Risk factors for Mycobacterium ulcerans infection (Buruli Ulcer) in Togo ─ a case-control study in Zio and Yoto districts of the maritime region

We conducted a case-control study in the Zio and Yoto districts of the maritime region (Fig. 1) between November 2014 and May 2015. Buruli ulcer cases were selected at the National Reference Center for BU Treatment (CNRT-UB) located at CHR Tsévié. Patients enrolled were recruited from March 2013 to May 2015. Controls were recruited by active search during the survey. Patients infected with the human immunodeficiency virus (HIV) or with active tuberculosis were excluded from the study.

A probable case of Buruli ulcer was defined as any person aged ≥2 years who lived in Zio or Yoto district showing clinical symptoms according to the WHO clinical definition of BU [3]. A confirmed case was defined as a probable case with detection of M. ulcerans using Ziehl-Neelsen (ZN) microscopy and IS2404 PCR [9, 10].

An eligible control was defined as any person aged ≥2 years without any history or clinical symptoms of Buruli ulcer. Up to two controls were randomly selected and matched to cases by sex and place of residence (home where lived the case or neighbor home in the same village).

This study was conducted in 17 villages in districts of Zio and Yoto where more than 85% of confirmed BU patients originate. These districts are in the Maritime Region (South of Togo) which covers an area of ​​6.359 km². With an estimated population of 1,762,518 inhabitants in 2012, the climate is tropical and humid with two rainy seasons and two dry seasons. The maritime region has a flat topography, with a low contrast characterized by a sedimentary basin that covers 4/5 of the region, a low altitude (50-80 m on average) and crossed by the depression of the Lama. The clay soil remains soggy and muddy in the rainy season. Water stagnates for several months in this region. The hydrographic network comprises 3 large rivers which are the Mono in the east, the Zio and the Haho in the center with several small tributaries that flow into the “lac Togo” (Fig. 1). All these streams have a low flow, closely linked to seasonal variations of precipitations [33].

Sample collection: Samples were collected according to standardized procedures as previously described [9, 10]. Briefly, fine needle aspirates (FNA) were collected from the center of non-ulcerative lesions or from undermined edges of ulcerative lesions including necrotic tissue. Swabs were collected by circling the entire undermined edges of ulcerative lesions. Samples taken were put in tubes containing cell lysis solution (CLS, Qiagen, Hilden, Germany) and sent for PCR analysis at INH.

Laboratory testing: Direct smears for microscopy were prepared from swab and FNA samples at peripheral care units or CNRT-UB and subjected to Ziehl-Neelsen staining for detection of acid fast bacilli at the laboratory of the CHR. Slides were analyzed by microscopy according to the WHO [34] recommended grading system. All slides were double checked at the INH by a second technician for external quality control.

All molecular analyses were conducted at the NRL for BU at INH as previously described [10]. For PCR analysis, DNA was extracted from FNA and swab samples with the Gentra Puregene DNA extraction kit (Qiagen) with minor modifications of the manufacturer’s protocol. The conventional IS2404-PCR with gel-based amplicon detection was applied using dry-reagent-based consumables (DRB-IS2404 PCR). Briefly, for the DRB-PCR, the primers MU5 (AGCGACCCCAGTGGATTGGT) and MU6 (CGGTGATCAAGCGTTCACGA) were lyophilized in reaction tubes. The Illustra PuReTaq Ready-To-Go PCR beads (GE Healthcare) containing Taq polymerase, dNTPs and Mg²⁺ were added and dissolved in water before adding DNA template. All PCR assays included negative extraction, positive, negative (no template) and inhibition controls. All inhibited samples were repeated after 10-fold dilution.

We used the power calculation tool of Epi-Info (version 7; 2012) to determine the sample size by setting α to 5% and power β to 80%. The health and population development survey (EDST; 2014) in Togo showed that 36,2% of households use water from unprotected sources [35, 36]. The odds ratio (OR) of the association between cases and controls was set at 2, yielding a sample size of 100 BU cases and 200 controls.

The survey was conducted by a team of four people including a clinician from CHR Tsévié, a focal point of the national program of BU surveillance, a community volunteer and a laboratory technician. Case residence was identified by the community volunteer. Once at home, we selected one or two matched control subjects. A well-structured questionnaire was administered to all selected participants (Additional file 1: Questionnaire form S1). For participants who could not respond in French, the interview was conducted in the local language. For children under 10 years, we interviewed their parents mainly for their activities and behavior. All the participants gave their consent prior to data collection on socio-demographic characteristics, behavior, occupational and environmental factors as well as administration of BCG vaccination.

Data collected were entered in a database designed through Epi-Info software (Version 7; 2012). Statistical analysis was carried out by SPSS software (Statistical Package for Social Science, Version 16.0, SPSS Inc. and Chicago, IL). Qualitative data were presented as number n (%) and quantitative data as mean ± standard deviation. Buruli ulcer was considered as the dependent variable and socio-demographic characteristics, occupational and environmental factors as independent variables. Student t-test was used for comparison of mean or median age and number of people in the household between patients and controls with significant level set at p ≤ 0.05. Univariate logistic regression was used to determine the risk factors of M. ulcerans infection by determining the odds ratio (OR) and 95% confidence interval (CI). All variables obtained from the univariate analysis with p-value ≤0.1 were retained for the multivariate model. The final model was obtained after a step-by-step backward elimination step using multiple logistic regressions.

After adjustment for potential confounders, we found that factors such as age (< 10 years (aOR = 11.48, 95% CI = 3.72–35. 43) and 10 to 14 years (aOR = 3.63, 95% CI = 1.22–10.83)), receiving insect bites near a river in children aged 10 to14 years (aOR = 7.8, 95% CI = (1.48–41.24)) and bathing with water from open borehole (aOR = 5.77, 95% CI = 1.11–29.27) (Table 5) remain as potential factors of increasing risk of M. ulcerans infection.

