Introduction

Leptospirosis is a growing global health threat. The zoonotic disease, caused by Leptospira spirochetes, is transmitted to humans through direct exposure to infected animals or more commonly through contact with soil or water contaminated with urine from infected mammals. Globally, there are over one million yearly cases of leptospirosis and approximately 60,000 deaths [1]. While Asian and Oceanic nations have reported the highest incidence of the disease to date, leptospirosis has become the most widespread and prevalent bacterial zoonotic disease globally, and its epidemiological reach spans tropical and subtropical areas, including many regions with limited laboratory diagnostics or well-established surveillance infrastructure [1,2].

The diagnosis of leptospirosis is challenging given its varied clinical presentations, ranging from asymptomatic or mild febrile infections to severe pulmonary hemorrhage and multi-organ failure, which overlap with the symptomatology of many tropical and subtropical diseases including dengue, chikungunya, typhoid, rickettsioses, and malaria. The accurate diagnosis relies on high clinical suspicion coupled with laboratory evidence of infection. While molecular tests have excellent test characteristics for the diagnosis of infection early in the disease course, they are costly and not widely available in many endemic settings including in the Dominican Republic, the location of the present study. The reference standard serologic assay is the microscopic agglutination test (MAT), that detects antibodies against leptospiral lipopolysaccharide. Anti-leptospira antibodies are generated within one week of infection and persist for up to eight years. While the utility of MAT is limited for the diagnosis of acute leptospirosis due to poor sensitivity early in the disease course, MAT provides valuable data on the presence of leptospira antibodies for population-based serological surveys [3,4].

Given the overlapping clinical features of leptospirosis and other common diseases, together with limited access to confirmatory testing, the burden and epidemiology of leptospirosis is poorly characterized in the Dominican Republic and the Caribbean more broadly [5]. Yet, despite this, the Caribbean region is known to have widespread endemic disease, has experienced numerous outbreaks, and reports among the highest estimated disease-specific mortality rates globally [1]. Regional risk factors for leptospirosis transmission include intensive farming and household livestock, as well as frequent extreme climactic events with excess precipitation and flooding, which are exacerbated by climate change and expected to increase [6,7]. In the present study, to inform public health planning and control efforts, we aimed to characterize the seroepidemiology of human leptospira infection among populations in two provinces in the Dominican Republic.

Methods

Results

Of 2,124 participants enrolled in the two study provinces during the national serosurvey, 2,091 samples (98.4%) were tested for evidence of prior leptospirosis infection using MAT and comprise the current study population. The median age was 39 years (IQR 22, 56) and 1,339 were female (64.0%). Demographic details are listed in Table 1. A total of 237 study participants registered agglutinating antibodies, assessed by the presence of MAT titers ≥1:100, for an overall unadjusted seroprevalence of 11.3% (95% confidence interval: 10.0–12.7%) and an adjusted seroprevalence of 11.3% (10.8–13.0%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of study population and Leptospira seroprevalence by microscopic agglutination test, Espaillat and San Pedro de Macoris Provinces, Dominican Republic, 2021.

