Prevalence, intensity and associated risk factors of soil-transmitted helminth and schistosome infections in Kenya: Impact assessment after five rounds of mass drug administration in Kenya

Background In Kenya, over five million school age children (SAC) are estimated to be at risk of parasitic worms causing soil-transmitted helminthiasis (STH) and schistosomiasis. As such, the Government of Kenya launched a National School Based Deworming (NSBD) program in 2012 targeting the at-risk SAC living in endemic regions, with the aim of reducing infections prevalence to a level where they no longer constitute a public health problem. The impact of the program has been consistently monitored from 2012 to 2017 through a robust and extensive monitoring and evaluation (M&E) program. The aim of the current study was to evaluate the parasitological outcomes and additionally investigate water, sanitation and hygiene (WASH) related factors associated with infection prevalence after five rounds of mass drug administration (MDA), to inform the program’s next steps. Materials and methods We utilized a cross-sectional design in a representative, stratified, two-stage sample of school children across six regions in Kenya. A sample size of 100 schools with approximately 108 children per school was purposively selected based on the Year 5 STH infection endemicity prior to the survey. Stool samples were examined for the presence of STH and Schistosoma mansoni eggs using double-slide Kato-Katz technique, urine samples were processed using urine filtration technique for the presence of S. haematobium eggs. Survey questionnaires were administered to all the participating children to collect information on their demographic and individual, household and school level WASH characteristics. Principal findings Overall, STH prevalence was 12.9% (95%CI: 10.4–16.1) with species prevalence of 9.7% (95%CI: 7.5–12.6) for Ascaris lumbricoides, 3.6% (95%CI: 2.2–5.8) for Trichuris trichiura and 1.0% (95%CI: 0.6–1.5) for hookworm. S. mansoni prevalence was 2.2% (95%CI: 1.2–4.3) and S. haematobium prevalence was 0.3% (95%CI: 0.1–1.0). All the infections showed significant prevalence reductions when compared with the baseline prevalence, except S. mansoni. From multivariable analysis, increased odds of any STH infections were associated with not wearing shoes, adjusted odds ratio (aOR) = 1.36 (95%CI: 1.09–1.69); p = 0.007; high number of household members, aOR = 1.21 (95%CI: 1.04–1.41); p = 0.015; and school absenteeism of more than two days, aOR = 1.33 (95%CI: 1.01–1.80); p = 0.045. Further, children below five years had up to four times higher odds of getting STH infections, aOR = 4.68 (95%CI: 1.49–14.73); p = 0.008. However, no significant factors were identified for schistosomiasis, probably due to low prevalence levels affecting performance of statistical analysis. Conclusions After five rounds of MDA, the program shows low prevalence of STH and schistosomiasis, however, not to a level where the infections are not a public health problem. With considerable inter-county infection prevalence heterogeneity, the program should adopt future MDA frequencies based on the county’s infection prevalence status. Further, the program should encourage interventions aimed at improving coverage among preschool age children and improving WASH practices as long-term infection control strategies.

Introduction Soil-transmitted helminths (STH: three common species being Ascaris lumbricoides, the hookworms; Necator americanus and Ancylostoma duodenale, and Trichuris trichiura), and schistosomiasis (mainly caused by Schistosoma mansoni and S. haematobium), are among the most Western, Rift Valley and Coast regions), and those sites that had not had routine parasitological monitoring since baseline but had been participating in deworming (the sites included Eastern and North Eastern regions). Further, during this evaluation survey, the program collected individual, household and school level WASH and demographic characteristics in order to determine factors associated with the risk of STH and schistosomiasis in the national program in Kenya.

Study design and sampling
The study utilized a cross-sectional design in a representative, stratified, two-stage sample of school children across six regions in Kenya. A sampling frame of all primary schools which were participating in the NSBD program within a county was taken. A sample size of 100 schools (five schools per county with highest STH prevalence) with approximately 108 children per school was calculated to be adequate to detect a 5% change in prevalence of STH infections, assuming power of 80% and test size of 5%, and considering the anticipated variance in prevalence. The schools were purposively selected based on Year 5 (the year 2017) STH infection endemicity prior to the study [15]. In each school, 18 children (nine girls and nine boys) were sampled randomly from each of the six classes; one early childhood development (ECD) class and classes two to six using random number tables, for a total of approximately 108 children per school.

