Leptospirosis in humans and selected animals in Sub-Saharan Africa, 2014–2022: a systematic review and meta-analysis

The methodology of this systematic review and meta-analysis was guided by the Centers for Reviews and Dissemination guidelines [18], and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. The review was registered in the PROSPERO International Prospective Register of Systematic Reviews on 14^th July 2022 (CRD42022345844).

We included peer-reviewed studies, both observational and experimental studies that reported on the seroprevalence or had sufficient data to estimate the seroprevalence of leptospirosis among humans, cattle, goats, and rodents across the SSA region. The inclusion and exclusion criteria are outlined below.

A comprehensive search was conducted in the following databases for articles published on leptospirosis and Leptospiral antibodies among humans and animals in SSA between 01 January 2014 and 30 August 2022: Scopus, PubMed, Google Scholar, CINAHL, Web of Science, and African Journals Online. A search strategy that employs medical subject headings (MESH) and keywords were developed and used while searching for literature. We categorized the search terms according to geographic location (SSA); participants (humans and animals), and outcome of interest (seroprevalence of leptospirosis). The final search strategy for PubMed was reported (Supp Table 1). The search terms used in PubMed were adopted and used in other databases. Furthermore, a search of the reference lists of the eligible papers was conducted to obtain other relevant articles.

Identified articles were uploaded into EndNote to remove duplicates. The articles were then screened based on title and abstract for eligibility by one author (JMG). The full texts of studies selected after screening were retrieved and screened to verify their conformance with the inclusion criteria. This process of screening was conducted by one author (JMG) and reviewed by the second author (LM).

For all included studies, data were extracted using a customized data abstraction tool designed for this study. The following information was extracted:

Where data such as sample size and number of human participants or animals that were seropositive were provided, the seroprevalence estimates were calculated using this data.

Data were analyzed using the meta package in R software (version 3.6.1). Forest plots were drawn to visualize the pooled seroprevalence and the 95% confidence intervals (95% CI) of leptospirosis in humans and animals in SSA. A random effects meta-analysis model was used to pool the seroprevalence data [56]. Heterogeneity was assessed using the χ2 test on Cochrane’s Q statistic [57] and the I² (values of 25%, 50%, and 75% representing low, medium, and high heterogeneity, respectively) [58]. Subgroup analysis based on the SSA region, diagnostic methods, and study setting (rural/urban), was done for studies that involved human participants. As a result of fewer animal studies, the pooled prevalence estimates of leptospirosis were categorized only based on the diagnostic criteria used and the study setting (rural vs urban setting).

The included articles were evaluated for methodological quality using a 10-item scale developed by Hoy et al. [59] for internal and external validity, generalizability, and response rate. The risk of bias assessment was conducted by JMG and reviewed by LM. The results of risk of bias (ROB) were presented for each study (Table 1).

The funnel plots and Egger’s weighted regression methods [60] were used to assess publication bias and a p-value < 0.10 was considered indicative of statistically significant publication bias. Funnel plots are presented in the supplementary files.

We found 509 articles in the literature search. When duplicates (n = 141 articles) were removed, 368 articles remained for both abstract and full-text screening. In the title and abstract screening, 242 articles were excluded because they did not meet the inclusion criteria (Fig. 1). The full-texts of the remaining 126 articles were obtained and after full-text screening, 89 articles were excluded because they didn’t meet the inclusion criteria. Finally, 37 articles were included in the systematic review. To report the seroprevalence of leptospirosis, PCR and culture results reported by studies that used these techniques to test for leptospirosis were excluded while conducting the meta-analysis.

Thirty-seven articles from 14 countries out of 46 countries in SSA were included in this review. Most of the studies (n = 24, 64.9%) were conducted in the East Africa (EA) region particularly in Tanzania (n = 14) (Fig. 2). The studies conducted in Central Africa (CA), West Africa (WA), and Southern Africa (SA) regions were 5 (13.5%), 6 (16.2%), and 2 (5.4%) respectively (Table 1). All the included studies used a cross-sectional study design, with 20 (54.1%) studies conducted particularly in a rural setting, 4 (10.8%) studies in an urban setting, and 13 (35.1%) studies conducted in both urban and rural settings. The most used diagnostic method was the MAT (n = 21, 56.8%) used either alone or in combination with other methods such as ELISA or culture or PCR. The studies that used the ELISA method alone were 8 and the studies that used the PCR method alone were also 8 (Table 1). All of the included studies had a low ROB (n = 34, 91.9%) except 3 (8.1%) that had a moderate ROB.

