CI: 0.74–1.10, P=0.32), humidity (IRR =1.13, 95% CI: 0.99–1.31, P=0.08), and flooding (IRR =0.94, 95% CI: 0.78–1.13, P=0.51) were not statistically significant. Prior to model selection, the relationships among climatic variables were examined using cross-correlation analysis (Figure S12). The correlation coefficients between the climatic variables were generally low to moderate, with rainfall showing a moderate positive correlation with humidity (r=0.35) and a negative correlation with temperature (r=−0.28). Temperature was weakly negatively correlated with humidity (r=−0.22) and flooding (r=−0.05), while humidity showed a weak positive correlation with flooding (r=0.10). Model selection was subsequently performed using a set of GLMMs that included all climatic variables, namely temperature, rainfall, humidity, and flooding. Candidate models were compared using both AIC and the corresponding ΔAIC (Table S11). Although the model including temperature and humidity produced the lowest AIC value (AIC =2077.73), the model including temperature, rainfall, and humidity (the third model) was selected as the final model because it had comparable empirical support (ΔAIC =0.73) and retained climatic variables considered substantively relevant to this study. In the final GLMM (Table 2), temperature remained significantly associated with leptospirosis incidence (IRR =0.88, 95% CI: 0.78–0.99, P=0.041), whereas rainfall (IRR =0.90, 95% CI: 0.74–1.09, P=0.26) and humidity (IRR =1.09, 95% CI: 0.95–1.26, P=0.22) were not statistically significant. These findings suggest that lower temperature levels were associated with higher leptospirosis incidence, while rainfall and humidity showed weaker and non-significant associations after adjustment in the final model. The set of these explanatory variables is mapped together with leptospirosis cases and shown in Figures S13-S16.

Discussion

Key findings

This study described the demographic characteristics, spatiotemporal patterns, and spatial density of leptospirosis in Malaysia between 2012 and 2022. The burden of disease was predominantly concentrated among the working-age population, with individuals aged 15 to 64 years accounting for more than three-quarters of reported cases. Males consistently represented the majority of infections throughout the study period. Spatiotemporal patterns demonstrated that higher numbers of cases were observed in several districts, particularly in Selangor, Kelantan, and Sarawak, peaking between 2014 and 2016, followed by fluctuations in subsequent years. Spatial density analysis demonstrated that higher density values were observed in several districts in Peninsular Malaysia, particularly along the western and eastern regions, while comparatively lower density values were noted across many districts in East Malaysia. Temporal variation in spatial density aligned with case trends.

Spatiotemporal analyses demonstrated clear and consistent clustering of leptospirosis cases throughout the study period. The NNI indicated a clustered distribution across all years, with values ranging from 0.19 to 0.28, and the strongest clustering was observed during 2014 and 2015, corresponding to peak transmission periods, while comparatively weaker clustering was noted in 2020 and 2021, although the pattern remained clustered. Global Moran’s I showed predominantly significant positive spatial autocorrelation throughout the study period, with the highest clustering observed in 2013, 2014, and 2015, while 2012 did not demonstrate a significant deviation from a random pattern. Local spatial analyses further identified persistent HH clusters across multiple years, particularly in districts within Kelantan, Terengganu, and Sarawak, with clustering expanding during peak years and subsequently becoming more concentrated in selected districts. In contrast, LL clusters were observed in parts of Johor, Perak, and Pulau Pinang, indicating areas of consistently lower incidence. These patterns were further supported by Getis-Ord Gi* findings.

At the state level, climatic factors showed varying associations with leptospirosis incidence. In univariable analyses, temperature was the only variable significantly associated with yearly incidence, with higher temperature levels corresponding to approximately 13% lower incidence. In contrast, rainfall, humidity, and flooding did not demonstrate statistically significant associations, although rainfall and flooding showed a tendency towards reduced incidence (approximately 9% and 6% lower, respectively), while humidity showed a tendency towards increased incidence (approximately 13% higher). Model selection identified several competing models with comparable empirical support; the model including temperature, rainfall and humidity was selected as the final model based on AIC, consideration of competing models within ΔAIC ≤2, and the substantive relevance of the included climatic variables. In the final model, temperature remained significantly associated with leptospirosis incidence, with lower temperature levels associated with approximately 12% higher incidence. Rainfall and humidity remained non-significant after adjustment, with rainfall showing an approximately 10% lower incidence and humidity an approximately 9% higher incidence, while flooding was not retained in the final model due to limited contribution.

Strengths and limitations

This study has several notable strengths that enhance the robustness and relevance of its findings. First, it utilised a large, nationally representative dataset comprising more than 50,000 reported leptospirosis cases over an eleven-year period, enabling a comprehensive assessment of long-term epidemiological trends. The use of national surveillance data allowed for coverage across all states and districts in Malaysia, providing a strong foundation for examining both spatial and temporal variations in disease occurrence. Such longitudinal data are particularly valuable for environmentally mediated infectious diseases, where transmission dynamics often fluctuate over time.