Survey procedures
The selected schools were visited three days prior to the survey date to explain the purpose of the survey to the school head teacher and the school committee; permission to conduct the study was sought at the school level. On the day of the survey, each selected child was given a container (poly pot) labeled with unique identifier and instructed to place a portion of his or her own stool sample in it. The stool samples were then processed in the laboratory within 24 hours and examined in duplicate for the presence of STH and S. mansoni eggs by two technicians using the Kato-Katz technique [16]. Additionally, urine samples were obtained only from children in the participating schools from Coastal, Eastern and North Eastern regions where S. haematobium is known to be focally prevalent. The urine samples were then processed using the urine filtration technique in the laboratory within 24 hours using polycarbonate membrane filters and examined in duplicate for the presence of S. haematobium eggs by two technicians [17]. However, children not present during the day of the survey were not included in the survey. As part of the NSBD program during MDA, all participating children were treated with albendazole (400 mg) and praziquantel (40 mg/kg) for STH and schistosomiasis respectively according to MoH and WHO guidelines [18].

Data collection and management
Data was collected in two phases between 29th January to 17th February, 2018 and 8th to 24th May, 2018, approximately 12 months after the Year 5 MDA in each region. Single stool specimens were collected to assess prevalence and intensity of STH (A. lumbricoides, T. trichiura, and hookworm) and S. mansoni infections. Single urine specimens were additionally collected in Coast, Eastern and North Eastern regions to assess prevalence and intensity of S. haematobium infections. In all areas, pilot-tested survey questionnaires were administered to all the participating children to collect information on the participants' demographic and their individual, household and school levels WASH related behaviours, practices and characteristics.
Both survey questionnaires and laboratory reporting forms were programmed onto androidbased smart phones and used to capture data electronically using the Open Data Kit (ODK) system which incorporated in-built data quality checks to reduce data entry errors [19].

Ethics statement
Ethical approval for the study protocol was obtained from the Kenya Medical Research Institute (KEMRI)'s Scientific and Ethics Review Unit (SSC Number 2206). At county-level, approval was provided by the respective county health and education authorities. At school, parental consent was obtained based on passive opt-out consent rather than written opt-in consent due to the routine and low risk nature of the study procedure. Additionally, individual assent was obtained from each child before participation in the study. All data used were anonymised.

Statistical analysis
Infection prevalence and average intensity of infection were calculated for STH and schistosomiasis and the 95% confidence intervals (CIs) determined using binomial and negative binomial regression models respectively, taking into account clustering by schools. Infection intensities were classified into light, moderate and heavy infections according to WHO guidelines (S1 Table) [20] and the prevalence of light, moderate and heavy infections together with 95%CIs obtained using a binomial regression model taking into account clustering by schools. We calculated the prevalence of each intensity class using two approaches; 1) when taking the denominator as the overall number of children examined, and 2) when taking the denominator as the total number of positive-children for each respective infection. The use of these two approaches enabled us to conveniently compare the morbidity due to these infections and for easy comparison to other studies.
WASH and sociodemographic conditions of interest from the questionnaires included reported individual, household and school-level variables that are known factors affecting STH or schistosomiasis prevalence. Individual factors included age, gender, handwashing, defecation and urination, soil-eating and shoe-wearing behaviours at school and home. Householdlevel factors included availability of toilet, anal cleansing material, handwashing facility equipped with water and soap, type of water source, as well as number of people living in an individual's household. Information regarding type of household latrine (being improved or unimproved) was not collected as the surveys were conducted at school locations. School-level factors included interviewer-verified availability and type of school toilet facility, availability and type of handwashing facility equipped with water and soap, and availability of anal cleansing material at school. Additionally, latrine structural integrity and cleanliness were assessed by interviewers. Latrine structural integrity was assessed by the evidence of all the following: roof, walls with no holes, a functional lockable door, and a stable floor slab, while assessment of latrine cleanliness was determined by the absence of strong smell, absence of visible faeces on the latrine floor, and clean floor. Index scores for latrine structural integrity and cleanliness were created using factor analysis [21], with a score range of between zero and one, and with higher scores indicating greater cleanliness/structural integrity.
Overall, the WASH factors associated with STH or schistosomiasis prevalence were analyzed, first using univariable analysis and described as odds ratio (OR) using mixed effects logistic regression model at two levels; pupils nested within schools selected within counties. To select minimum adequate variables for multivariable analysis, an inclusion criterion of pvalue <0.1 was pre-specified in a sequential (block-wise) variable selection method which selected covariates meeting the set criterion; however, sex and age were retained as fixed terms in the final models regardless of statistical significance due to their known importance. Adjusted OR (aOR), of the most parsimonious model, were obtained by mutually adjusting all minimum generated variables using multivariable mixed effects logistic regression model at 95%CI taking into account the hierarchical nature of the data.
All the statistical analyses were carried out using STATA version 14.1 (STATA Corporation, College Station, TX, USA). Graphs were developed using the ggplot package implemented in R [22]. School locations were mapped using ArcGIS Desktop version 10.2.2 software (Environmental Systems Research Institute Inc., Redlands, CA, USA).