Based on the ELISA method, the overall seroprevalence of leptospirosis across the regions in SSA was 12.7% (95% CI 7.5,20.8) with substantial heterogeneity between studies (I² = 96.0%, p < 0.01) (Fig. 3). Based on the funnel plot (supp Fig. 1) and Egger's test, there was symmetry and no evidence of potential publication bias. The EA region had the highest pooled seroprevalence compared to other regions in studies that used the ELISA method.

Based on the MAT method, the overall seroprevalence of leptospirosis was 15.1% (95% CI 9.4,23.5) with substantial heterogeneity between studies (I² = 97.0%, p < 0.01) (Fig. 4). Based on the funnel plot (supp Fig. 2) and Egger's test, there was symmetry and no evidence of potential publication bias. CA region had the highest pooled seroprevalence compared to other regions among studies that used MAT diagnostic method. Based on the PCR method, the overall seroprevalence of leptospirosis across the regions in SSA was 4.5% (95% CI 0.4, 35.6) (supp table 2). When studies were sub-grouped based on whether they were conducted in a rural or urban setting, the pooled seroprevalence of studies conducted in rural areas was higher than the pooled estimate for studies conducted in urban or a mixture of both urban and rural areas (supp Figs. 3 and 4).

Most of the studies that estimated the prevalence of Leptospira in cattle used MAT and ELISA methods, with only 2 studies and 1 study using PCR and culture methods respectively. Overall, the seroprevalence of leptospirosis in cattle was 29.2% (95% CI 16.1,46.9), 30.1% (95% CI 28.0,32.2), and 9.7% (95% CI 0.1,88.6) based on the ELISA, MAT, and PCR methods respectively (Fig. 5). The pooled seroprevalence estimated by the MAT method was higher compared to ELISA and PCR methods and the pooled seroprevalence differed significantly between the diagnostic methods (p < 0.01) (Fig. 5).

Most of the studies that estimated the seroprevalence in goats used the MAT diagnostic method, with only one study using the PCR method and no study using ELISA. The pooled seroprevalence estimates of Leptospira infection in goats among studies that used the MAT method was 30.0% (95% CI 1.1,94.0) with substantial heterogeneity among the studies (I² = 97.0%, p < 0.01) (Fig. 6).

Among rodents, most of the studies used the PCR method followed by the MAT method, only one study used culturing method and no study used the ELISA method. The pooled seroprevalence estimates of leptospirosis among the studies that used the MAT method was 21.0% (95% CI 15.6,27.7) with no heterogeneity among the pooled studies (p < 0.5). Among the studies that used the PCR method, the pooled seroprevalence estimate was 9.6% (95% CI 21.,34.3) with substantial heterogeneity (I² = 95.0%, p < 0.01) Fig. 7.

The data included in this meta-analysis to a large extent is a tip of an ice bag of leptospirosis morbidity in SSA and therefore it’s not conclusive. Several factors such as limited awareness and paucity of diagnostic facilities likely drive the issues of underreporting of Leptospira infection both in humans and animals. Other factors such as over-representation of certain countries or regions such as Tanzania may have contributed to reporting bias, particularly in the spatial distribution of the studies. This, therefore, necessitates that more studies on Leptospira infection in humans need to be conducted in CA, WA, and SA regions and some countries in the EA region to explicitly decipher the epidemiology of leptospirosis in SSA. In addition, the level of heterogeneity between the pooled studies was quite high in this review, a challenge common to meta-analyses of prevalence studies. Sub-group analysis based on SSA region (EA, WA, CA, and SA), and study setting (rural and urban) was conducted to ascertain the sources of heterogeneity among studies that involved human participants. However, the heterogeneity persisted, and it could largely be attributed to differences in study participants' characteristics and varying case definitions. Lastly, unpublished data or grey literature were not included in this review, hence some relevant unpublished/ grey literature may have been missed. The synthesized data from animal studies should also be interpreted with caution because most animals in the studies were sampled from an abattoir, and therefore this creates a selection bias since most animals for slaughter tend to be older, and fluctuation in leptospirosis occurrence based on the season of the year was not adjusted for because most studies often did not report this data. Notwithstanding the limitations, this meta-analysis provided a current synthesis of the prevalence of Leptospira infection in humans and animals based on diagnostic methods and regions in SSA.