Mwandawiro et al. Parasites & Vectors 2013, 6:198 102537 http://www.parasitesandvectors.com/content/6/1/198 RESEARCH Open Access Monitoring and evaluating the impact of national school-based deworming in Kenya: study design and baseline results Charles S Mwandawiro1*, Birgit Nikolay2, Jimmy H Kihara1,3, Owen Ozier4, Dunstan A Mukoko3, Mariam T Mwanje3, Anna Hakobyan5, Rachel L Pullan2, Simon J Brooker2,6 and Sammy M Njenga1 Abstract Background: An increasing number of countries in Africa and elsewhere are developing national plans for the control of neglected tropical diseases. A key component of such plans is school-based deworming (SBD) for the control of soil-transmitted helminths (STHs) and schistosomiasis. Monitoring and evaluation (M&E) of national programmes is essential to ensure they are achieving their stated aims and to evaluate when to reduce the frequency of treatment or when to halt it altogether. The article describes the M&E design of the Kenya national SBD programme and presents results from the baseline survey conducted in early 2012. Methods: The M&E design involves a stratified series of pre- and post-intervention, repeat cross-sectional surveys in a representative sample of 200 schools (over 20,000 children) across Kenya. Schools were sampled based on previous knowledge of STH endemicity and were proportional to population size. Stool (and where relevant urine) samples were obtained for microscopic examination and in a subset of schools; finger-prick blood samples were collected to estimate haemoglobin concentration. Descriptive and spatial analyses were conducted. The evaluation measured both prevalence and intensity of infection. Results: Overall, 32.4% of children were infected with at least one STH species, with Ascaris lumbricoides as the most common species detected. The overall prevalence of Schistosoma mansoni was 2.1%, while in the Coast Province the prevalence of S. haematobium was 14.8%. There was marked geographical variation in the prevalence of species infection at school, district and province levels. The prevalence of hookworm infection was highest in Western Province (25.1%), while A. lumbricoides and T. trichiura prevalence was highest in the Rift Valley (27.1% and 11.9%). The lowest prevalence was observed in the Rift Valley for hookworm (3.5%), in the Coast for A. lumbricoides (1.0%), and in Nyanza for T. trichiura (3.6%). The prevalence of S. mansoni was most common in Western Province (4.1%). Conclusions: The current findings are consistent with the known spatial ecology of STH and schistosome infections and provide an important empirical basis on which to evaluate the impact of regular mass treatment through the school system in Kenya. Keywords: Soil-transmitted helminths, Schistosomiasis, School-based deworming, Monitoring and evaluation, Kenya * Correspondence: cmwandawiro@kemri.org 1 Eastern and Southern Africa Centre of International Parasite Control (ESACIPAC), Kenya Medical Research Institute (KEMRI), P.O Box 54840–00200, Nairobi, Kenya Full list of author information is available at the end of the article © 2013 Mwandawiro et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 2 of 14 http://www.parasitesandvectors.com/content/6/1/198 Background Centre of International Parasite Control (ESACIPAC) at In Africa, an increasing number of countries have the Kenya Medical Research Institute (KEMRI) and the launched national Neglected Tropical Diseases (NTDs) Japan International Cooperation Agency [6]. In addition Strategic Plans, with Kenya being among the first to to the operational experience in implementing SBD launch its multi-year plan in November 2011. The main programmes in Kenya, there exists a clear policy con- strategy of these plans in tackling NTDs in Africa is the text, with school health policy and guidelines developed delivery of preventative chemotherapy, including school- jointly by the Ministry of Public Health and Sanitation based deworming (SBD) to control soil-transmitted and the Ministry of Education [7,8]. In 2011, the na- helminths (STH) and schistosomiasis. There are also tional government programme (with technical support attempts to integrate mass drug administration (MDA) from Deworm the World) received five years of funding for STH and schistosomiasis with other NTDs, includ- from the Children Investment Fund Foundation and in ing lymphatic filariasis and onchocerciasis. 