Results
Overall, 100 schools (9,801 children) with the median age of 10 years (range:1-21 years) were surveyed across 20 counties in Western, Nyanza, Rift Valley, Coast, Eastern and North Eastern regions prior to the Year 6 MDA. Five schools with a total of 108 children per school were surveyed from each county. Approximately half (50.2%) of the children were males. It is important to note that the wider age range here is due to inclusion of few younger children (<2 years) that were found in ECD classes, who properly might have accompanied their siblings or mothers to school on the day of the survey, on the other hand it is normal to get older children (�15 years) in most primary schools in Kenya due to variety of reasons. Table 1 provides the number of schools and children examined by county as well as the range of school-level prevalence for both STH and schistosomiasis.  (Table 2). Overall, the undifferentiated STH prevalence reduced by 61.7% (p<0.001) from the baseline of 32.3%; similarly, specific species indicated significant declines over the period. Hookworm reduced by 93.6% (p<0.001) from a baseline prevalence of 15.4%, A. lumbricoides reduced by 52.9% (p<0.001) from baseline prevalence of 18.1% and T. trichiura reduced by 42.7% (p<0.001) from initial prevalence of 6.7%. Similar declines were observed for mean intensity of infections ( Table 2).
Analysis of undifferentiated STH prevalence by demographics (sex and age group) showed that there was no significant difference between male and female children (χ 2 = 0.31, p = 0.578) or by age groups (χ 2 = 4.08, p = 0.130). Whilst non-significant slightly higher hookworm infection prevalence was seen in older children (<5 years: 0%, 5-14 years: 1.0% and >14 years: 1.2%), analyzing differences at this low level of prevalence is affected by the limits of diagnostic accuracy. For A. lumbricoides, non-significantly higher infection prevalence was seen in younger children (<5 years: 14.7%, 5-14 years: 9.7% and >14 years: 4.9%) and for T. trichiura, non-significantly higher infection prevalence was observed among those aged 5 to 14 years at 3.6%, compared to those aged <5 years or >14 years (Table 3).

PLOS NEGLECTED TROPICAL DISEASES
Sixth year impact assessment of the national school based deworming program in Kenya prevalence of any STH above 20%. All surveyed schools in Eastern and North Eastern regions had STH prevalence below 1%. There was no infection among surveyed schools in Garissa and Wajir counties. Three counties (Kitui, Makueni and Taita Taveta) had STH prevalence below 1%, six counties (Bungoma, Kilifi, Kisumu, Kwale, Migori, and Mombasa) had STH prevalence between 1% and 10%, one county (Kericho) had STH prevalence between 10% and 20%, and the remaining eight counties had STH prevalence between 20% and 50% (Table 4). County-specific mean intensities of infection are shown in Table 5.
At regional level, STH infections were generally more prevalent in Rift Valley Region (21.8%) with species-specific prevalence of 17.4% for A. lumbricoides, 5.5% for T. trichiura and 0.1% for hookworm, followed by Western Region (20.9%) with species-specific prevalence of 15.8% for A. lumbricoides, 7.7% for T. trichiura and 1.3% for hookworm. Low percentages of infections were observed in Eastern Region where only hookworm was present at 0.5% (noting that diagnostic accuracy of Kato Katz at this level of prevalence is questionable). However, no STH infection was observed in any of the surveyed schools of North Eastern Region.
Using both approaches, we found that STH infections were predominantly of light intensity, followed by moderate and then heavy intensity. In particular, using the first approach, the prevalence of light infections were 5.8% (n = 561) for A. lumbricoides, 5.2% (n = 322) for T. trichiura and 2.8% (n = 91) for hookworm, and prevalence of heavy infections were observed for A. lumbricoides only at 1.9% (n = 3). The prevalence of moderate to heavy infections for any STH infections was 6.0% (n = 399), with the majority of the moderate to heavy infections being for A. lumbricoides. The first approach showed that the prevalence of moderate to heavy   intensity of the STH infections has significantly reduced since baseline for all the STH infections except that of T. trichiura (Table 6).