Second, the study applied a rigorous spatiotemporal analytical framework incorporating multiple complementary spatial methods, including NNI, Global Moran’s I, LISA, and Getis-Ord Gi*. The integration of these approaches allowed for consistent identification of clustering patterns at both global and local scales, strengthening confidence in the observed spatial heterogeneity and enabling the detection of persistent hotspots and cold spots.

Third, climatic determinants were analysed using a GLMM with state-level random effects and population offsets, allowing for estimation of IRRs while accounting for underlying population structure and clustering over time. Model selection based on the AIC, while considering competing models with comparable empirical support (ΔAIC ≤2) and the substantive relevance of the included climatic variables, further enhanced the methodological rigour by balancing model fit, parsimony, and interpretability. The inclusion of multiple climatic variables within the same model also allowed for examination of their relative contributions rather than treating them in isolation.

Lastly, the findings of this study are likely to be generalisable to other tropical and subtropical settings with similar climatic, environmental, and socioeconomic conditions. Malaysia’s equatorial climate, characterised by consistently high temperatures, substantial rainfall, and elevated humidity, provides an ecological context comparable to many countries in Southeast Asia, South Asia, and parts of Latin America where leptospirosis remains endemic. The identification of a predominantly working-age male population at higher risk, alongside geographically clustered hotspots, reflects patterns that have been consistently reported across diverse endemic regions, suggesting that these findings have broader relevance beyond the Malaysian context.

Several limitations should be considered when interpreting the findings. First, the study relied on routinely collected surveillance data that were not originally designed for research purposes. As such, misclassification of cases, underreporting, and incomplete capture of mild or subclinical infections may have occurred. These factors could lead to underestimation of the true burden and introduce bias in the observed associations.

Second, the use of aggregated state-level climatic data introduces the potential for ecological bias. Climatic exposures such as temperature, rainfall, humidity, and flooding were measured at broader geographic scales and may not accurately reflect local environmental conditions experienced by individuals. This mismatch between exposure and outcome measurement could attenuate observed associations.

District-level incidence and mortality rates may also be unstable in sparsely populated interior districts, such as Kapit, Kanowit, Julau, and Sipitang, because of their relatively small population denominators. In these settings, even a small number of cases or deaths may produce disproportionately high rates. Comparisons involving these districts should therefore be interpreted cautiously, particularly where estimates are based on few events.

A key methodological limitation is the absence of lagged climatic effects in the modelling approach. Previous studies have consistently demonstrated that climatic variables such as rainfall, temperature, and humidity influence leptospirosis transmission with delayed effects, often ranging from several weeks to months (35,36). These lag structures reflect the time required for environmental contamination, bacterial proliferation, and subsequent human exposure. In the present study, climatic variables were analysed at an annual level, which limits the ability to capture short-term temporal variability and lagged relationships. By modelling climatic variables without incorporating lag terms, the analysis may not have fully captured the temporal dynamics of transmission, potentially attenuating or obscuring true associations, particularly for rainfall, humidity, and flooding.

Furthermore, interactions between climatic variables and non-linear relationships were not extensively explored. Given the complex interplay between temperature, rainfall, humidity, and hydrological processes, the use of linear and contemporaneous associations may oversimplify underlying mechanisms. In reality, the relationship between climatic factors and leptospirosis transmission is unlikely to be strictly linear. For instance, incremental increases in rainfall may have minimal impact up to a certain threshold, beyond which substantial environmental changes such as flooding, soil saturation, or overflow of drainage systems may occur, leading to a disproportionate increase in exposure risk. Similarly, temperature may exert both facilitating and limiting effects depending on its range, where moderate conditions support bacterial survival while extreme heat may reduce environmental persistence.

Comparison with similar research

The present study found that the burden of leptospirosis in Malaysia was concentrated among the working-age population, with individuals aged 15 to 64 years accounting for more than three-quarters of reported cases, while males consistently represented the majority of infections throughout the study period. These findings are consistent with previous literature. Evidence from Pakistan has shown that individuals aged 19 to 50 years demonstrate the highest seroprevalence and significantly increased odds of infection, indicating that economically productive age groups are at greater risk of exposure (37). Malaysian epidemiological data have similarly reported the highest incidence among adults aged 30 to 39 years, with comparatively lower rates among younger age groups (38). Findings from the Netherlands further reinforce this pattern, demonstrating lower incidence among younger and older populations compared with middle-aged individuals (39).

The predominance of male cases observed in this study is also consistent with prior evidence. Malaysian data from 2004 to 2012 showed that males accounted for the majority of infections, with substantially higher incidence compared with females (38). Studies conducted in Pakistan have similarly reported higher odds of infection among males (37), while evidence from the Netherlands demonstrated markedly higher incidence among males compared with females (39). Although some studies have reported no statistically significant differences in susceptibility between sexes (40), the overall epidemiological pattern across settings consistently shows a higher burden among males.