2012, an estimated 4.6 million school children received The STHs, Ascaris lumbricoides, Trichuris trichiura albendazole treatment for STH infection (Government and hookworms, are estimated to infect over 1 billion of Kenya, pers. comm.). individuals worldwide and in 2010 caused 5.18 million This paper describes the overall study design of the disability adjusted life years (DALYs), while schisto- monitoring and evaluation (M&E) of the Kenya national somiasis contributes 3.31 million DALYs [1]. Chronic school-based deworming programme and presents re- infections can have insidious effects on childhood devel- sults from the baseline survey. Patterns of infection by opment, including growth and cognitive development, age, sex and geography are also reported. whilst heavy infections may result in serious clinical dis- ease. Both chronic and intense infections are most com- Methods mon in school-age children who are the natural targets Study design for school-based chemotherapy programmes. Reaching The M&E includes a series of pre- and post-intervention, the school-aged population is most effectively achieved repeat cross-sectional surveys in a representative, stra- via the school infrastructure, and SBD programmes have tified, two-stage sample of schools across Kenya. Dis- been shown to be a simple and cost-effective strategy to trict stratification was based on both geography and reduce the disease burden of STH [2]. In 2001, the anticipated infection prevalence. The programme con- World Health Assembly endorsed the WHA 54.19 reso- tains three tiers of monitoring: i) a national baseline lution that urged countries to control morbidity due to survey including 200 schools in 20 districts, which STH infection through regular deworming of school- aims to establish an accurate national measurement of aged children, setting a target to deworm 75% of the infection levels; ii) surveys conducted pre and post school children [2]. In support of this resolution, an intervention (pre-post surveys), which monitor 60 of important recent development was the large scale dona- the 200 schools before and immediately after the tion of deworming drugs by pharmaceutical companies deworming activity to evaluate reductions in infec- in 2010, with GlaxoSmithKline donating 400 million tions that can be directly attributed to programme im- albendazole tablets per year and Johnson & Johnson plementation; and iii) high frequency surveys in 10 donating 200 million mebendazole tablets per year. Con- schools, distinct from the 60 pre-post schools, at four comitantly, there has been increased demand for govern- time points in a single year, before, during, and after ment led SBD programmes and donors are willing to treatment (Figure 1). fund national programmes either as part of school health Two hundred schools were examined at baseline and programmes or integrated NTD control programmes. will be re-examined in year 3 and 5 in order to monitor As programmes are scaled-up there is a scientific im- long-term changes in worm infection at a national level perative to monitor the efficacy of treatment and to both in terms of prevalence and intensity of infection. rigorously document the impact of treatment on infec- This sample size was chosen in order to be able to tion and health outcomes. detect a five-percentage-point change in prevalence In Kenya, a national SBD programme was launched in across years, assuming power β = 0.80 and test size α = 2009, with financial support from the Ministry of Educa- 0.05, and considering the anticipated variance in preva- tion and technical support and drugs provided by lence. Sixty schools (a subset of the 200) will be surveyed Deworm the World and the Partnership for Child Devel- every year for 5 years, before each treatment round opment. The programme successfully treated over 3.6 to evaluate programme impact and 3–5 weeks post- million school-age children. This programme built on treatment to evaluate treatment efficacy [2]. The same previous pilot programmes in Kenya [3-5], including a schools will be surveyed each year in the 60 pre-post pilot school health programme in Mwea District in cen- survey schools, whereas in the remaining 130 schools, a tral Kenya supported by the Eastern and Southern Africa different random sample of schools will be undertaken Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 3 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 1 Schematic of M&E programme design. Three tiers of monitoring are conducted: i) a national baseline survey; ii) pre-post surveys, and iii) high frequency surveys. each year. In the 10 high frequency schools, a cohort of Ethical approval children will be followed-up longitudinally and assessed Ethics approval for this study was granted by the KEMRI for haemoglobin concentration in addition to parasito- Ethics Review Committee in Kenya. logical outcomes. The 200 schools were selected based on the geograph- Data collection ical distribution of population and STH endemicity. The baseline surveys were conducted between January Based on available data and predictive maps [9,10], STH and April 2012. In each school, 18 children (9 girls was assumed to be endemic in 66 districts. From these and 9 boys) were sampled randomly from each of six clas- districts, grouped into strata, 20 districts were randomly ses - one Early Childhood Development (ECD) class and selected for M&E in the first sampling stage, with number classes 2–6 - using computer generated random num- of districts per province proportional to population: ber tables, for a total of approximately 108 per school. six districts from Western Province, three from the The sampling within these specified classes aimed Rift Valley, five from the Coast, and six from Nyanza to target children aged 5–16 years. Stool samples (Table 1). At the second