https://doi.org/10.1371/journal.pntd.0012463.t001

Serocatalytic models estimated the annual attack rate in the Dominican Republic to be 0.30% (0.27–0.35%) using the catalytic model (model 1) which assumes a constant FOI and no waning immunity, and 1.00% (0.70%–1.69%) for the reverse catalytic model (model 2a) which assumes a constant FOI whilst allowing for waning immunity. The catalytic model (model 1) fitted the data better compared with the reverse catalytic model (model 2a), with a DIC difference of 31. A third model, which was the reverse catalytic model but assumed there was no transmission in the six years prior to the serosurvey (model 2b), fitted the data better compared with model 2a (DIC difference 16), with an estimated annual attack rate of 1.19% (0.90–0.17%) (S2 Table and S1 Fig). Using data from these models, we estimated the number of annual leptospirosis infections and assessed the gap between reported cases and serologically inferred infections. Using the lowest estimated FOI (0.30%) applied to the population of the Dominican Republic (~11 million), we infer about 33,000 infections per year which is approximately 145-fold higher than the average number of nationally reported cases between 2013 and 2022 (226 cases per year) (S3 Table). If we consider the highest estimated FOI (1.19%), the ratio of reported to serologically estimated cases is approximately 1:580. While the spatial heterogeneity of leptospirosis transmission necessitates caution when extrapolating from two provinces to the entire country, when considering only the two study provinces (population ~680,000) we estimate 2,040 annual leptospirosis cases assuming a FOI of 0.30% and 8,092 cases assuming a FOI of 1.19%. In turn, this suggests that there may be between nine and 36-fold more leptospirosis infections in just the two study provinces than reported cases across the entire country.

Seropositive results were reported for 16 of the 20 serovars included in the MAT panel, which included 14 of 17 serogroups. Among the 237 seropositive individuals, 91 (38.4%) registered the highest titers to L. interrogans serogroup Icterohaemorrhagiae, 38 to Australis (16.0%), 25 to Canicola (10.5%), 20 to Djasiman (8.4%), and 12 to L. santarosai serogroup Pyrogenes (5.1%) (Fig 1 and Table 2 and S1 Table). The four most common serogroups accounted for 73.4% of seropositive results. Among 27 (11.4%) individuals that registered highest titers against two or more serogroups (i.e., Mixed serogroups), the mostly commonly reacting serogroups were Icterohaemorrhagiae (23, 85.2%), Canicola (14, 51.9%), and Australis (12, 28.6%). Most seropositive individuals registered titers of 1:100 (N = 119, 50.2%), 1:200 (49, 20.7%), 1:400 (44, 18.6%), and 1:800 (14, 5.9%), with 11 (4.6%) registering titers of 1:3200 or higher (Fig 1). The distributions of titers when stratified by serogroup were largely similar (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Distribution of Leptospira serogroups by highest titer, age and study setting, Dominican Republic, 2021.

(A) Bar plot indicates the number of participants stratified by the highest MAT titer and serogroup. (B) Ridge plot shows the distribution of MAT titers for each serogroup. (C) Stacked bar plot indicates the distribution of serogroups by age group. (D) Distribution of serogroups stratified by urban vs rural setting and Northwest (Espaillat) vs Southeast (San Pedro de Macoris) regions. Serogroup is defined as the serogroup that registered the highest MAT titer for each seropositive study participant. Individuals that registered the same highest titer to two or more serogroups are categorized as ‘Mixed.’ N = 237 for all plots. Serogroups aggregated as ‘Other’ are those with counts less than 10, with details listed in S1 Table.

https://doi.org/10.1371/journal.pntd.0012463.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Prevalence of positive Leptospira microscopic agglutination tests by primary reacting serogroup, Espaillat and San Pedro de Macoris Provinces, Dominican Republic, July-Oct 2021.

https://doi.org/10.1371/journal.pntd.0012463.t002

Analysis of leptospirosis seroprevalence across demographic and environmental factors identified distinct trends (Fig 2 and S4 Table). Increasing age was associated with higher seropositivity, with the seroprevalence of 2.8% among those 5–19 years, 13.7% among those 35–49 years, and 19.3% among those 65 years and older. Similarly, multivariable adjusted odds ratios demonstrated a sharp increase in seroprevalence during early adulthood, followed by a plateau in middle age and then a subsequent increase in older age (Fig 2). Seropositivity was higher among males (17.8%) than females (7.7%) (adjusted Odds Ratio 2.41 [95% confidence interval 1.79–3.25]), among residents of the Northwest (15.7%) versus Southeast Region (8.6%) (aOR 1.84, 1.31–2.58), and among those reporting frequently seeing rats in the household (aOR 1.64 [1.06–2.52]). While being a farmer was associated with high seroprevalence (35.1%) with an unadjusted odd ratio of 4.67 (2.80–7.63) vs non-professionals, no association was identified on multivariable analysis (aOR 1.63 [0.94–2.80]). Similarly, residing in a rural vs urban setting was associated with higher odds of seropositivity on univariable (OR 1.61 [1.22–2.11]) but not multivariable analysis. Residing in a barrio (settlement) and indoor versus outdoor work were not associated with leptospirosis serostatus.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Risk factors for Leptospira seropositivity by microscopic agglutination test in the Dominican Republic, 2021.