The temporal patterns observed in this study are also in agreement with existing evidence. The increase in cases between 2014 and 2016, followed by subsequent variation, reflects the well-documented fluctuation of leptospirosis incidence over time. Previous literature has shown that in tropical regions, disease occurrence varies with climatic conditions, with higher incidence often observed during periods of increased rainfall and environmental moisture (5,6). Malaysian surveillance data have similarly reported fluctuating case numbers, with peaks occurring in certain years, particularly following periods associated with environmental changes and flooding events (8,10).

The geographical distribution identified in this study also corresponds with previous findings highlighting spatial heterogeneity in leptospirosis burden. The observation that higher numbers of cases were concentrated in Selangor, Kelantan, and Sarawak is consistent with reports indicating variation in incidence across Malaysian states (8). In particular, states in East Malaysia, especially Sarawak, have been reported to experience higher incidence compared with many regions in Peninsular Malaysia (8). Differences in rainfall patterns, environmental conditions, land use, and occupational activities have been suggested as contributing factors to these spatial variations (6,8).

The spatial density patterns observed in this study are also in line with existing understanding of disease distribution. Higher density values observed in several districts in Peninsular Malaysia, particularly along the western and eastern regions, are consistent with the concept that areas with greater environmental exposure or population concentration may experience higher aggregation of cases. Conversely, the lower density observed across many districts in East Malaysia reflects a more dispersed pattern of cases across larger geographic areas. Temporal variation in spatial density, corresponding with peaks in case numbers between 2014 and 2016, is also consistent with previous observations that periods of increased transmission are associated with greater spatial concentration of cases (6).

The spatial clustering patterns identified in this study are similarly supported by regional evidence. Ecological modelling in western Java, Indonesia, demonstrated that predicted probabilities of leptospirosis occurrence were higher in specific geographic zones, particularly in the northern and central regions (41). Spatial analyses from Thailand have also reported significant positive Global Moran’s I values, indicating a non-random and clustered distribution of cases, with high-risk provinces concentrated in defined regions (42). These findings are consistent with the present study, which identified persistent clustering and hotspot areas across multiple years.

Notably, the clustering patterns identified, including persistent HH clusters in Kelantan, Terengganu, and Sarawak, are consistent with evidence suggesting that certain areas function as sustained hotspots of transmission. The overlap between high-burden districts and identified hotspots in this study is also comparable with findings from other settings, where areas with elevated incidence tend to coincide with persistent clustering. The dynamic pattern of clustering, with expansion during peak transmission periods and contraction in subsequent years, further aligns with previous observations of temporally shifting clusters in endemic regions (42). Similarly, the identification of LL clusters in Johor, Perak, and Pulau Pinang is consistent with reports from other settings where certain regions consistently demonstrate lower incidence.

However, the findings of this study differ from more localised analyses conducted in Selangor, where spatial patterns were reported to be largely random, with only limited clustering and scattered hotspots (43). This difference may reflect variation in spatial scale, as broader national-level analyses are more likely to detect structured clustering patterns compared with more localised studies.

With regard to climatic factors, the findings of this study both align with and differ from existing literature. The absence of a significant association between flooding and leptospirosis incidence contrasts with a substantial body of evidence describing flooding as a key driver of outbreaks. Studies from India, the Philippines, and Argentina have reported marked increases in cases following flood events (44-46), while Malaysian evidence has identified flood-related indices as important predictors of disease occurrence (47) and highlighted monsoon-associated flooding as a recurring context for outbreaks (48). Nevertheless, more recent studies have reported non-significant associations between flooding and leptospirosis incidence when analysed alongside other environmental variables (49), which is consistent with the findings of the present study.

The non-significant association observed for rainfall is also consistent with the mixed findings reported in the literature. While several studies have demonstrated positive associations between rainfall and leptospirosis incidence in tropical settings (1,50), other studies have reported inverse or non-significant relationships, particularly in urban environments (51). Malaysian studies have similarly reported variable findings, including negative correlations (52) and evidence suggesting that rainfall may influence disease occurrence indirectly through hydrological changes such as river water levels (43).

The inverse association observed for temperature in this study differs from many studies that report positive relationships between temperature and leptospirosis incidence. Evidence from Argentina and Guadeloupe has shown higher case occurrence during periods of elevated temperature and rainfall (46,53), while other studies have identified temperature as an important determinant of environmental suitability for transmission (1,41). However, heterogeneous findings have also been reported, including negative associations for maximum temperature and non-significant relationships in certain settings (49,52), indicating variability across contexts.

The findings for humidity in this study are also consistent with the mixed evidence reported in the literature. While biological and epidemiological studies have shown that high humidity supports environmental persistence of Leptospira and is associated with increased incidence (5,35,49), Malaysian studies have reported non-significant or variable associations (52). This variability suggests that the influence of humidity may depend on interactions with other climatic and environmental factors.