sampling stage, primary were obtained for each child and two slides prepared schools were randomly selected from within the chosen and examined for the presence and intensity of STH 20 districts. species and S. mansoni using the Kato Katz method, with the concentration of eggs expressed as eggs per gram (epg) of faeces. Urine samples were obtained Table 1 Number of schools to be sampled in each only from children in Coast Province (where Schistosoma province for each tier of monitoring haematobium is widespread) and investigated for pre- Province National baseline Pre-post High-frequency Totals sence and intensity of S. haematobium using the urine Western 39 18 3 60 filtration method, with the concentration of S. haema- Rift Valley 19 9 1 30 tobium eggs estimated in eggs per 10 ml urine. Egg counts Coast 33 15 4 50 were performed only up to 24,000 epg and 1,000 eggs/ Nyanza 39 18 2 60 10 ml urine, respectively. Infection intensities above these values were, therefore, not further quantified. In the 10 Totals 130 60 10 200 “high frequency” schools, finger-prick blood samples were Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 4 of 14 http://www.parasitesandvectors.com/content/6/1/198 obtained and analysed using a HemoCue photometer The presence of spatial autocorrelation in prevalence (HemoCue, Angelhom, Sweden) to estimate haemoglobin data was investigated using Moran’s I statistic. Moran’s concentration. The geographical coordinates of schools I coefficient of autocorrelation is a measure of similarity were collected using an eTrex global positioning system of outcome variables among spatially related areas, using (Garmin Ltd., Olathe, KS, US). a weighted matrix to define spatial relationships. A coef- ficient of 0 indicates the null hypothesis of no clustering, Geographic data while a positive coefficient indicates positive spatial au- Survey data were linked by school location and mapped tocorrelation [19]. Therefore, large-scale spatial effects using ArcGIS 9.3.1 (Environmental Systems Research were first removed using binary logistic regression Institute Inc. Redlands, CA, US). A set of ecological models that modelled prevalence as a function of sur- and climatic covariates were assembled from a variety of vey location and environmental characteristics. To sources and linked to school locations. Estimates of reduce the number of variables and to prevent multi- monthly mean temperature and annual precipitation at collinearity, variables were first subjected to factor ana- 1 km resolution were obtained from WorldClim [11] and lysis. Five factors with Eigenvalues greater than 1 were elevation data at 100 m resolution were obtained from the retained: factor 1 was mainly determined by mean, Consortium for Spatial Information [12]. Monthly mean, minimum, maximum temperature and altitude, factor 2 maximum and minimum NDVI estimates at 1 km reso- by Euclidean distance to any and permanent water bod- lution for 2011 were retrieved from the SPOT Vegetation ies, factor 3 by mean and maximum NDVI, factor 4 by Programme [13] at a resolution of 1 km. Location of rivers distance to any and permanent rivers, and factor 5 by and water bodies were obtained from the Digital Chart of population density and population density < 5 year- the World [14] and Euclidean distances were calculated at olds. Factor values for each school were predicted and 50 m resolution. Population density estimates at 1 km included into the binomial regression models. Land resolution were obtained from the Afripop project [15] cover, NDVI standard deviation, and annual precipita- and land cover information from the GlobCover project tion did not strongly contribute to these factors and [16] at a 300 m resolution. were, therefore, included additionally into the models. The resultant normally distributed Pearson residuals Statistical analysis from the regression analysis were used to estimate the Observed prevalence of each STH species was calculated Moran’s I statistic. All statistical analyses were carried at school, district and province levels, and 95% confidence out using STATA version 12.0 (STATA Corporation, intervals (95% CI) were obtained by binomial logistic re- College Station, TX, US). gression, taking into account clustering by schools. Com- parisons of prevalence by location, age group and sex were Results tested for significance on the basis of the Wald test. For Overall, data were collected from 21,528 children, the purposes of analysis, the following age groups were including 4,944 (23.0%) from Coast Province, 3,658 used: 3–5, 6–7, 8–9, 10–11, 12–13, and >14 year-olds. (17.0%) from Rift Valley Province, 6,018 (28.0%) from Mean egg counts were expressed as arithmetic mean epg Western Province, and 6,908 (32.1%) from Nyanza Province and since egg counts are not normally distributed, 95% (Table 2). Errors in recording information in the field CIs were estimated using a negative binomial regression meant that information on age and sex of children were model taking into account clustering. Anaemia was de- available for 21,312 children (99.0%) and 21,342 chil- fined