Schistosome infections
The overall prevalence of S. mansoni infection was 2.2% (95%CI: 1.2-4.3) and for S. haematobium, 0.3% (95%CI: 0.1-1.0). Respective mean intensities of infection of 12 epg (95%CI: 5-31) and 0 eggs/mL (95%CI: 0-1) were seen ( Table 2). Prevalence of schistosomiasis overall was very low, and at these levels only S. haematobium prevalence reduced significantly, by 98.5% (p<0.001) from baseline prevalence of 18.0%. S. mansoni showed a non-significant reduction of 7.9% (p = 0.779) from initial prevalence of 2.4%. However, it is difficult to assess significant reductions in both species of Schistosoma due to poor diagnostic performance of Kato-Katz at low levels of prevalence, and also because the baseline prevalence of S. mansoni was very low to commence with. Similar decline patterns were observed for mean intensities of infection ( Table 2).

PLOS NEGLECTED TROPICAL DISEASES
Sixth year impact assessment of the national school based deworming program in Kenya   -$ Indicate that prevalence of intensity was not assessed at that particular cut-off point n; indicates the number positive for each intensity class # Prevalence of each intensity class was calculated using two approaches 1) � when taking the denominator as the overall number of children examined, and 2) �� when taking the denominator as the total number of positive-children for each infection. The use of these two approaches enabled us to conveniently compare the morbidity due to these infections and for easy comparison to other studies. https://doi.org/10.1371/journal.pntd.0008604.t006 prevalence of S. mansoni above 10%. There was no S. mansoni infection in the surveyed schools in ten counties, below 1% prevalence in four counties, between 1% and 10% in five counties and above 10% only in Busia County. Similarly, S. haematobium prevalence was zero in Makueni, Wajir and Taita Taveta counties, below 1% in Garissa and Kilifi counties, and between 1% and 10% in Kwale and Mombasa counties (Table 4). County-specific mean intensities of infection are shown in Table 5. At regional level, S. mansoni infection was most prevalent in Western (6.3%), followed by Eastern (2.4%), Nyanza (1.9%) and Coast regions (0.3%), however no S. mansoni infection was observed in North Eastern and Rift Valley regions. Additionally, S. haematobium infection was only observed in Coast (0.4%) and North Eastern (0.3%) at very low levels, with no observed infection in Eastern Region.
Similarly, using both approaches, we found that intensity of schistosomiasis were predominantly of light intensity, except that of S. mansoni which were predominantly of heavy infections according to first approach. In particular, using the first approach, the prevalence of light infections were 0.9% (n = 85) and 65.2% (n = 6), and prevalence of heavy infections were 2.5% (n = 55) and 0.1% (n = 3) respectively for S. mansoni and S. haematobium. The prevalence of moderate to heavy infections for any schistosome was 3.2% (n = 26) with the majority of the moderate to heavy infections being for S. mansoni. The first approach showed that the prevalence of moderate to heavy intensity of S. haematobium significantly reduced since baseline (RR = 98.5%, p<0.001) but it instead increased by more than two-folds for S. mansoni (Table 6).

Individual, household and school WASH characteristics
All the 9,801 children surveyed from 100 schools were administered with a questionnaire, where they reported on their WASH practices and behaviours both at school and at home. However, we would like to point out that some of the indictors measured like number of household occupants and shoe-waering behaviour were proxy indicators of poverty but not risk factors per se. Table 7 gives the WASH characteristics, overall and stratified by region.