Explanations of findings

The findings of this study demonstrate that leptospirosis in Malaysia is characterised by a distinct concentration of cases within specific demographic groups, alongside marked spatial heterogeneity and temporal variability. The predominance of cases among individuals aged 15 to 64 years suggests that disease transmission is closely linked to patterns of daily activity and environmental exposure. This age group represents the economically productive segment of the population and is more likely to be engaged in occupational or recreational activities that involve direct or indirect contact with contaminated environments (1,4). In contrast, the relatively lower burden observed among younger and older age groups may reflect reduced exposure to such high-risk environments (39).

The consistent predominance of male cases further reinforces the role of exposure-related factors in shaping disease distribution. Rather than reflecting inherent biological susceptibility, this pattern likely arises from differential engagement in outdoor or labour-intensive activities, where contact with contaminated soil and water is more frequent (1,4). This is supported by epidemiological evidence indicating that occupational and behavioural factors contribute substantially to sex-based differences in leptospirosis risk (38,39). The demographic profile observed therefore points towards exposure-driven transmission dynamics, in which occupational and environmental interactions play a central role.

The temporal pattern identified, with a clear peak between 2014 and 2016 followed by subsequent variation, suggests that leptospirosis transmission in Malaysia is not constant over time but instead fluctuates in response to broader environmental conditions. These fluctuations may reflect interannual variability in climatic conditions, environmental disruption, or changes in human behaviour that influence exposure pathways (6,8). The recurrence of increased cases in later years indicates that transmission remains sustained within the population, with periodic amplification rather than isolated outbreaks (8,10).

Spatially, the findings reveal that leptospirosis is not evenly distributed across Malaysia but instead demonstrates clear geographic concentration in selected regions. The higher number of cases observed in states such as Selangor, Kelantan, and Sarawak suggests that local environmental and contextual factors play an important role in shaping transmission. These may include differences in rainfall patterns, land use, population density, occupational structure, and infrastructure (6,8). The presence of higher spatial density in several districts in Peninsular Malaysia, particularly along the western and eastern regions, further indicates that certain areas experience greater aggregation of cases, potentially reflecting both environmental suitability and population concentration.

The spatial clustering analyses provide additional evidence that leptospirosis transmission is spatially structured. The consistent finding of clustering across all years, with stronger clustering during peak transmission periods, indicates that cases tend to occur in proximity to one another rather than being randomly distributed. This suggests the presence of localised transmission dynamics, where environmental conditions, reservoir presence, or human exposure pathways create conditions that sustain disease occurrence within specific geographic areas (41,42). The identification of persistent HH clusters, particularly in districts within Kelantan, Terengganu, and Sarawak, indicates that these areas may function as long-standing hotspots of transmission.

Importantly, the overlap between areas of high cumulative incidence, elevated mortality rates, and identified spatial hotspots suggests that certain districts experience a convergence of increased transmission intensity and more severe disease outcomes. This pattern may reflect a combination of environmental risk factors, delayed healthcare access, or differences in health system capacity (1). In contrast, the identification of LL clusters in parts of Johor, Perak, and Pulau Pinang indicates that some regions consistently experience lower disease burden, further highlighting the presence of geographic disparities.

The spatial density findings complement these observations by demonstrating that areas with higher case counts also tend to exhibit greater concentration of cases, particularly during peak transmission periods. The alignment between temporal increases in case numbers and higher spatial density suggests that periods of intensified transmission are accompanied by greater geographic clustering of cases, rather than uniform increases across all regions. This reinforces the interpretation that leptospirosis transmission occurs within defined geographic contexts rather than being evenly dispersed (6).

The analysis of climatic factors indicates that environmental influences on leptospirosis are complex and not uniformly expressed across variables at the state level. Temperature emerged as the most consistent variable, with lower temperature levels associated with higher incidence. This inverse relationship may suggest that within a tropical setting where baseline temperatures are already high, relatively lower temperature conditions may create more favourable environments for bacterial persistence or alter patterns of human and animal interaction with water sources (5,6). Alternatively, temperature may be acting as a proxy for broader environmental conditions that influence transmission dynamics (1).

In contrast, rainfall, humidity, and flooding did not demonstrate statistically significant associations after adjustment, although directional trends were observed. The tendency for humidity to be associated with increased incidence is consistent with its role in maintaining environmental moisture, which supports bacterial survival (5,6). However, the lack of statistical significance suggests that humidity alone may not be sufficient to explain variations in disease occurrence at the state level. Similarly, the observed tendency for rainfall and flooding to be associated with reduced incidence may reflect the influence of more complex hydrological processes, such as runoff, drainage, or dilution of contamination, rather than direct linear effects (6,49).

The absence of strong independent associations for rainfall, humidity, and flooding also suggests that these variables may operate through interconnected pathways rather than acting as isolated determinants. Environmental processes such as water accumulation, soil saturation, and river level changes may mediate the relationship between climatic conditions and human exposure (43). As such, simple annual measures of these variables may not fully capture the dynamic and context-dependent nature of leptospirosis transmission (35,36).