using age- and sex-corrected WHO thresholds [17], dren (99.1%), respectively. The mean age of children adjusted by altitude [18]. Infection intensities were further was 9.8 years (standard deviation, SD 2.8 years) and the classified into moderate-heavy infections according to age range was 3 to 21 years. The percentage of boys WHO guidelines [2] and the prevalence of moderate- (49.9%) and girls (49.2%) was comparable. heavy infections and 95% CIs were calculated by binomial logistic regression taking into account clustering. Ninety Soil-transmitted helminths five percent CIs of mean haemoglobin concentrations and Overall, 32.4% (95% CI 30.1-34.8%) were infected with at prevalence of anaemia were obtained by linear and bino- least one STH species. A. lumbricoides was the most mial regression analysis, respectively, both taking into ac- prevalent STH species (18.0%, 95% CI 15.7- 20.6%), count clustering by schools. The associations of any or followed by hookworm (15.6%, 95% CI 13.7-17.7%) and moderate-heavy infections with anaemia were investigated then T. trichiura (6.6%, 95% CI 5.4- 8.1%). The overall by binomial regression analysis adjusting for age and sex mean intensity of A. lumbricoides was 1,653 epg (95% CI of children and taking into account clustering. Compari- 1372–1991), mean hookworm intensity was 64 epg (95% sons were tested for significance using the Wald-test fol- CI 51–81) and mean T. trichiura intensity was 33 epg lowing regression analysis. (95% CI 10–104). Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 5 of 14 http://www.parasitesandvectors.com/content/6/1/198 Table 2 Prevalence and intensity of infections, by province and range of school-level prevalence Province/species N examined N infected % infected (95%CI1) School prevalence range (%) Mean epg (95%CI2) Coast Hookworm 4944 898 18.2 (14.0- 23.5) 0-59.3 64 (34–121) A. lumbricoides 4944 51 1.0 (0.7-1.6) 0-8.3 29 (15–54) T .trichiura 4944 390 7.9 (5.7-10.9) 0-44.4 10 (5–21) Rift Valley Hookworm 3658 129 3.5 (2.1 -6.0) 0-21.9 18 (6–57) A. lumbricoides 3658 992 27.1 (21.9-33.5) 0-55.6 3445 (2707–4385) T. trichiura 3658 435 11.9 (7.7-18.3) 0-64.4 31 (17–57) Western Hookworm 6018 1,513 25.1 (21.4-29.5) 0-57.9 144 (116–177) A. lumbricoides 6018 1459 24.2 (20.4-28.7) 0.9-63.0 1616 (1293–2020) T. trichiura 6018 351 5.8 (3.8-8.8) 0-40.7 17 (10–28) Nyanza Hookworm 6908 812 11.8 (9.8-14.1) 0-35.2 19 (14–26) A. lumbricoides 6908 1373 19.9 (16.0-24.7) 0-71.3 1898 (1356–2655) T. trichiura 6908 250 3.6 (2.6-5.1) 0-26.9 64 (10–399) Total Hookworm 21528 3352 15.6 (13.7-17.7) 0-59.3 64 (51–81) A. lumbricoides 21528 3875 18.0 (15.7- 20.6) 0-71.3 1653 (1372–1991) T. trichiura 21528 1426 6.6 (5.4- 8.1) 0-64.4 33 (10–104) 1 Confidence intervals obtained by binomial regression taking into account school clusters. 2 Confidence intervals obtained by negative binomial regression taking into account school clusters. Overall, the prevalence and intensity of hookworm was infection was lower among boys than girls (27 epg, 95%CI higher in boys than girls (16.8%, 95% CI 14.8-19.0 vs. 10–74 vs 40 epg, 95% CI 11–139, p = 0.002), while the in- 14.5%, 95% CI 12.6-16.7, p < 0.001, 70 epg, 95% CI 55–88 tensity of A. lumbricoides did not differ significantly by sex vs. 58 epg, 95% CI 44–76, p = 0.015). Prevalence of A. (p = 0.310). Figure 2 presents prevalence and intensity of lumbricoides and T. trichiura did not differ significantly by infection by age and sex. For both boys and girls, the sex (p = 0.527 and p = 0.701). The intensity of T. trichiura prevalence of hookworm varied significantly by age group Figure 2 Prevalence of infection and mean intensity of infection by age and sex for hookworm (A, D), A. lumbricoides (B, E), and T. trichiura (C, F). 95% Confidence intervals were obtained by binomial regression and negative binomial regression, respectively taking into account school clusters. Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 6 of 14 http://www.parasitesandvectors.com/content/6/1/198 (p < 0.001), with prevalence highest among >14 year-olds; Table 3 Moran’s I statistic (I) of hookworm, A. however, hookworm intensity did not vary significantly by lumbricoides, T.trichiura, S. mansoni, and S. haematobium age group (p = 0.250 and p = 0.051). Both the prevalence prevalence spatial autocorrelation after removal of and intensity of A. lumbricoides varied significantly by age large-scale environmental effects (p < 0.001), among boys and girls, and was highest among Coast Western/Rift Valley/ Nyanza 6–7 year-olds. The prevalence of T. trichiura did not vary I p-value I p-value by age group (p = 0.355 for boys and p = 0.284 for girls), Hookworm 0.118 <0.045 0.168 <0.001 whereas the intensity of T. trichiura did vary significantly A. lumbricoides 0.432 <0.001 0.121 <0.001 by age group (p < 0.001), with intensity being highest T. trichiura 0.092 0.077 0.067 0.008 among 3–7 year-old male children and 8–9 and >14 year- S. mansoni 0.221 0.002 0.030 0.113 old female children. Table 2 summarizes the prevalence and intensities of S. haematobium −0.055 0.414 NA NA STH infection by province. Patterns of STH