Study site

All field research was conducted in Bugoto, Mayuge District, Uganda, which has approximately 3,500 inhabitants, divided over two villages. Bugoto A is a densely populated village situated on the shores of Lake Victoria and the majority of the adult population are fisherfolk or fishmongers [25]. Bugoto B, to the north and inland of Bugoto A, is less densely populated, and most inhabitants are farmers. There is a health clinic in Bugoto A, serving both villages. The main language spoken is Lusoga, a language spoken across the Busoga region in the south east of Uganda.

Terminology

We will use the term ‘gender’ throughout the discussion of our methods and results to describe self-reported identity of study participants and acknowledge the sociocultural aspects of exposure and infection, and limit discussion of our results to gender differences. Differences based on biological ‘sex’ are explicitly discussed in the introduction and discussion and reflect the data presented in those studies.

Ethnographic observations

To compare water contact behaviour between different risk groups, participants for the ethnographic observations were selected from a larger cohort (n = 274), recruited as part of the ongoing SCHISTO_PERSIST project, comprising SAC with an equal age/gender distribution. This cohort was recruited in March 2017 at Bugoto Lake View Primary School, the largest primary school in the community situated between Bugoto A and Bugoto B. Samples were taken from this cohort at four time points in 2017; March, September, October and December, with treatment administered after the September sample collection. Detected infection with schistosome parasites and mean infection intensities were recorded at every time point. In March 2018, eight children with no/very low and eight children with high baseline and reinfection intensities were recruited from the larger cohort for ethnographic observation. An effort was made to equally represent ages, gender and village of residence across each group using unique identifiers.

Ethnographic data were collected from the 5^th to 24^th March 2018, covering school days, and the 19^th September to 10^th October 2018, covering non-school days. A local village health team member was recruited to carry out translations between the researcher and the community. Observations of individual children were conducted from 10.30am (beginning of first school break, for the first data collection period) or 9.00am (for the second data collection period) to approximately 9.00pm. The researcher and translator spent the day accompanying and observing the child in their daily activities (including all water contact moments) and participating where possible and appropriate. Participation excluded high-risk activities such as water contact, and participation and observation excluded privacy-sensitive (toilet/shower use), culturally inappropriate (mosque attendance) or disruptive (class attendance) activities. In addition, we asked the child if they had any water contact on the day before our arrival. Data collection stopped when the child went to bed or no further water contact activities would be carried out before going to bed. Notes were taken during the day and later written up in detail. Transcripts were reviewed, coded and categorised using NVivo software [26], using thematic analysis [27], with a coding frame based on previously published literature complemented with codes emerging during the coding process.

Household survey

A household survey was administered in Bugoto A and B in February 2019 as part of a wider study on attitudes and behaviours for different water, sanitation and hygiene (WASH) interventions aimed at reducing S. mansoni transmission. The sampling methods have been described elsewhere [28]. The required sample size was 125 and based on the latest national census of the village of 2018 the household sampling interval was therefore six. We interviewed one adult (≥18 years old) respondent per household and aimed to have an equal gender distribution among respondents. One of five research assistants read out the survey to each participant in the local language and responses were collected on tablets using Sawtooth Computer Assisted Personal Interview software. Collected data for this study included demographic and socio-economic characteristics, knowledge of schistosomiasis, and water collecting practices. Data were cleaned with STATA software [29], and frequency and percentages were calculated. Gender differences were calculated with exact binomial tests and two-proportions z-tests.

Epidemiology diagnostics cohort

A random sample of 55 Schistosoma mansoni positive children (gender matched and evenly distributed across the 6–14 years age range) were selected from the SCHISTO_PERSIST cohort. The individuals were all positive for S. mansoni in October 2017 and were treated with 40 mg praziquantel per kilogram body weight. These individuals were resampled up to twice per week for five months following treatment (S1 Fig). At each timepoint, children were recruited at 9:00am and asked to give a single stool sample for duplicate thick smear Kato-Katz (KK) [30]. If an individual was diagnosed as having over 100 eggs per gram (epg) at nine weeks post-treatment, she/he was retreated with 40 mg/kg praziquantel (n = 3) for ethical reasons. At pre-treatment and five months post-treatment, children were sampled for three consecutive days to improve diagnostic accuracy [31]. Significant predictors of infection intensity (measured as epg per slide) were evaluated using a generalised linear mixed effects model with a negative binomial distribution to account for overdispersion of these data [32]. Gender and relative treatment (pre-treatment vs five months post-treatment) were used as fixed effects and unique child ID included as a random effect to account for repeated samples from individuals from multiple slides per day and days per time point. Models were created using the package lme4 [33] in R v. 3.5 [34]. Egg counts (as measured by KK) are highly variable between days, particularly in individuals with low infections [35, 36], so clearance of schistosomiasis was defined as two consecutive days of 0 epg in duplicate KK slides. The time to reinfection was identified as the first day that eggs were detected in stool following this clearance event. We used survival analysis methods to evaluate the time to event (here, reinfection rather than death) to determine if there are significant differences between genders in the proportion reinfected or the time to reinfection. The Kaplan-Meier method was used to fit the time to reinfection and estimate 95% confidence intervals (CI), and significant differences were evaluated using a log-rank test. The survival analysis was conducted with the survivfit function within the package survival [37] in R v. 3.5 [34].