Implications and actions needed

Public health

At the national level, the identification of clear spatiotemporal clustering and persistent hotspots indicates a need to strengthen surveillance systems to move beyond passive case detection towards more proactive, predictive approaches. Traditional surveillance mechanisms, while essential, may not adequately capture early signals of increased transmission risk, particularly in environmentally mediated diseases such as leptospirosis. Integrating climatic and environmental indicators into surveillance systems could enhance the ability to anticipate outbreaks. For example, monitoring trends in temperature variability, rainfall patterns, and hydrological changes may allow for the development of early warning systems that can trigger pre-emptive public health responses before case numbers rise.

Such systems would require strengthening data integration across sectors, including meteorological, environmental, and health surveillance platforms. Real-time or near real-time data linkage could support more responsive decision-making, particularly in districts that have demonstrated recurrent clustering. The temporal peaks identified between 2014 and 2016, and subsequent fluctuations in the following years, suggest that transmission intensifies during specific periods, reinforcing the value of seasonal preparedness strategies. These could include targeted risk communication, pre-positioning of diagnostic resources, and reinforcement of preventive measures in anticipation of higher-risk periods.

Preventive strategies should also be geographically targeted. The consistent identification of hotspot areas, particularly in Kelantan, Terengganu, and Sarawak, indicates that resource allocation should prioritise districts with sustained transmission. This may involve intensified environmental sanitation programmes, rodent control initiatives, and community-based interventions focused on reducing exposure to contaminated water and soil. At the same time, the identification of low-incidence regions suggests that tailored approaches are necessary, as a uniform national strategy may not be equally effective across diverse settings.

The demographic findings further highlight the importance of targeting high-risk populations. Working-age individuals, particularly males, are more likely to be engaged in occupations involving environmental exposure, such as agriculture, construction, and sanitation work. Occupational health strategies should therefore be strengthened, including the provision of protective equipment, education on safe work practices, and workplace-based health promotion initiatives. In addition, recreational exposures, particularly in flood-prone or water-rich environments, should be addressed through public awareness campaigns that emphasise risk-reduction behaviours.

Healthcare system preparedness also warrants attention. The overlap between high incidence and higher mortality in certain districts suggests potential disparities in access to timely diagnosis and treatment. Strengthening clinical awareness, improving diagnostic capacity at peripheral health facilities, and ensuring prompt referral pathways may help reduce severe outcomes. Given the non-specific clinical presentation of leptospirosis, enhancing diagnostic suspicion among healthcare providers remains an important component of early case management.

Global health

The patterns observed in this study reflect broader trends seen in tropical regions, where leptospirosis is closely linked to climatic variability and environmental conditions. The consistency of findings across multiple countries underscores that leptospirosis is not solely a localised issue but part of a wider global health challenge associated with zoonotic diseases in tropical and subtropical settings.

This has implications for regional collaboration, particularly within Southeast Asia, where countries share similar climatic conditions, ecological environments, and population movement patterns. Cross-border transmission risks may arise through shared river systems, agricultural practices, and labour migration. As such, strengthening regional surveillance networks and promoting data sharing between neighbouring countries could improve collective preparedness and response capacity.

Standardisation of surveillance definitions, reporting systems, and analytical approaches would further enhance the comparability of data across countries. This is particularly important for environmentally mediated diseases, where regional climatic patterns such as monsoon systems influence transmission dynamics across national boundaries. Collaborative research initiatives could also support the development of region-specific predictive models that incorporate climatic and environmental variables.

In addition, global health strategies should recognise the role of socioeconomic and infrastructural disparities in shaping disease risk. Informal settlements, inadequate sanitation, and occupational exposure are common features across many endemic regions. Addressing these underlying determinants requires coordinated efforts that integrate health with broader development agendas, including housing, water and sanitation, and labour protection.

One health

The findings of this study strongly support the relevance of a One Health approach to leptospirosis control. The disease is inherently zoonotic, with transmission occurring at the interface of human, animal, and environmental systems. The observed spatial clustering and hotspot patterns likely reflect localised ecological conditions, including the presence of reservoir hosts, environmental contamination, and human exposure pathways.

Integrating human health surveillance with animal and environmental monitoring systems is therefore essential for a more comprehensive understanding of transmission dynamics. Surveillance of rodent populations, livestock, and other potential reservoirs could provide valuable information on infection sources and transmission pathways. Environmental monitoring, including water quality and soil contamination assessments, could further inform risk mapping and targeted interventions.

Multi-agency collaboration is critical in this context. Effective control of leptospirosis requires coordination between public health authorities, veterinary services, environmental agencies, and local government bodies. For example, urban sanitation programmes aimed at reducing rodent populations require collaboration between municipal authorities and health departments. Similarly, agricultural interventions may involve engagement with veterinary services and farming communities.

Operationalising a One Health approach also involves integrating data systems across sectors. Shared platforms that combine human case data with environmental and animal surveillance data could enable more precise identification of high-risk areas and transmission pathways. This would support more targeted and effective interventions, particularly in hotspot districts.