varied markedly by province (p < 0.001), with hookworm most infection prevalence and intensity by age and gender is prevalent in Western Province and A. lumbricoides and shown in Figure 6. The prevalence (plinear = 0.001 and T. trichiura most common in Rift Valley Province. Inten- plinear = 0.018) and intensity (p < 0.001 and p = 0.031) of sities of hookworm and A. lumbricoides infection also S. mansoni varied significantly by age-group among both varied by province (p < 0.001), but there was only sexes. While prevalence increased linearly with age, suggestive evidence for a difference in T. trichiura inten- intensity of infection was lowest among 3–5 year-olds. sity by province (p = 0.055). The prevalence of moderate- The evidence for a variation in S. haematobium pre- heavy infection intensities was highest for A. lumbricoides valence or intensity by age-group was weak among boys in the Rift Valley (15.7%, 95% CI: 12.4-19.9, p < 0.001) and and girls (prevalence p = 0.308, p = 0.066; intensity: for hookworm in the Coast and Western Proince (0.5%, p = 0.640, p = 0.265). 95% CI: 0.2-1.1 and 0.5, 95% CI: 0.3-0.9, p < 0.001). The Prevalences and intensities of Schistosoma infections overall prevalence of moderate-heavy T. trichiura infec- are summarized in Table 4. S. mansoni infection preva- tions was 0.3% (95% CI: 0.2-0.6, p = 0.133). Figures 3,4,5 lence and intensity varied by provinces (p < 0.001) and present the geographical variation of prevalence and both were highest in Western Province. The prevalence intensity of infection by school and by district. Prevalence of moderate-heavy S. mansoni infection intensity was of infection varied markedly by school across the coun- also highest in Western Province (3.4%, 95% CI: 1.5-8.1. try: A. lumbricoides ranged from 0–71.3%; hookworm from 0–59.3%, T. trichiura from 0–64.4% (Table 2 and Table 4 Prevalence and intensity of Schistosoma mansoni and S. haematobium, by province and variation by school Figures 3A-5A). Prevalence by district varied from 0.3- 45.5%, 0.2-44.3% and 0.2-30.2%, respectively (Figures 3D- N N % School Mean examined infected infected prevalence epg (95% 5D). After removal of large-scale environmental effects, 1 (95%CI ) range (%) CI2) small-scale spatial patterns of prevalence, as indicated by Coast Moran’s I statistic of spatial autocorrelation could be S. mansoni 4944 1 0.0 (0.0- 0-0.9 0 (0–0) observed for hookworm (p = 0.045, p < 0.001) and A. 0.1) lumbricoides (p < 0.001) in all regions, and for T. trichiura S. 3019 448 14.8 0-37.0 16 (10–26) in Western, Rift Valley and Nyanza Provinces (p = 0.008) haematobium (11.3- (Table 3). 19.5) Rift Valley Schistosome infection S. mansoni 3658 13 0.4 (0.1- 0-12.0 1 (0–9) The overall prevalence of S. mansoni was 2.1% (95% CI 2.5) 1.2- 3.5) and the mean infection intensity was 12 epg Western (95%CI 4–36). In the Coast Province, where urine sam- S. mansoni 6018 246 4.1 (1.9- 0-64.8 40 (12– ples were collected, the prevalence of S. haematobium 9.0) 126) was 14.8% (95% CI 11.3-19.5) and the mean infection Nyanza intensity was 16 eggs/10 ml urine (95% CI 10–26). S. mansoni 6908 190 2.8 (1.5- 0-47.2 3 (1–7) There was no significant difference by sex in the 5.0) prevalence of S. mansoni (p = 0.868), prevalence of S. Total haematobium or intensity of S. mansoni (p = 0.362), S. mansoni 21528 450 2.1 (1.2- 0-64.8 12 (4–36) whereas the intensity of S. haematobium was higher 3.5) among boys than girls (32 eggs/10 ml, 95%CI 14–74 vs 1 95%CIs obtained by binomial regression taking into account school clusters. 12 eggs/10 ml, 95%CI 7–20, p = 0.033). Schistosoma 2 95%CIs obtained by binomial regression taking into account school. Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 7 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 3 Spatial distribution of hookworm prevalence and intensity, by school and district. Spatial distribution of hookworm school prevalence (A) and average school infection intensity (C), school prevalence distribution (B), and average district prevalence (D). p < 0.001). There was also marked variability by school Anaemia and district: school prevalence ranged from 0–64.8% for Finger-prick blood samples were collected from 492 S. mansoni and 0-37% for S. haematobium; and mean children from 10 schools. Although too small to be district prevalences were 0–27.5% and 10-23%, respect- meaningfully representative of the country, these ten ively (Figures 7 and 8). schools were chosen to include schools in each of the After removing environmental large-scale effects, posi- four provinces where STH endemicity had been pre- tive spatial autocorrelation was observed only for S. dicted. In these ten schools, the overall prevalence of an- mansoni prevalence in the Coast Province (p = 0.002), aemia was 31.3% (95% CI 21.7-45.1%) and the overall while the evidence for small-scale spatial structures was mean haemoglobin concentration was 12.6 g/dL (95% CI weak for S. mansoni in the other areas (p = 0.113) or for S. 12.0-13.2 g/dL). No significant differences were found in haematobium (p = 0.414). Moran’s I statistics of spatial anaemia and