Genetic diversity of schistosomes

To evaluate the genetic diversity of S. mansoni within individual children, miracidia were collected throughout the epidemiological survey. Collection protocols followed established methods [38], but briefly, after stool samples were submitted for KK, the remainder of the sample was washed through a Pitchford funnel with spring water [39]. The washed sample, potentially containing eggs, was left overnight in the dark and then exposed to indirect sunlight the following day. Individual miracidia were picked up in 2.0–2.5 μl of water with a micropipette under a stereomicroscope and placed on Indicating FTA Whatman cards. Samples were left to dry a minimum of one hour before being stored and transported in sealed plastic bags with desiccants. Data presented here are from miracidia from eight boys and eight girls collected in October 2017 (pre-treatment) and March 2018 (five months post-treatment). These individuals did not overlap with children selected for the ethnographic observations; rather these individuals were chosen because we were able to obtain miracidia from children at both timepoints (details in S2 Table).

DNA from individual miracidia was eluted from FTA cards using established protocols [40]. Seventeen microsatellite loci were amplified per miracidium using two separate PCR reactions (S1 Table). Type-IT microsatellite PCR kits (Qiagen) were used according to the manufacturer’s protocol for each multiplex PCR. Each PCR reaction was carried out in 12.5μl total volume using 2μl of eluted DNA from each miracidia. Multiplex amplicons were visualised on a 2% agarose gel stained with Sybr Safe (Invitrogen). Successfully amplified DNA from miracidia was diluted 1:50 in LIZ500 Size Standard (ThermoFisher) and HiDi Formamide before sequencing on an Applied Biosciences. Allele peaks were visualised and called in GeneMapper (ThermoFisher).

Genetic diversity measures were calculated for all miracidia and loci for each individual at each time point. Observed heterozygosity (Ho), expected heterozygosity (He), and allelic richness were calculated in poppr v. 2.8.1 [41]. Genetic diversity pre- and post-treatment was compared used a Wilcoxon signed rank test implemented with the function wilcox.test in R. The number of populations and the ordination of different parasite genotypes was investigated using Discriminant Analysis of Principal Components (DAPC) within the adegenet package [42, 43]. Variation among genders and timepoints was summarised by calculating Nei’s G_ST with functions in the package mmod [44].

Ethical clearance

Ethical approval for the survey and ethnographic research was granted by the Ugandan National Council for Science and Technology Social Sciences (reference: UNCST-SS 4241) and the University of Glasgow College of Social Sciences (reference: 400160134). Methods for the epidemiological and genetic data were reviewed and approved by the Vector Control Division Research Ethics Committee (VCDREC/062), the Uganda National Council of Science and Technology (UNCST-HS 2193) and the University of Glasgow Medical, Veterinary and Life Sciences Research Ethics Committee (200160068). Community consent was obtained from the District Education Officer, the District Health Officer and the Chairmen of the study villages. Individual written consent (signature or thumb print) was obtained from all participants 18 and over. Participants under 18 years old provided assent and additional consent was provided from a parent or guardian. Praziquantel treatment was not conditional on participation in the study or giving consent and all individuals could leave the study at any time without affecting their drug treatment.

Few differences in water contact between genders observed during ethnographic observations

Nine females and seven males, aged between 6 and 14 years old (median 10 years) were selected for ethnographic observations. Nine children were living in Bugoto A, seven children in Bugoto B. Seven identified as Muslim and nine as Christian (different branches). All consented to participate in the study. Two children were lost to follow up for the second data collection period; these were replaced by children with similar demographic characteristics. Differences in water contact by gender were assessed by types of activity, frequency, contact sites, level of submersion and duration of contact. No major cultural events that may have influenced behaviours were observed to have taken place during the data collection period.

We observed a range of activities carried out by the participating children that involved contact with lake water, in various settings and for different purposes. The main exposures to water were household tasks including fetching water, washing plates and jerrycans, washing clothes and flip flops, washing food items (potatoes, fish, vegetables, fruit), mopping by hand and personal care such as bathing (self and helping younger or disabled household members) and washing hands. Other types of water exposure were of a recreational (swimming) or a religious nature (washing hands, feet and face before attending the mosque). Occasional water contact also occurred at school, as children took turns in fetching water for the cook to prepare meals for students and when they washed their cup/plate after eating. Children were not seen to contact water in occupational settings such as working in rice paddies or helping with fishing, however some children carried out ‘commercial fetching’, fetching water for others in the community, receiving a small amount of money in return. Lake water was not observed to be processed by filtering, purification or sun exposure although large amounts of soap are used for washing plates, clothes and bathing. All water for drinking was fetched from a borehole or (paid-for) tap. Rain was considered ‘clean’ water and safe for drinking and household use by participants and their families, however it was not regularly collected due to the absence of water containers or tanks.