Capacity building is another key consideration. Training programmes that promote interdisciplinary collaboration and understanding of zoonotic disease dynamics could strengthen the implementation of One Health strategies. This includes not only technical training but also institutional mechanisms that facilitate coordination across traditionally separate sectors.

Planetary health

The broader environmental context of leptospirosis transmission highlights the relevance of a planetary health perspective. The findings of this study suggest that environmental changes, including those associated with climate variability, urbanisation, and land use transformation, play a significant role in shaping disease patterns. Although climatic variables did not demonstrate strong independent associations in the final model, the observed spatiotemporal patterns and clustering suggest that environmental processes remain central to transmission dynamics.

Urbanisation, particularly when accompanied by inadequate infrastructure, may increase exposure to contaminated environments. Informal settlements with poor drainage and sanitation create conditions that facilitate the persistence and spread of Leptospira. Similarly, deforestation and land use changes may alter the habitats of reservoir species, bringing them into closer contact with human populations.

Flooding, while not independently associated with incidence in this analysis, remains an important contextual factor that can amplify exposure pathways. The broader impacts of climate change, including increased frequency and intensity of extreme weather events, may therefore influence future disease patterns. These considerations underscore the need for climate-resilient health strategies that anticipate and adapt to changing environmental conditions.

Ecosystem-based approaches to disease prevention should be prioritised. This includes improving water and sanitation infrastructure, promoting sustainable land use practices, and enhancing environmental management to reduce contamination risks. Urban planning that incorporates health considerations, such as effective drainage systems and waste management, can also play a critical role in reducing exposure.

Importantly, planetary health emphasises the interconnectedness of human health and environmental sustainability. Addressing leptospirosis therefore requires not only disease-specific interventions but also broader efforts to create healthier environments. This includes policies that mitigate environmental degradation, reduce climate vulnerability, and promote sustainable development.

Conclusions

This study provides a comprehensive national assessment of leptospirosis in Malaysia between 2012 and 2022, demonstrating that the disease remains widely distributed with clear demographic, spatial, and temporal patterns. The burden of disease was concentrated among working-age individuals, with males consistently representing the majority of cases, indicating that exposure-related factors play a central role in transmission. Marked spatial heterogeneity was observed, with persistent clustering and identifiable hotspot areas, particularly in districts within Kelantan, Terengganu, and Sarawak. These hotspots largely overlapped with areas of elevated cumulative incidence and mortality, suggesting that certain geographic areas experience sustained transmission and potentially greater disease severity. Temporal variation was evident, with peak incidence occurring between 2014 and 2016, followed by fluctuations in subsequent years, indicating dynamic transmission patterns over time. Climatic drivers demonstrated heterogeneous associations with yearly state-level leptospirosis incidence. Temperature emerged as the only consistent predictor, with lower temperature levels associated with higher incidence. In contrast, rainfall, humidity, and flooding were not significantly associated after adjustment, suggesting that their effects may be mediated through more complex environmental or hydrological processes rather than direct linear relationships. Overall, the findings indicate that leptospirosis in Malaysia is shaped by a combination of demographic vulnerability, spatial clustering, and environmental context. The identification of high-risk populations and persistent hotspot areas provides important evidence to support targeted public health interventions. Strengthening surveillance systems, incorporating spatial and environmental indicators, and prioritising prevention efforts in high-burden areas are essential steps to improve disease control.

Acknowledgments

The authors would like to acknowledge the Director General of Health Malaysia and the Ministry of Health Malaysia for granting access to the leptospirosis surveillance data through the national electronic communicable disease notification system. Appreciation is also extended to the DID Malaysia and the DOE Malaysia for providing climatic data and to the DOSM for providing state-level population data used in this study. The contributions of all personnel involved in data collection, management, and reporting at the national and state levels are gratefully acknowledged. An abstract based on this study was presented as an oral presentation at the APRU Global Health Conference 2025, held from 28 to 31 October 2025, and was subsequently published as abstract AB121 in the Journal of Public Health and Emergency (JPHE): https://jphe.amegroups.org/article/view/11334/html.

Footnote

Provenance and Peer Review: This article was commissioned by the Guest Editor (Mellissa Withers) for the series “Towards Planetary Health Equity: A Global Call for Shared Solutions” published in the Journal of Public Health and Emergency. The article has undergone external peer review.