haemoglobin concentration by sex (an- autocorrelation are summarised in Table 3. aemia: p = 0.264; haemoglobin: p = 0.947), but there was Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 8 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 4 Spatial distribution of A. lumbricoides prevalence and intensity, by school and district. Spatial distribution of A. lumbricoides school prevalence (A) and average school infection intensity (C), school prevalence distribution (B), and average district prevalence (D). significant variation by age (anaemia: p = 0.046; haemo- positive association of anaemia with any or moderate- globin: p = 0.048), with anaemia decreasing and haemo- heavy hookworm, A. lumbricoides, T. trichiura, or S. globin increasing with increasing age. haematobium infections and moderate-heavy S. mansoni The overall prevalence of STH infection in these 10 infections. schools was 33.3% (95% CI 25.7-43.1%, range of school prevalence: 18.5-65.7%) and the prevalence of schisto- Discussion some infection was 14.3% (95% CI 7.1-28.7%, range of This survey provides an up-to-date assessment of STH school prevalence: 0–34.3%). There was suggestive evi- infections in the regions of Kenya targeted for school- dence for an association of anaemia with S. mansoni in- based deworming, and provides a rigorous basis for fection (p = 0.052) after adjusting for age and sex of evaluating programme impact. The most common STH children. However, there was no strong evidence for a species detected among children was A. lumbricoides, Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 9 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 5 Spatial distribution of T. trichiura prevalence and intensity, by school and district. Spatial distribution of T. trichiura school prevalence (A) and average school infection intensity (C), school prevalence distribution (B), and average district prevalence (D). followed by hookworm and T. trichiura. The prevalence District in Nyanza Province (17.4%) and Kilindini in and intensity of species infection varied markedly at Coast Province (18.0%). However, the ranges of school province, district and schools levels. At an individual prevalence in these districts show spatial heterogeneity level, patterns of infection and intensity varied by age- and schools in western parts of Kisumu and in the group and sex of the children. North of Kilindini exceed indeed the threshold preva- In Kenya, STHs are assumed to be endemic in 66 dis- lence. A deviation from the prediction can be observed tricts that were identified based on historical data and for Transmara District in Rift Valley Province. While the predictive maps created using a Bayesian space-time predictive map suggested prevalence between 10 and geostatistical modelling approach [9,10]. When compar- 20% and no intervention was recommended, the district ing the predictions with the survey data, two districts has the highest mean district STH prevalence in this that were suggested to require MDA have an actual study (53.1%). It is noteworthy that no historical survey mean STH prevalence < 20%. These are Kisumu East data were previously available for this district. Furthermore, Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 10 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 6 S. mansoni and S. haematobium prevalence (A-B) and mean infection intensity (C-D) by age and sex. 95% Confidence intervals of prevalence were obtained by binomial regression and confidence intervals of infection intensity by negative binomial regression, both taking into account school clusters. the newly obtained survey data should be used to revise structure of T. trichiura infection. The observed small- the predictions, especially for areas with previous high scale patterns of infections might originate from specific uncertainty. characteristics of the locations that influence the risk of Interestingly, when comparing the baseline survey re- infection, such as access to water, sanitation and hygiene sults to historical data, STH infection prevalence has (WASH) [35,36]. strongly decreased over the last decade in certain re- Even though the overall prevalence of S. mansoni was gions of Kenya. These are specifically Bunyala District, low, prevalence reached up to 64% in some schools in Western Province, where a deworming programme has Western Province and up to 47.2% in schools in Nyanza been implemented since 1998 [20,21] and Kwale District, Province. This is consistent with data collected between Coast Province, where the National Programme for 1980 and 2009, showing small levels of infection in Elimination of Lymphatic Filariasis led to four rounds of coastal regions and higher prevalence near the shores of Diethylcarbamazine citrate (DEC) and albendazole dis- Lake Victoria [10,37]. Previously collected data for S. tribution [22,23]. In other districts where chemotherapy haematobium infections, however, show higher preva- programmes have not been implemented previously, lence than actually observed in this survey [10]. This such as Homa Bay in Nyanza Province, the prevalence highlights that historical