Multivariable logistic regression with all modeled covariates shown. Serogroups Australis and Canicola not presented given limited number of seropositive individuals (N = 38 and 25 respectively). Full model outputs including univariable odds ratios and model metrics shown in S4 and S5 Tables. SPM, San Pedro de Macoris (Southeast study region). Espaillat (Northwest study region).

https://doi.org/10.1371/journal.pntd.0012463.g002

Given the distinct host and ecological features of different Leptospira serogroups, we examined serogroup specific demographic and environmental risks. Analyzing serogroup-specific seroprevalence for the most common serogroups (Icterohaemorrhagiae, Australis, and Canicola) identified similar overall seroprevalence trends across demographic and environmental factors (Fig 2 and S4–S8 Tables). However, there were exceptions. Males registered higher seropositivity (vs females) against serogroup Icterohaemorrhagiae (aOR 2.69 [1.71–4.26]) but not Australis (aOR 1.14 [0.51–2.41]) or Canicola (aOR 1.44 [0.59–3.34]). Serogroup Australis was associated with higher seroprevalence among farmers (aOR 5.55 [2.03–14.98)] versus the reference group, but not Icterohaemorrhagiae or Canicola. Icterohaemorrhagiae and Canicola serogroups were both associated with higher seroprevalence in the Northwest study region (aOR 2.82 [1.65–4.99] and 3.24 [1.23–9.37] respectively), but not Australis. Lastly, Icterohaemorrhagiae and Canicola seroprevalence was higher in rural vs urban setting (aOR 1.62 [1.02–2.61] and 2.1 [1.0–4.7] respectively) but trended lower for Canicola (aOR 0.57 [0.23–1.35]). When excluding individuals that were seropositive to serogroups other than the serogroup of interest (e.g., when assessing risk factors for serogroup Icterohaemorrhagiae, all seropositive individuals with highest titer to non-Icterohaemorrhagiae serogroups were excluded from the analysis) risk factors and odds ratios were essentially unchanged.

Given the unusual age distribution of seroprevalence, with the highest rates among the older population rather than the common pattern of peaking in middle age and then either plateauing or declining in older age [19,20], we examined the age distribution in more detail, stratifying by study region, gender, and urban vs. rural setting. We found that the trend of increasing seroprevalence with age was consistent across all variables, except for female gender, which showed a biphasic curve with peaks in middle age and again in older age (Fig 3). Similar trends by age group were observed when examining gender-specific seroprevalence in the Northwest vs. the Southeast region (S2 Fig). To further assess this finding, we used GAMs to model seroprevalence by age and gender, while accounting for documented or potential confounding covariates such as study region, work environment, residence in a barrio (settlement), and urban vs. rural residence. The model aligned with the unadjusted analyses showing females with a biphasic age-associated seroprevalence distribution and males with a near-linear increase in seroprevalence with age (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Age-stratified seroprevalence by region, gender and urban vs. rural setting.