Reporting Checklist: The authors have completed the RECORD reporting checklist. Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0038/rc

Data Sharing Statement: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0038/dss

Peer Review File: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0038/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jphe.amegroups.com/article/view/10.21037/jphe-2026-0038/coif). The series “Towards Planetary Health Equity: A Global Call for Shared Solutions” was commissioned by the editorial office without any funding or sponsorship. The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study received ethical approval from the Medical Research and Ethics Committee (MREC), Ministry of Health Malaysia, with the ethical identification number NMRR ID-24-01790-LOK. Individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Costa F, Hagan JE, Calcagno J, et al. Global Morbidity and Mortality of Leptospirosis: A Systematic Review. PLoS Negl Trop Dis 2015;9:e0003898. [Crossref] [PubMed]
2. Adler B, de la Peña Moctezuma A. Leptospira and leptospirosis. Vet Microbiol 2010;140:287-96. [Crossref] [PubMed]
3. World Health Organization, International Leptospirosis Society. Human Leptospirosis: Guidance for Diagnosis, Surveillance and Control. Geneva: World Health Organization; 2003.
4. Bharti AR, Nally JE, Ricaldi JN, et al. Leptospirosis: a zoonotic disease of global importance. Lancet Infect Dis 2003;3:757-71. [Crossref] [PubMed]
5. Levett PN. Leptospirosis. Clin Microbiol Rev 2001;14:296-326. [Crossref] [PubMed]
6. Lau CL, Smythe LD, Craig SB, et al. Climate change, flooding, urbanisation and leptospirosis: fuelling the fire?. Trans R Soc Trop Med Hyg 2010;104:631-8. [Crossref] [PubMed]
7. Desvars A, Cardinale E, Michault A. Animal leptospirosis in small tropical areas. Epidemiol Infect 2011;139:167-88. [Crossref] [PubMed]
8. Philip N, Ahmed K. Leptospirosis in Malaysia: current status, insights, and future prospects. J Physiol Anthropol 2023;42:30. [Crossref] [PubMed]
9. Prevention and Control of Infectious Diseases Act 1988 (Act 342). 1988.
10. Mohd Radi MF, Hashim JH, Jaafar MH, et al. Leptospirosis Outbreak After the 2014 Major Flooding Event in Kelantan, Malaysia: A Spatial-Temporal Analysis. Am J Trop Med Hyg 2018;98:1281-95. [Crossref] [PubMed]
11. Sullivan KM, Dean A, Soe MM. OpenEpi: a web-based epidemiologic and statistical calculator for public health. Public Health Rep 2009;124:471-4. [Crossref] [PubMed]
12. Ministry of Health Malaysia. Guidelines for the Diagnosis, Management, Prevention and Control of Leptospirosis in Malaysia. Putrajaya: Ministry of Health Malaysia; 2011.
13. Center For International Earth Science Information Network-CIESIN-Columbia University. Gridded Population of the World, Version 4 (GPWv4): Population Count, Revision 11. Version 4.11 ed. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC); 2018.
14. Donovan TM, Welden CW. Spreadsheet Exercises in Ecology and Evolution. Sunderland, MA: Sinauer Associates; 2002.
15. Gotelli NJ. A Primer of Ecology. 4th ed. Sunderland, MA: Sinauer Associates; 2008.
16. Keyfitz N, Caswell H. Applied Mathematical Demography. 3rd edition. New York: Springer; 2005.
17. Malthus TR. An Essay on the Principle of Population. London: J. Johnson; 1798.
18. Vermote E. MOD09A1 MODIS/Terra Surface Reflectance 8-Day L3 Global 500m SIN Grid. Version 6.1 ed. Sioux Falls, SD: NASA EOSDIS Land Processes Distributed Active Archive Center (LP DAAC); 2021.
19. Ledien J, Sorn S, Hem S, et al. Assessing the performance of remotely-sensed flooding indicators and their potential contribution to early warning for leptospirosis in Cambodia. PLoS One 2017;12:e0181044. [Crossref] [PubMed]
20. Xu H. Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery. International Journal of Remote Sensing 2006;27:3025-33.
21. Little RJA. A test of missing completely at random for multivariate data. Journal of the American Statistical Association 1988;83:1198-202.
22. van Buuren S. Flexible Imputation of Missing Data. Boca Raton: CRC Press; 2018.
23. Department of Statistics Malaysia. Current population estimates, Malaysia, 2024. Putrajaya: Department of Statistics Malaysia; 2024.
24. Comber A, Brunsdon C. An Introduction to R for Spatial Analysis and Mapping. London: SAGE Publications; 2018.
25. Gatrell AC, Bailey TC, Diggle PJ, et al. Spatial point pattern analysis and its application in geographical epidemiology. Transactions of the Institute of British Geographers 1996;21:256-74.
26. Baddeley A, Turner R, Mateu J, et al. Hybrids of Gibbs point process models and their implementation. Journal of Statistical Software 2013;55:1-43.
27. Clark PJ, Evans FC. Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 1954;35:445-53.