Children fetched lake water for the household between one and three times per day, and we found that girls fetched water for the household more often (on 11 of 14 observation days) than boys (on 10 of 18 days). Girls also more often fetched water more than once per day (on seven observation days vs four in boys); most of these children who collected water multiple times a day lived in Bugoto A, which is located closer to the lake. Commercial fetching was only carried out by children in Bugoto A, probably due to the higher number of businesses requiring water and the short distance to the lake. Two boys and two girls were found to carry out commercial fetching, up to five times per day.

Most children washed plates during the observation days, either one or two times per day. Although many children and parents said that children take turns in washing plates, more girls (on 12 out of 14 days) were observed to carry out this task than boys (on 10 out of 18 days). A few children washed clothes, however while spending time at participants’ homes, adult women seemed to be mainly responsible for this household chore. Washing of food items and mopping of the house was carried out by two girls and one boy. Most children bathed once a day, those who did not mentioned that the temperature in the evening was too low to bathe outside. Four girls bathed twice a day, however reasons for this are unknown. Hand washing was performed regularly by all children, mainly after meals, likely due to the fact that the hands are used for eating. The highest frequencies of hand washing were seen with children who were not in school, who either carry out many household tasks or handle different items at jobs.

Four of the boys engaged in swimming (including playing in the water, diving off boats, submerging to cool down during a hot day or small races) during the observations whilst none of the girls did. In addition, no other girls of the community were seen swimming at these sites. The translator confirmed that only a few girls swim and all children stop swimming when they reach puberty. Swimming activities ranged in frequency from one to four times per day.

Only one boy was observed to wash his hands, feet and face before every visit to the mosque. The other male Muslim participants did not visit the mosque at all during the observations and during the two data collection periods no women were seen entering the mosque or praying elsewhere.

No patterns of difference between gender were found when comparing sites of fetching water. Participants mentioned that their site of choice was closest to their home or cleanest and it was observed they consistently accessed water at the same site. Only when their preferred source dried up in the dry season, or water was considered too dirty by the child, would they move to another site. Observations while accompanying children to water contact sites did not reveal any access restrictions by gender and both genders were found to contact water at the sites. Washing plates, clothes, food items, body or hands were all carried out at the house compound with water previously collected at the lake.

The level of submerging and duration of water contact was not noticeably different between boys and girls for the activities both genders carried out, as all children showed similar approaches to the various activities. Level of submerging was more related to activity (fetching, swimming) and varying levels of submerging (until ankles, knees or hips pulling trousers or skirt up) during fetching were only seen when a participant sought for clearer water which was often further from the shore. Duration of fetching water, cleaning food items and washing plates and clothes was mainly based on the number of items. Children of both genders who carry out these activities tended to have similar and a continuous pace until the work was complete, without visible rushing or loitering. The weather affected the duration of some activities, for example bathing was observed to be rushed when the temperature was low, whilst on days with high temperatures boys were seen swimming for extended lengths of time.

Adult females responsible for the majority of household tasks involving water contact

A total of 130 household heads (52% female) participated in the household survey: 85 (65%) in Bugoto A and 45 (35%) in Bugoto B, answering on behalf of their whole household. During the ethnographic observations we identified fetching water and washing clothes as two key household tasks involving water exposure. Respondents were therefore asked who fetches the majority of water and who washes the majority of clothes for the household, specified by gender and age. In 106 (82%) households one person fetches the majority of the water, in 24 (18%) households this was between two to four persons. Almost all respondents (n = 127, 98%) reported that a member of the household, rather than someone outside the household, fetches the majority of the water. The remainder said other members of the community, unrelated to household members and over 18 years old, fetch water for their household. Of these, two were paid for their help.

For 118 (91%) households, there was one main person responsible for washing of clothes, for 12 households this was more than one person. Washing of clothes is mostly undertaken by household members, with only one respondent reporting clothes washing by other members of the community. No respondents reported that neighbours, or other family members not living in the household, fetched water or washed clothes for them.

For both fetching water and washing clothes, women were mainly responsible (68% and 89% of households respectively; Table 1), and especially those over 18 years old. Men were less often reported responsible for fetching (15%) and washing clothes (4%), whilst men and women shared the responsibility for the tasks in 17% and 7% of households respectively. In households with a single gender responsible for these tasks, women significantly more often fetched water (probability of 0.82, CI [0.74–0.89], p<0.001) and washed clothes (probability of 0.96, CI [0.91–0.99], p<0.001) as compared to an equal distribution between the two genders. No differences in the gender distribution for these two household tasks were found by demographic or socio-economic characteristic (Table 2).