STH infections
Any STH infection prevalence The overall reported average number of household occupants was 6.8 people (standard deviation (SD) = 2.6 people). At the time of the interview, the majority 8,101 (84.5%) of the pupils were wearing shoes. Geophagy was not uncommon at 2,596 (27.1%) of the pupils. Nearly half 4,866 (49.7%) of the pupils reported use of an improved water source for drinking at their household. Reported latrine coverage (any type of latrine) at household level was high 9,329 (97.3%), however, fewer pupils reported always having a handwashing facility equipped with water and soap 1,422 (14.5%), or tissues/water for anal cleansing 5,174 (54.0%) available in their households. School WASH conditions varied considerably by region as well as by county ( Table 7). The average number of pupils per school was 527 (SD = 377). Improved water sources were interviewer-observed in 40

Univariable and multivariable analysis of factors associated with STH infections
A large number of individual, household and school level WASH and socioeconomic factors were assessed in a univariable model and revealed significant associations with STH infections as shown in Table 8. They were then fitted in a multivariable model after adjusting for other factors, however, many of them did not remain significant (Table 9). Not wearing shoes on day of interview (adjusted odds ratio (aOR) = 1.36, p = 0.007), and household membership of more than five members per household (aOR = 1.21, p = 0.015), showed significant association with increased odds of any STH infections. Additionally, a gradient effect relative to never being absent from school was evident; although one day's absence was non-significant (aOR = 1.05, p = 0.612), two days' absence (aOR = 1.32, p = 0.029), and more than two days' absence (aOR = 1.33, p = 0.045) were both significantly associated with increased odds of STH infection. Children aged 5-14 years (aOR = 3.25, p = 0.015), relative to those aged over 14 years, were shown to have three times greater odds of STH infections; on the other hand, children aged less than 5 years had over four times greater odds of STH infection than those aged over 14 years (aOR = 4.68, p = 0.008). Availability of tissue/newspaper/water for anal cleansing at home (aOR = 0.77, p = 0.005), and households possessing electricity (aOR = 0.75, p = 0.001) were associated with lower odds of any STH infections. Sex was not significantly associated with STH infection.
Nearly all the above factors; not wearing shoes (aOR = 1.46, p = 0.003), high household membership (aOR = 1.24, p = 0.001) and school absenteeism of two days (aOR = 1.40, p = 0.016), and more than two days (aOR = 1.35, p = 0.072) similarly showed higher significant risk for A. lumbricoides infection. The availability of tissue/newspaper/water for anal cleansing at home (aOR = 0.78, p = 0.015) and availability of television at home (aOR = 0.68, p = 0.001) showed significant association with lower odds of A. lumbricoides infection (Table 10). Children aged less than 5 years had significantly greater odds of infection than children aged over 14 years (aOR = 3.56, p = 0.045), but children aged 5-14 years did not (aOR = 1.72, p = 0.308). Sex was not significantly associated with A. lumbricoides infection.
For T. trichiura infection, children aged 5-14 years (aOR = 10.01, p = 0.027) and two days' school absenteeism (aOR = 1.72, p = 0.020) were the only risk factors identified; though more than two days' absenteeism was mildly non-significantly associated with increased odds of infection, (aOR = 1.59, p = 0.072). Always using school latrine (aOR = 0.26, p = 0.010) was shown as a protective factor (Table 11). A multivariable model for hookworm infection was not run due to insufficient observations.

Univariable and multivariable analysis of factors associated with schistosomiasis
Univariable analysis of individual, household and school level WASH and socioeconomic factors showed no significant associations between prevalence of schistosomiasis and any of the variables of interest as shown in Table 12. However, factors such as not wearing shoes (odds ratio (OR) = 1.20, p = 0.501), not receiving treatment during the last MDA (OR = 1.08, p = 0.779), sharing toilet/latrine with other households (OR = 1.12, p = 0.515), and school absenteeism (OR = 1.04, p = 0.873) showed increased odds of S. mansoni infection, though not significant. Additionally, only school absenteeism (OR = 1.13, p = 0.894) showed non-significant increased odds of S. haematobium infection. As such, and also due to low number of observations, multivariable analysis for factors associated with schistosomiasis prevalence was not conducted.

Discussion
Since Kenya has implemented the NSBD program for five years, it is important to conduct an evaluation of the program with the view to refine MDA requirements in line with the WHO guidelines [5]. Additionally, this evaluation enables us to understand the variation of the program impact between and within counties as well as risk factors associated with infection.