data can serve as an indicator estimates did not change over time [24-32]. for endemic regions, however, programme implementers After removal of large scale environmental effects, need to be aware of a certain degree of uncertainty. small-scale spatial patterns could be observed for hook- As for STH infections, S. mansoni infection varied sig- worm and A. lumbricoides in all regions and T. trichiura nificantly between and within provinces. Additionally, in Western, Rift Valley and Nyanza Provinces. Previous small-scale spatial patterns of S. mansoni were observed studies in Kenya, Uganda, Tanzania, and Zambia showed within Coast Province, which is consistent with studies spatial correlation up to 95–166 km for hookworm and in Cameroon, Mali, and Uganda, where spatial correl- 36–92 km for A. lumbricoides [33,34]. The lack of spatial ation ranged up to 70 km [38]. For Western, Nyanza, correlation of T. trichiura was reported in several other and Rift Valley Provinces, significant evidence for spatial studies, although Sturrock et al. were able to show spatial autocorrelation was observed only before removal of en- dependency up to a distance of 46 km in Kenya’s Coast vironmental effects. Interestingly, there was no spatial Province [33,34]. Nevertheless, the spatial autocorrelation pattern for S. haematobium prevalence before or after p-value for Coast Province in this study is quite small (p = de-trending the data. This is surprising, as studies from 0.077), and might still be indicative of a certain spatial Tanzania and Sierra Leone demonstrated autocorrelation Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 11 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 7 Spatial distribution of S. mansoni prevalence and intensity, by school and district. Spatial distribution of S. mansoni school prevalence (A) and average school infection intensity (C), school prevalence distribution (B), and average district prevalence (D). of S. haematobium infections [39,40]. Nevertheless, frequency samplings are performed, which allow a infection prevalence was associated with proximity to more direct attribution of reductions in infection to water bodies and annual rainfall. programme implementation and help to distinguish The main limitation of this study is the lack of a treatment effects from any other time-varying fluctua- control group to evaluate the effect of the deworming tions in infection rates. A challenge experienced in the programme. As the deworming programme is im- field was the use of paper questionnaires to collect plemented nationally and the ministries of health and data and in rare instances there was inaccurate record- education plans to deworm all children in the 66 ing of age and sex. To overcome this issue, future data endemic districts, an exclusion of groups of children collection intends to use electronic data collection. A from this effort would be unethical. As compensation further study limitation is the use of a single stool for this constraint, series of pre-post and high- sample for Kato-Katz analysis and the truncation of Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 12 of 14 http://www.parasitesandvectors.com/content/6/1/198 Figure 8 Spatial distribution of S. haematobium prevalence and intensity, by school and district. Spatial distribution of S. haematobium school prevalence (A) and average school infection intensity (C), school prevalence distribution (B), and average district prevalence (D). egg counts at a certain maximum level for the meas- Nonetheless, there was evidence of an association be- urement of infection intensities. The sensitivity of tween S. mansoni infection and the risk of anaemia. Kato-Katz analysis might be lower for the detection of light infections than for example formol-ether concen- Conclusions tration techniques and could have been improved by The analysis of collected data provided insight into the processing more than one sample [41]. This poten- current prevalence and distribution of spatial patterns of tially leads to lower than actually detected infection STH and schistosome infections in Kenya and will allow prevalence and mean infection intensities. Finally, be- monitoring of the impact of the national deworming cause anaemia data were only collected from ten programme on the prevalence and intensity of infection schools, we cannot make generalizations across the and on child health outcomes. Furthermore, areas of entire country, and the small sample makes associa- high prevalence of schistosome infection that merit mass tions with helminth infection difficult to assess. drug administration were identified, while other areas Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 13 of 14 http://www.parasitesandvectors.com/content/6/1/198 should be further assessed for S. haematobium infec- control of schistosomiasis haematobia. II. Metrifonate vs. praziquantel in tions. Additional studies using individual data on poten- control of infection-associated morbidity. AmJTrop Med Hyg 1990, 42:587–595. tial risk factors of infections might enable us to explain 4. Magnussen P, Ndawi B, Sheshe AK, Byskov J, Mbwana K, Christensen NO: the observed small-scale spatial patterns of infection and The impact of a school health programme on the prevalence and provide further insights into determinants of STH and morbidity of urinary schistosomiasis in Mwera Division, Pangani District, Tanzania. Trans R Soc Trop Med Hyg 2001, 95:58–64. schistosome infections in Kenya. Monitoring the impact 5. Miguel EA, Kremer M: Worms: Identifying Impacts on Education and of school-based deworming using these baseline findings Health in the Presence of Treatment Externalities. Econometrica 2004, will allow the Government of Kenya to make informed 72:159–217. 