Download:

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

Table 1. Key household activities involving water contact by gender and age, as reported by household head (n = 130).

https://doi.org/10.1371/journal.pntd.0008266.t001

Download:

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

Table 2. Gender distribution for fetching water and washing clothes by demographic and socio-economic characteristics (n = 130).

https://doi.org/10.1371/journal.pntd.0008266.t002

The occupation involving most water contact was fishing. Twenty-nine per cent of the population in Bugoto A reported fishing as their occupation, and 9% of those living in Bugoto B. There was a significant difference between genders; 28 of the 62 men (45%) were fisherfolk compared to one of 68 women (χ² = 33.23, CI [0.29–0.58], p<0.001). Water contact can also occur when farming, reported by 35% of those in Bugoto A compared to 64% in Bugoto B, however a similar proportion of men and women reported this as their occupation (45% of men and 46% of women (χ²<0.001, CI [-0.17–0.18], p = 1).

Rapid reinfection with S. mansoni in both genders

Fifty-four SAC (out of 55) gave samples for at least 40% of sample collection days throughout the five months and were included in the analysis. Both genders had a moderate mean infection intensity pre-treatment: 162.3 epg in males and 122.5 epg in females. At five months post-treatment, the mean infection intensity pre-treatment was lower in each gender; 99.8 epg in males and 80.8 epg in females. Our mixed-effects modelling results suggest that drug treatment affects observed infection intensity, with infection intensity significantly higher pre-treatment than post (slope estimate = 1.03, z-value = 3.88, p = 0.0001). However, gender was not a significant predictor of infection intensity (p = 0.37).

Fifty-two out of the 54 (96%) individuals were identified as reinfected during the five month follow-up period (egg positive after a period of clearance). Among these reinfected individuals, the period of no egg production varied from seven to 125 days. All males were reinfected in the time period observed and the mean time to reinfection was 54.4 days. All but two females were reinfected within the observation period and the mean time to reinfection for females was 39.3 days. Despite this difference, Kaplan-Meier curves found no significant difference in the likelihood of reinfection between genders (Fig 1; p = 0.62).

Download:

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

Fig 1. Reinfection by gender.

Reinfection curves and 95% confidence intervals by gender for the cohort of 54 children that were sufficiently sampled during the longitudinal study. There is no significant difference in reinfection rates or time to reinfection between genders.

https://doi.org/10.1371/journal.pntd.0008266.g001

No differences in within-host parasite genetic diversity between genders

A total of 386 miracidia were successfully genotyped from the 16 children (eight males, eight females) at pre-treatment and five months post-treatment. The average genetic diversity of S. mansoni collected from female hosts (pre-He = 0.65; post-He = 0.67) was not significantly different to those collected from male hosts (pre-He = 0.66; post-He = 0.67) before (Fig 2; t = -0.62; p = 0.54) or after treatment (t = 0.12; p = 0.97). Parasite genetic diversity pre-treatment was not significantly correlated with parasite genetic diversity post-treatment in the same host (p = 0.17). This was also true when host genders were analysed separately: genetic diversity pre- and post- treatment were not correlated in parasites from either females (p = 0.11) or males (p = 0.74).

Download:

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

Fig 2. Mean genetic diversity by gender before and five months after treatment.

There are no significant differences in average genetic diversity (measured as expected heterozygosity) between male and females, either before or after treatment. Violin plots show distribution of genetic diversity and internal boxplots show mean and first and third quartile range per gender and timepoint. Mean genetic diversity per child is shown by individual open circles.

https://doi.org/10.1371/journal.pntd.0008266.g002

We asked whether population structure of schistosome parasites is differentiated by gender and how population changes with treatment. There was significant, yet low differentiation within the population—the mean Nei’s G_ST (fixation index for multi-locus data) between all individuals was 0.02 (p = 0.002). Despite this differentiation, there is not significant differentiation between genders or timepoints. Four distinct sub-population clusters were identified using discriminant analysis of principle components from the entire group of miracidia (Fig 3A). However, these clusters were not significantly correlated with treatment or gender (Fig 3B). Although there was population differentiation in parasites, differentiation was not associated with host gender or timing relative to treatment.

Download:

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

Fig 3. Population differentiation with discriminant analysis of principle components (DAPC).

A. DAPC identified 4 clusters of parasite genotypes within the population sampled. B. However, these clusters were distributed relatively evenly between both genders and timepoints, suggesting no population structure determined by gender or treatment effects.

https://doi.org/10.1371/journal.pntd.0008266.g003