Individual plots show seroprevalence point estimates (black dots) and 95% confidence intervals (black vertical lines) by five-year age group. A smoothed line is fitted to the data using generalized additive models with a cubic spline and weighted for number of observations. Colored ribbons represent the 95% CI. Number of observations per plot listed in Table 1. Southeast study region (San Pedro de Macoris province). Northwest study region (Espaillat province). Additional plots showing age stratified seroprevalence by both gender and study region are presented in S2 Fig.

https://doi.org/10.1371/journal.pntd.0012463.g003

Discussion

This multistage household survey based on a carefully sampled population across two discrete study regions provides new insights into the epidemiology of human Leptospira infection in the Dominican Republic. Our findings suggest a wide diversity of circulating serogroups, regional variation in transmission, and a marked difference between national reported cases and infections.

Discerning the true extent of leptospirosis infections is challenging, particularly in most endemic setting where access to testing is limited. Accordingly, population-based serological studies are valuable for characterizing the burden of Leptospira infections. In this study we used serological data to estimate the annual attack rate and in turn the difference between nationally reported leptospirosis cases and serologically estimated infections. Our findings suggest marked under ascertainment with 100-fold or greater difference in reported clinical cases versus estimated infections. While our estimates vary based on the modeling approach it appears clear that only a small fraction of infections are identified through existing reporting mechanisms. The application of serocatalytic methods to estimate leptospirosis FOI and, in turn, under ascertainment has only been reported once [3]. In that study, based on data from a national serological survey in Fiji, the ratio of reported cases to estimated infections ranged from about 1:28 to 1:140. While surveillance infrastructure and criteria for reporting leptospirosis cases varies markedly between countries, thus making direct comparisons challenging, our findings support the concept that under-ascertainment of Leptospira infections in many endemic settings is likely on the order of magnitudes. The reason for such marked discrepancy between reported cases and estimated infections is likely due to a range of factors including subclinical infections or mild infections that do not seek medical care, limited access to healthcare or leptospirosis testing, or gaps in clinician knowledge about the clinical features of leptospirosis precluding suspicion of leptospirosis as the infecting pathogen.

While Icterohaemorrhagiae, Australis, Canicola, and Djasiman were collectively implicated as the infecting serogroups in almost three quarters of cases, our study suggests high diversity with 14 of the 17 serogroups included in our panel identified as the primary infecting serogroup, similar to Trinidad where 17 serovars have been isolated from humans [21]. This is in contrast to some other island settings such as Barbados where only three serovars of Leptospira interrogans have been identified, or Fiji where a single serovar (Pohnpei) was implicated in over 80% of seropositive cases identified through a national serological survey [21,22]. We examined both the overall epidemiology of human Leptospira infection and the epidemiology of the three most common serogroups: Icterohaemorrhagiae, Australis, and Canicola. Icterohaemorrhagiae, primarily transmitted by exposure to rat urine, was the most commonly identified serogroup. Given Icterohaemorrhagiae is considered to cause the most severe clinical disease, including Weil’s Disease which is characterized by hepatic and renal failure and high case-fatality rates, it is of particular public health importance [23]. The primary reservoirs for Icterohaemorrhagiae are the Rattus norvegicus (or brown rat) and Rattus rattus (Black rat), while other rodents including mice and voles may be involved in transmission. Australis and Canicola serogroups are usually associated with milder clinical disease, although severe disease and deaths have been reported. The primary reservoirs for the Australis serogroup include various wild and domestic animals, such as farm animals, rats, and mice. Dogs are the major recognized reservoir for the Canicola serogroup, although livestock and wild mammal are other potential reservoirs [24].

