Many studies have attempted to measure individuals' exposure, either by directly observing behavior at water sources [6-18], or indirectly by interview/questionnaire [19-27]. Only few studies attempted to quantify variables that are important in determining age and sex patterns of exposure from the corresponding patterns of water contact [12,16,28-31] and even less tried to relate these to infection levels and patterns [5,11,32,33].

Figure ​Figure33 and ​and44 show the duration and frequency of water contact per season, age and sex. Although in the hot dry season water contacts were slightly more intense than for the other seasons, patterns of both duration and frequency depend much more on age than on the season. In the age group of 10-19 years, longer and more frequent water contact was observed than in the other age groups. Women showed substantially more and longer water contact than men; for the older age groups (10-19 and more than 20 years), duration and frequency of water contact in women were even twice that of men. Female adolescents (10-19 years) showed the most intense water contact, which is mainly due to collecting water more often and a longer time spent on household activities. Most intense water contact occurred between noon and 2 p.m. for men; for women two distinct peaks were observed, one between 8 and 10 a.m., and one between 2 and 5 p.m.

Figure ​Figure55 shows for each age- and sex group the infection intensity in relation to duration and frequency of water contact. It is clear that EPG-values (both pre-control and 1 year after treatment) are not linearly related with water contact. Especially adult females had lower egg counts than would be expected from the duration and frequency of their water contacts, while male adolescents appeared to have the highest EPG/water contact ratio. The average number of water contacts per person per day in this population was 0.42; the average time spent in the water per person per day was 4.3 minutes.

Figure ​Figure66 shows for each age- and sex group the infection intensity in relation to duration and frequency of water contact, after adjustment for body surface and time of day of contact. Both proxies for exposure, especially frequency, showed a better fit with intensities of infection when adjusting for these two factors (Figure ​(Figure66 vs Figure ​Figure5).5). Nevertheless, the relationship between infection and water contact patterns still deviated from a clear proportional one, especially in women.

Also in terms of the number of water contacts per person per day, our results in Northern Senegal did not show marked differences with other schistosomiasis endemic countries where intensive observational water contact studies have been performed. Indeed, our finding of an average of 0.42 contacts per person per day is within the range of Fulford et al. [10], who reported a mean annual frequency of 12.8 to 162 water contacts/person among seven S. mansoni endemic communities in Kenya, which is 0.04 to 0.44 contacts/person/day (median 0.17). Exact average water contact durations for these communities were not reported, but from the graphs it can be deduced that this was about 880 minutes/person annually, and thus 2.4 minutes per day. Again, this is of the same order of magnitude as the average duration of 4.3 minutes spent in the water per day as presented here for Northern Senegal. Chandiwana & Woolhouse [38] reported a mean rate of water contact of 0.43 contacts/person/day (ranging from zero to 3.3 contacts/person/day), in a S. haematobium endemic area in Zimbabwe, which is remarkably similar to our values for Senegal. Other water contact studies based on direct observations used different exposure indices, which precludes straightforward comparison [32,39,40]. Nevertheless, from the available observational water contact studies we can conclude that the water contact levels in Ndombo are not exceptionally high and thus cannot explain the extremely high infection intensities in this area as compared to the other studies.