STH and schistosome infections
From this evaluation, STH infections have declined tremendously after five rounds of treatment to prevalence of 12.9%; a highly significant reduction of 61.7% since baseline. Less marked but also important, are reductions in schistosomiasis prevalence to 2.2% (S. mansoni) and 0.3% (S. haematobium). Mean intensities for these parasites similarly reduced over this period. These results indicate that Kenya, via its NSBD, is making good headway towards achieving elimination of these diseases as a public health problem. WHO recommendations indicate that where STH five-year evaluation prevalence is assessed at 10-20%, preventive chemotherapy should continue once per year, and where schistosomiasis five-year evaluation prevalence is assessed at 1-10% it should continue once after every two years [5].
Whilst drops in each STH species prevalence were significant, the species indicated different decline rates perhaps reflecting their different dynamic mechanisms and reactions to albendazole [23,24]. A. lumbricoides has been previously shown to have high levels of re-infections [14], while single-dose oral albendazole, as is given in STH control programs, has been shown to be less efficacious against T. trichiura [25]. For schistosomiasis, whilst only the decline in S. haematobium was significant, numbers overall for both schistosome species were very low. The non-significant decline in S. mansoni might be attributed to several factors such as the poor sensitivity of the Kato-Katz diagnostic technique at low prevalence, its already low baseline prevalence [26], along with a known gap in treatment coverage as there was a lack of availability of praziquantel in one year of the NSBD program. These factors might have resulted in the observed high prevalence of heavy-intensity of any schistosomiasis especiacially S. mansoni infection and as such this prevalence is still above the newly-defined WHO target of <1% [27], indicating that schistosomiasis in the country is still a public health problem, and there is no question that praziquantel treatments should continue. WHO guidelines further provide for conducting of serological surveys where five-year follow-up prevalence of schistosomiasis is <1%, before making any decision to stop MDA [5]. Further, it is concerning that a few pocket of schools mainly in the western part of Kenya had high S. mansoni prevalence above 10%, this could be influenced by their nearness to Lake Victoria and hence indication of high schistosomiasis transmission around that lake and its associated islands, a finding that has been supported by various studies [28][29][30], and calls for critical need for targeted control of schistosomiasis.
Examples of large-scale helminth programs achieving levels of reduction warranting reducing or ceasing MDA are still relatively unusual, and whilst WHO decision trees provide guidance on reduction and cessation thresholds, it is also generally acknowledged that MDA needs to continue for some time after achieving the definition of elimination as a public health problem to prevent resurgence of infection. Whilst as noted above the national program has not achieved this level yet, with infection heterogeneity within counties, some but not all counties potentially meet definitions of elimination of STH or schistosomiasis as a public health problem. As the survey was not powered to county level, this is a picture that is emerging that warrants careful investigation going forward. This evaluation did not find any STH infection among the schools surveyed in Garissa and Wajir counties, either a feature of the non-random selection of schools or possibly indicating no ongoing biological transmission of STH in those counties and probably in the greater North Eastern Region of Kenya. This is in agreement with past empirical and research studies in the region, which have indicated that there could be little or no active biological transmission of these infections mainly due to the harsh arid climate in the region [31]. Future surveys should encompass random (for STH) and a mixture of both random and purposive (for schistosomiasis) selection of sites to further verify these results. However, considering the results, Kitui and Makueni counties in the Eastern Region and Taita Taveta County in the upper part of the Coast Region, which all had STH prevalence below 1%, could benefit from development and implementation of surveillance strategies including careful monitoring of coverage and compliance, until there is further evidence that planning to stop MDA would not be followed by resurgence of infection. Six counties, Bungoma, Kisumu, Migori, Kilifi, Kwale and Mombasa with STH prevalence between 1 to <10%, and Kericho County with prevalence between 10 to <20% would need to continue MDA once every year. Eight counties with prevalence between 20 to <50% would be required to maintain the previous MDA pattern (which is also once every year) under the current treatment guidelines (Fig  4, [11]). Heterogeneity within counties for schistosomiasis indicated that prevalence was zero in ten counties, below 1% in four, between 1% and 10% in five and above 10% in only Busia County. Again noting that survey power being for the overall NSBD geographic area rather than county-level, plus the above mentioned recommendation for some new sites to be included in future surveys of schistosomiasis, means that care must be exercised before making a full recommendation that MDA treatment frequencies for schistosomiasis could change. However, again this picture indicates a lot of potential that such decisions may be able to be arrived at in future. According to the WHO guidelines for schistosomiasis, these decisions could in future be as follows: counties with zero prevalence could potentially consider stopping MDA, and counties with prevalence of below 1% would require serological surveys to determine cases of infection [5]; if from these serological surveys, positive cases are found then MDA would be recommended once after every two years. Counties with prevalence ranging between 1 to <10% for S. mansoni and S. haematobium respectively, could potentially conduct MDA once after every two years. Any county with prevalence of between 10 to < 50% would be warranted to maintain the previous MDA schedule of once every year (Fig 5, [5]). The other important observation to note from these low levels of schistosomiasis prevalence is that future large-scale surveys to assess schistosomiasis will not be very effective, as it will not be possible to power them to detect disease. For a highly focal infection it is instead likely that recent evidence of where schistosomiasis is in the country will be brought to bear, possibly with a change of surveying strategies towards more focal mapping (precision mapping) in the known endemic areas.