The present study showed that children under 15 years of age were at higher risk of contracting Buruli ulcer than adults. This result is in accordance with other studies conducted in Benin [11] and Ivory Coast [13] as well as WHO reports [37]. Indeed, in this age group children appeared to be often less protected especially at the head and feet [11]. Also, children’s behavior is usually driven by their parents’ activities as they accompanied them to the river for washing and for farming where they were highly exposed to aquatic areas that are associated with an increasing risk of BU infection.

We found that bathing with water from an open borehole was associated with higher risk of contracting BU. Similar results were found in Ghana [38], Ivory Coast [39] and Cameroon [14]. Indeed, other studies [6, 28, 40] have also shown that using unprotected water sources for bathing was associated with M. ulcerans infection. It has also been observed that even when used with soap, unprotected water sources constitute an increased risk of M. ulcerans infection [37]. However, Raghunathan et al. [38] in Ghana found that using a detergent while bathing provides significant reduction in Buruli ulcer risk. This difference could be explained by the antibacterial power of the soap used. Besides, in our study people from villages commonly used the local soap. On the other hand, we found that using soap to wash clothes or dishes was reducing the risk. This time, the type of the soap used for the laundry is provided from commercial brands which are strongly enriched in detergents and acids. Our study also identified other water sources of M. ulcerans infection such as swimming in a river, frequently crossing a river, receiving insect bites or injuries of cuts near rivers. However, after adjustment for potential confounders, only receiving insect bites near a river remained as an independent predictor of acquiring BU infection. Similar results were found in Ghana [38] but in Ivory Coast [13] and in other study [12], it was found that swimming or wading in water did significantly increase the risk of BU infection. To explore the difference of our finding with other studies, we looked to determine any potential age confounding or effect modification. Therefore, we found that insect bites increase the risk of BU only in 101-14 years age group (aOR = 7.80, 95%CI = 1.48–41.21). Though, other studies did not determine in which age group swimming or wading in water significantly increased the risk of BU, we could explain the difference between these studies by the age of BU cases. Further, in our study, 63% of BU cases were aged < 15 years while in Ivory Coast, 75% of cases were aged more than 15 years who are able to swim or wad in a river.

Most of the people surveyed were perform agricultural activities. However, we did not find any significant association with the risk of contracting BU. Among agricultural activities, planting and harvesting activities were associated with decrease risk of M. ulcerans infection. Similar results were found in Cameroon [14]. We observed that wearing a long-sleeved shirt or a long dress while performing agricultural activities did not provide significant reduction of the risk of contracting of Buruli ulcer. This observation is in accordance with the study conducted in Cameroon [14]. On the other hand, we found that wearing pants or hats is associated with reduction in the risk of mycobacterial infection. This would explain the low frequency of wounds on head and legs observed in our investigation. These results are consistent with those found in Ghana [12, 38] and Ivory Coast [13].

In Australia, Lavender et al. [20] showed that mosquito bites were significantly associated with Buruli ulcer. However, we did not find any risk of M. ulcerans infection associated to mosquito bites in Togo. In general, results of studies on mosquito bites associated with the use of mosquito coils or bednets during M. ulcerans infection are often contradictory [11, 14, 38, 39, 41].

Some studies [12, 13, 42] have shown that animals such as chickens, goats, cats and pigs could harbour M. ulcerans and exposure to these animals may increase the risk of contracting BU disease. During this study, we did not observe significant increase in risk of contracting BU associated with contact with domestic animals.

BCG vaccine is delivered against a mycobacterium. This vaccination could therefore provide a cross-protection against M. ulcerans infection [43]. In our study, we did not observe any significant difference in the percentage of BCG vaccination scar between patients and controls. The lack of a significant association with BCG vaccination with M. ulcerans infection has been also described in the literature [12, 16, 43]. However, data from Benin [11], Ivory Coast [13] and Cameroon [14] showed negative correlation between BCG vaccination and BU. Studies conducted to explore this possible cross-protection have often led to contradictory results. Indeed, a multicenter study [44] conducted in the DR Congo, Ghana and Togo did not reveal any significant association between BCG vaccination and BU disease.

The attitude of the participants interviewed has considerably improved with their capacity to recognize some BU symptoms and their ability to refer suspected cases to medical treatment compared to the situation 5 years before [33]. This finding could be attributable to several awareness campaigns in the community that had influenced their behavior toward this disease [33]. However, there remains some effort to help recognizing early symptoms by the community as well as the herbalists because 5.3% of BU patients continue to believe in herbal treatment as the first preferred treatment option. Poor individual hygiene and dirty surrounding were recognized as a potential risk factor for participants in the present study. The impact of poor hygiene and its possible role as a risk factor has been underlined in studies in Benin [11, 45] and Ghana [12].

This study had some limitation. We did not reach all participants especially some BU cases due to their unavailability during the survey time. The sample size was calculated based on the proportion of households using water from unprotected sources which was higher than the prevalence of BU. The number of newly confirmed BU cases in Togo every year is low and varies from 30 to 65 patients. During the study period, we found 91 BU cases but 8 patients were not available at the survey time. The main concern with the limit number of controls was due to the fact that in many households, there were often two to three patients and exceptionally in one house up to six. In those households, it was difficult to enroll two folds of controls. Moreover, as 47% of BU patients were under 10 years, it was difficult to interview children who were not capable to describe their activities which are driven by their parent’s duties. The reason we had decided to use their parents as controls sometimes.