Both infection- and water contact patterns in this community were found to be clearly age-and sex-related. Looking at these patterns more closely however, we found that in the respective age- and sex groups more or longer water contact did not unequivocally lead to high infection intensities (Figure 5 +​+6).6). In the oldest two age groups, women showed substantially more and longer water contact than men, while infection levels were comparable. An explanation for this finding could be that male water-related activities entail relatively intense water contact. Indeed, the main male activity was bathing, which can be considered a more risky behaviour in terms of body exposure as compared to typical female activities such as collecting water or doing laundry/dishes. Also the degree of using soap, which may have a 'cercaricidal' effect [10], for washing clothes and/or bathing may have influenced infection intensities in women and men in different ways. Moreover, in men most intense water contact occurred between noon and 2 p.m., the part of the day when cercarial emission has been reported to be highest in the Senegal River Basin [41], while women's water contact peaked in the morning and afternoon. Indeed, after adjusting for body surface and time of day, the association between water contact and intensities of infection appeared to be stronger than without including these factors, but differences in this relationship across the demographic groups remained (Figure ​(Figure6).6). In addition, male adolescents showed much higher infection levels than the other male age groups, while they did not bath significantly more or longer than young or adult men, nor at different times of the day. Another hypothetical explanation may be that certain typical male activities have not been captured, as these may have taken place outside the designated observation time or sites (e.g. fishing). However, we have no indications for such an effect in males only.

Despite observed differences in infection- and water contact patterns depending on age and sex, exposure appears not to be the only factor of importance in determining intensity of infection before or one year after treatment in this Schistosoma epidemic focus. Similar conclusions resulted from studies in stable endemic situations [11,39,42,43]. A few have found some relationship between water contact and infection intensity, although not very strong/convincing [12,38,44]. It should be noted that for any of these studies, including ours, it cannot be excluded that exposure factors other than those taken into account, may have somehow contributed to the observed age- and sex-related differences in infection levels [5,39]. In the specific case of Ndombo, very high numbers and infection rates of Biomphalaria pfeifferi, the snail intermediate host of S. mansoni, were found at the start of the epidemic [2], which could also have (partly) accounted for the extremely high S. mansoni infection intensities of the community. An obvious alternative explanation other than exposure would be the absence of acquired immunity in this recently exposed, supposedly non-immune community, but this does not correspond with the observed endemic-like age-related infection patterns in this community [1,3].

For women, the ratio of infection intensity to water contact (i.e. the slope of a hypothetical line from the origin to a point) in Figure ​Figure55 and - to a lesser extent - Figure ​Figure66 decreased with age, suggesting an increasing degree of resistance with age. In an epidemic focus like Ndombo, immunity should not, or at most partially, have developed at the time of the data collection. Thus, this resistance is more likely to be due to some other age-related, innate factor [2,4,5]. In men, however, this pattern seemed to be absent, with the youngest age group showing a slightly lower ratio than the older groups, particularly after adjusting for body surface and time of day (Figure ​(Figure6).6). As yet, there is no biological evidence that could explain why such resistance would only occur in adult women and not in adult men. Post-pubertal hormonal or other (both age- and sex-related) factors may play a role. For example, Fulford et al. [45] suggested that gonadal steroids affecting the immune system may lie behind the common observation, originally made by Butterworth et al. [46], that women are usually infected less heavily than men yet generally have more water contact.

A non-proportional relationship between egg counts and water contact may also indicate that egg production is suppressed in individuals with higher water contact, and thus suggesting some form of density-dependence. For example, density-dependent fecundity has been widely implicated as a major regulatory force in maintaining helminth population stability, and is generally considered to result either from competition between parasites for host resources or from immunological control [47]. The phenomenon is still controversial for schistosomiasis [48-54]. And yet again, we cannot explain why such density-dependence would only occur in women and not in men.

To our knowledge, this is the largest dataset published so far of directly observed recorded water contacts, spanning a two year time period in the early years of an S. mansoni epidemic. Direct water contact observations have more quantitative and qualitative value than water contact information based on questionnaires, which are easy to perform but have inherent well-known limitations such as overreporting, recall bias and information bias [55,56]. This is illustrated by a questionnaire-based study in four villages in the same S. mansoni affected area as Ndombo [5]. This study reported a mean of 4.4 water contacts per day with a median duration of 57 min per day, which is in sharp contrast with the relatively low numbers found in the present study. It has been noted before that levels and patterns of contact can vary dramatically between culturally similar communities, and even within a single village [10,36], but these extreme numbers are more likely to be due to the way the water contact information was collected [56].