28. Evans JS, Murphy MA, Ram K. spatialEco: Spatial Analysis and Modelling Utilities. 2019. Available online: https://cran.r-project.org/package=spatialEco
29. Tobler W. On the first law of geography: A reply. Annals of the Association of American Geographers 2004;94:304-10.
30. Elhorst JP. Specification and estimation of spatial panel data models. International Regional Science Review 2003;26:244-68.
31. Anselin L. Local indicators of spatial association—LISA. Geographical Analysis 1995;27:93-115.
32. Peeters A, Zude M, Käthner J, et al. Getis–Ord’s hot- and cold-spot statistics as a basis for multivariate spatial clustering of orchard tree data. Computers and Electronics in Agriculture 2015;111:140-50.
33. Brooks ME, Kristensen K, van Benthem KJ. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R. Journal 2017;9:378-400.
34. Zuur AF, Ieno EN, Walker NJ. Mixed Effects Models and Extensions in Ecology with R. New York: Springer; 2009.
35. Joshi YP, Kim EH, Cheong HK. The influence of climatic factors on the development of hemorrhagic fever with renal syndrome and leptospirosis during the peak season in Korea: an ecologic study. BMC Infect Dis 2017;17:406. [Crossref] [PubMed]
36. Rees EM, Lotto Batista M, Kama M, et al. Quantifying the relationship between climatic indicators and leptospirosis incidence in Fiji: A modelling study. PLOS Glob Public Health 2023;3:e0002400. [Crossref] [PubMed]
37. Sohail ML, Khan MS, Ijaz M, et al. Seroprevalence and risk factor analysis of human leptospirosis in distinct climatic regions of Pakistan. Acta Trop 2018;181:79-83. [Crossref] [PubMed]
38. Benacer D, Thong KL, Min NC, et al. Epidemiology of human leptospirosis in Malaysia, 2004-2012. Acta Trop 2016;157:162-8. [Crossref] [PubMed]
39. Obels I, Mughini-Gras L, Maas M, et al. Increased incidence of human leptospirosis and the effect of temperature and precipitation, the Netherlands, 2005 to 2023. Euro Surveill 2025;30:2400611. [Crossref] [PubMed]
40. Meny P, Menéndez C, Ashfield N, et al. Seroprevalence of leptospirosis in human groups at risk due to environmental, labor or social conditions. Rev Argent Microbiol 2019;51:324-33. [Crossref] [PubMed]
41. Dhewantara PW, Riandi MU, Wahono T. Effect of climate change on the geographical distribution of leptospirosis risk in western Java, Indonesia. IOP Conference Series: Earth and Environmental Science. 2022;1089:012074.
42. Chadsuthi S, Chalvet-Monfray K, Geawduanglek S, et al. Spatial-temporal patterns and risk factors for human leptospirosis in Thailand, 2012-2018. Sci Rep 2022;12:5066. [Crossref] [PubMed]
43. Ab Kadir MA, Abdul Manaf R, Mokhtar SA, et al. Identifying leptospirosis hotspots in Selangor: uncovering climatic connections using remote sensing and developing a predictive model. PeerJ 2025;13:e18851. [Crossref] [PubMed]
44. Maskey M, Shastri JS, Saraswathi K, et al. Leptospirosis in Mumbai: post-deluge outbreak 2005. Indian J Med Microbiol 2006;24:337-8. [Crossref] [PubMed]
45. Amilasan AS, Ujiie M, Suzuki M, et al. Outbreak of leptospirosis after flood, the Philippines, 2009. Emerg Infect Dis 2012;18:91-4. [Crossref] [PubMed]
46. Vanasco NB, Schmeling MF, Lottersberger J, et al. Clinical characteristics and risk factors of human leptospirosis in Argentina (1999-2005). Acta Trop 2008;107:255-8. [Crossref] [PubMed]
47. Jayaramu V, Zulkafli Z, De Stercke S, et al. Leptospirosis modelling using hydrometeorological indices and random forest machine learning. Int J Biometeorol 2023;67:423-37. [Crossref] [PubMed]
48. Lea JSX, Reduan MFH, Choong SS, et al. Leptospirosis in humans and animals in Malaysia: A review from 1976 to 2023. Vet World 2025;18:673-85. [Crossref] [PubMed]
49. Saputra YA, Armawan LVA, Lisa M, et al. The Role of Hydrometeorological Factors in Leptospirosis Transmission in Central Java, Indonesia. J Prev Med Public Health 2025;58:553-62. [Crossref] [PubMed]
50. Matsushita N, Ng CFS, Kim Y, et al. The non-linear and lagged short-term relationship between rainfall and leptospirosis and the intermediate role of floods in the Philippines. PLoS Negl Trop Dis 2018;12:e0006331. [Crossref] [PubMed]
51. Hacker KP, Sacramento GA, Cruz JS, et al. Influence of Rainfall on Leptospira Infection and Disease in a Tropical Urban Setting, Brazil. Emerg Infect Dis 2020;26:311-4. [Crossref] [PubMed]
52. Azman NA, Rasudin NS, Atan EH, et al. The association of climatic pattern and leptospirosis cases in Malaysia from 2013 to 2021. Malays J Pathol 2025;47:77-84. [PubMed]
53. Storck CH, Postic D, Lamaury I, et al. Changes in epidemiology of leptospirosis in 2003--2004, a two El Niño Southern Oscillation period, Guadeloupe archipelago, French West Indies. Epidemiol Infect 2008;136:1407-15. [Crossref] [PubMed]