WASH factors and their association with STH and schistosomiasis
Examination of the school and household WASH infrastructure as well as children's behaviour and the infection prevalence was done across two domains; (i) helminth species, and (ii) WASH exposure levels i.e. individual-level, household-level and school-level exposures. Generally, coverage and access to key WASH indicators at home and school was relatively good, but was not always associated with lower infection prevalence and the associations were not always consistent between helminth species, likely reflecting both the different pathways of exposure between the species [32] and also some of the complexities in measuring WASH associations with helminth outcomes. Other studies have investigated the association between WASH and STH or schistosome species in Kenya [32][33][34][35][36], but few of them if any had assessed these associations for both individual, household and school level exposures at a national-level. Although no significant associations between the WASH factors and any of the evaluated schistosomiasis prevalence were reported, driven primarily by lack of sufficient observations of schistosomiasis-positive individuals to measure these associations, past studies have investigated and reported possible significant risk factors associated with schistosomiasis [34,35,[37][38][39][40][41].
The association of A. lumbricoides with soil-eating behaviour is not surprising since it is typically acquired through ingestion of helminth egg-contaminated food or untreated water, and so eating of soil (presumably contaminated soil) exposes individuals to high risk of A. lumbricodes as well as T. trichiura infections [42]. Walking bare foot has been previously reported to mostly increase the risk of hookworm infections since hookworm, unlike A. lumbricoides, is primarily acquired through skin contact especially walking bare footed on contaminated soil [43][44][45]. However, a recent study on a sub-cohort of school pupils participating in the Kenyan NSBD program similarly found that children not wearing shoes were equally predisposed to A. lumbricoides infection [32]. Additionally, we found that younger children (<5 years), were 4-fold more likely to be infected with STH while those in the SAC age group (5-14 years) were 3-fold more likely to be infected. Whilst traditionally children below five years were excluded from treatment due to them not attending school and a lack of suitable drug formulation, this finding reiterates the need to widen efforts within the deworming program to cover the PSAC cohort in Kenya. Even though helminth control efforts have been traditionally directed towards SAC, this finding echoes other evidence from within and outside Kenya over the past few years of the high risk of STH infections among PSAC, hence the need to include and strengthen deworming activities in this age group [46][47][48]. Importantly, a higher infection prevalence among this particular age group indicates the underlying risk of infection in a given area, which then become visible when a particular vulnerable age group is not treatment.
The study findings at household level of high number of household occupants, as well as children with no or low level of parents/guardians education being at significant risk of STH infections, points to possible increased risk from household overcrowding, which, especially in the rural areas could feasibly result in poor household hygiene practices and inadequate latrine facilities and safe water for drinking. When parents/guardians have no or low level of education, then the level of hygiene education they offer to their children might also be limited. Both these features are markers of general poverty and thus may be reflecting this well-established influencer of STH and schistosomiasis infection. From this result, it is important for control programs to emphasize continual strengthening of community-wide health education programs delivered through schools, health centres, places of worship, village/community meetings and such like channels to improve on the community understanding on STH control and prevention measures [49][50][51][52][53].
At school level, self-reported school absenteeism was the only notable significant risk factor, showing a gradient association with STH infections specifically A. lumbricoides, implying that, whilst one-day absence was not significantly associated with increased odds of STH infections, two-day and more than two-day absence were increasingly significantly associated with these infections. This result is consistent with previous studies that had found strong association between school absenteeism and A. lumbricoides and hookworm infections but not T. trichiura infection [54][55][56]. Whereas this study reports significant association between STH infections and school absenteeism, most studies have acknowledged difficulty in exclusively quantifying the effect of STH infections on school absenteeism, mainly attenuated by the low levels of infection intensity in the sample population [57][58][59]. However, a post-trial non-randomized study has previously shown that pupils with mostly moderate-to-heavy STH infections especially for A. lumbricoides were found to miss up to 2.4% of school days during follow-up periods compared to their uninfected counterparts [60]. Our study provides associational evidence of an effect of STH infections on school absenteeism among SAC. Evidence from robust randomized control trials may be needed to conclusively assess the effect of STH infections on key educational outcomes [59]. However, such intervention trials may be impossible to conduct over sufficient time periods to assess deworming impacts in the manner that such programs are delivered in real-world settings (i.e., repeated rounds administered throughout childhood) [61].
Additionally, the study found some factors to be protective against STH infections. Availability of tissue/newspaper/water for anal cleansing at home significantly lowered the odds of STH infections especially for A. lumbricoides. Proper use of tissue or water for anal cleansing has previously been reported as an important deterrent to STH infections [62]; if the cleansing material is used improperly by children or if inconsistently or completely unavailable, then anal cleansing after defecation may be done using one's hand, which can lead to fecal contamination of the hands and thus exposure to helminth infection [32,63]. Children who reported always use of school drinking water were found to have lower odds of T. trichiura. Availability of drinking water to children especially when in school is key in controlling STH infections especially A. lumbricoides and T. trichiura since these infections are transmitted mainly through ingestion of contaminated food items like vegetables, fruits or drinking water [64]; this finding is consistent with other previous studies conducted among primary school children [8,32,65]. Children from households possessing assets such as television sets and electricity had lower odds of any STH infection. These household assets like television sets can be viewed as sources of information and can be used to increase the spread of STH advocacy, awareness and control messages through mass media adverts and talk shows [66]. More broadly than this, these are variables associated with increased household wealth relative to those households that do not have such items, providing more information on the likely underlying associations of these diseases with poverty. Interventions aimed at improving hygiene practices like proper anal cleansing after defecation, and access to safe drinking water, as well as continuing socioeconomic improvements, should be highlighted as recommendations for the long-term control of STH infections.