Similar to studies on the epidemiology of leptospirosis from other settings, we confirmed that being male, being exposed to rats in the home, and older age were associated with higher seroprevalence [19]. We found that the prevalence of Leptospira antibodies increased through middle age and then plateaued, a typical finding that reflects the increased occupational risk of infection among young and middle aged adults. However, we also observed an increase in the oldest age groups, particularly among males, a somewhat unusual trend given reported infection rates generally decline among older age groups [1], with seroprevalence typically remaining flat or declining during late adulthood [19]. Given antibodies to Leptospira decay over time and are thought to be detectable for only about eight years post-infection [3,25], the older population would be anticipated to register declining seroprevalence in the absence of ongoing exposures. The reason that this was not the case suggests unexpectedly high ongoing transmission among older adults, although why age-specific transmission would be different in this versus other settings is unclear. We further explored age-specific seroprevalence stratifying by gender, study region and urban vs rural setting and found interesting differences, particularly between males and females. Males demonstrated a near linear increase in seroprevalence with age while females demonstrated a distinct biphasic pattern with peaks around early middle-age and again after 70 years of age. The reason for these differences by gender is unclear and the few studies that have reported Leptospira seroprevalence stratified by both age and gender (or sex) contrast with our findings, although those studies were either limited by size [26] or from markedly different settings [19]. We also found differences by study region, with seroprevalence in the Northwest Region (Espaillat) about 1.8-fold higher than the Southeast Region (San Pedro de Macoris) even after adjusting for potential confounders, highlighting distinct spatial heterogeneity in leptospirosis transmission. However, there are nuances to these findings, and there were differences in serogroup-specific risks that were not identified when assessing risks aggregated across all serogroups. Given differences in mammal reservoirs across serogroups, and therefore differences in exposure risks and transmission pathways, this may be anticipated. For example, adjusted serogroup Australis seroprevalence was about five-fold higher among farmers than the reference group, but not higher for Icterohaemorrhagiae or Canicola. This likely reflects the fact that livestock is a common reservoir for the Australis serogroup but less so for Icterohaemorrhagiae and Canicola. Similarly, the adjusted odds of being seropositive for Icterohaemorrhagiae was almost three-fold higher for males than females, while no gender differences were identified for serogroups Australis and Canicola. These data highlight how serogroup-specific risks may be leveraged to prioritize targeted public health measures such as rodent control or livestock and dog vaccination [27].

While our study has multiple strengths including a relatively large population selected using a multilevel sampling approach, testing of samples using the reference-standard serological assay, and sampling across two discrete regions to understand geographic variations, there are some limitations. First, the geographic scope is limited to two out of 34 provinces, potentially restricting the generalizability of our findings across other regions of the country. Second, cross-reactivity may affect the accuracy of our serogroup-specific analyses. However, when we excluded seropositive individuals with highest titer against a serogroup other than the one being assessed, findings were essentially unchanged. Irrespective, given well established limitations of MAT for definitely identifying the infecting serogroup, and given we were reporting putative serogroups based on the highest reacting titer, our serogroup-specific findings should be considered as suggestive rather than definitive. Third, while some studies use an MAT titer cutoff of 1:50, we used a titer cutoff of 1:100 and therefore our estimated seroprevalence was likely lower than if a lower cutoff was used. Fourth, self-reported exposures and socio-demographics factors enumerated through our survey may be subject to recall or social desirability biases. Fifth, serogroup specific analyses were limited by the number of seropositive individuals, particularly for serogroups Australis and Canicola. Lastly, this was a cross-sectional study and therefore we were not able to capture temporal transmission dynamics.

In conclusion, this seroepidemiological study provides insights into the epidemiology and under-reported burden of human Leptospira infection in the Dominican Republic. We identified a range of serogroups, geographic variation in infection and risk factors, and a substantial discrepancy between reported cases and likely infections. While seroprevalence is high, risks varied markedly by age, gender, and study region. Analyzing serogroups individually suggested distinct transmission patterns, emphasizing the importance of understanding transmission pathways to ensure targeted and effective public health interventions. Our findings highlight the complex epidemiology of leptospirosis in the Dominican Republic, underlining the need for more granular investigation into environmental factors, animal reservoirs, and serovar-specific dynamics to effectively control this pathogen.

Supporting information

Acknowledgments

We would like to thank the many study participants that volunteered to participate in this study. We would also like to thank the study staff that collected the field data, the Dominican Republic Ministry of Health and Social Assistance, and the Pedro Henriquez Ureña National University, for their commitment and support for the study. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US CDC.