Study limitations
This study was not without limitations. First, individual and household WASH indicators were self-reported by the sampled children at school and not directly collected or observed at home, thus this could potentially lead to reporting bias especially where children are young to recall some of the questions or provide the answer they believe the interviewer wants to hear. Secondly, since the infection prevalence has substantially reduced following several rounds of MDA delivered in the past years, building of robust multivariable models was limited by low prevalence (resulting in few positive values needed to run the models) of infection at Year 6 of the program. Thirdly, whilst the use of purposively sampling of schools within counties was deemed adequate since only schools with highest STH prevalence were targeted, representativeness and generalizability of our results might have been reduced by use of this sampling technique. Lastly, the program used the Kato-Katz technique for examination of STH and S. mansoni eggs in line with the WHO guidelines for examination of these infections in high endemic settings [16], however, this technique has been proven to be less sensitive in low endemic areas [26]. Therefore, our prevalence estimates could have under-estimated the true population prevalence.

Conclusions
After five rounds of treatment, this impact assessment shows low levels of both STH and schistosomiasis in the National School-Based Deworming Program in Kenya. Ascaris lumbricoides is still the leading STH infection followed by T. trichiura and hookworm among SAC populations. The northern counties, Wajir and Garissa, showed zero prevalence of STH and schistosomiasis, an indication that there may be no ongoing biological transmission of the infections in the region. There was observed infection heterogeneity within counties, warranting careful investigation from this time going forward to determine whether reductions in treatment frequency can be proposed in some counties in future. Our assessment of the individual, household and school WASH practices and behaviour on STH and schistosomiasis prevalence suggested mixed associations and differed across individual helminth species; this is expected given the different mechanisms of infections. Based on the survey findings, the Kenyan NSBD program may wish to adopt county-level treatment frequencies based on the WHO guidelines, prioritize strategies to increase treatment coverage amongst the PSAC population, and incorporate integrated control approaches emphasizing health education and WASH interventions to the communities and schools.
Supporting information S1