6. Kihara JH, Muhoho N, Njomo D, Mwobobia IK, Josyline K, Mitsui Y, Awazawa decisions on the allocation of resources and on what T, Amano T, Mwandawiro C: Drug efficacy of praziquantel and strategies to adopt when infection-level has significantly albendazole in school children in Mwea Division, Central Province, dropped. Kenya. Acta Trop 2007, 102:165–171. 7. Ministry of Public Health and Sanitation and Ministry of Education: National Competing interests School Health Policy. Republic of Kenya: Ministry of Education; 2009. The authors declare that they all have no competing interests. 8. Ministry of Public Health and Sanitation and Ministry of Education: National School Health Guidelines. Republic of Kenya: Ministry of Education; 2009. Authors’ contributions 9. Pullan RL, Gething PW, Smith JL, Mwandawiro CS, Sturrock HJ, Gitonga CW, CSM, OO and SMN designed the study, with support from AH, RLP and SJB. Hay SI, Brooker S: Spatial modelling of soil-transmitted helminth JHK coordinated the data collection and DAM and MTM provided infections in Kenya: a disease control planning tool. PLoS Negl Trop Dis parasitological expertise. BN conducted the data analysis. CSM, BN and SJB 2011, 5:e958. wrote the first draft and all authors read and approved the final version of 10. Brooker S, Kabatereine NB, Smith JL, Mupfasoni D, Mwanje MT, the manuscript. All authors read and approved the final manuscript. Ndayishimiye O, Lwambo NJ, Mbotha D, Karanja P, Mwandawiro C, et al: An updated atlas of human helminth infections: the example of East Africa. Int J Health Geogr 2009, 8:42. Acknowledgements We acknowledge with appreciation the support from the Ministry of 11. WorldClim- Global Climate Data. www.worldclim.org/bioclim. Education, Ministry of Public health and Sanitation and the local provincial 12. Consortium for Spatial Information - CGIAR-CSI. http://srtm.csi.cgiar.org. and district administration in Coast, Western, Nyanza and Rift Valley 13. SPOT Vegetation Programme. http://www.spot-vegetation.com. Provinces. We thank the head teachers, teachers and students and parents in 14. Digital Chart of the World. www.diva-gis.org. each of the schools that participated in this study and are grateful to the 15. Afripop project. http://www.clas.ufl.edu/users/atatem/index_files/AfriPop.htm. fieldworkers, project staff and laboratory technologists for their assistance in 16. GlobCover. http://dup.esrin.esa.int/globcover/. field operations that included sample collection and examination. This study 17. Benoist B, McLean E, Egli I, Cogswell M: Worldwide prevalence of anaemia received great support from the Division of Vector-Borne and Neglected 1993–2003: WHO global database on anaemia. Geneva: WHO; 2008. Tropical Diseases’ (DVBNTD’s) district facilities and personnel for which we 18. Sullivan KM, Mei Z, Grummer-Strawn L, Parvanta I: Haemoglobin are highly grateful. The fieldwork was financially supported by the Children’s adjustments to define anaemia. Trop Med Int Health 2008, 13:1267–1271. Investment Fund Foundation (CIFF). SJB is supported by a Wellcome Trust 19. Pfeiffer DU, Robinson TP, Stevenson M, Stevens KB, Rogers DJ, Clements AC: Senior Fellowship in Basic Biomedical Science (098045), which also supports Spatial Analysis in Epidemiology. Oxford: OUP; 2008. RLP. BN is supported by a grant from the Bill & Melinda Gates Foundation. 20. Baird S, Hicks JH, Kremer M, Miguel EA: Worms at Work: Long-run Impacts of This paper is published with the permission of the Director of the Kenya Child Health Gains. Berkeley: Department of Economics at the University of Medical Research Institute (KEMRI). The findings, interpretations and California; 2011. conclusions expressed in this paper are entirely those of the authors, and do 21. Brooker S, Miguel EA, Moulin S, Luoba AI, Bundy DA, Kremer M: not necessarily represent the views of the World Bank, its Executive Directors, Epidemiology of single and multiple species of helminth infections or the governments of the countries they represent. among school children in Busia District, Kenya. East Afr Med J 2000, 77:157–161. Author details 22. Kenya Ministry of Public Health and Sanitation: Annual Report of the 1 Eastern and Southern Africa Centre of International Parasite Control National Programme to Eliminate Lymphatic Filariasis. Nairobi: Division of (ESACIPAC), Kenya Medical Research Institute (KEMRI), P.O Box 54840–00200, Vector-borne and Neglected Tropical Diseases; 2012. Nairobi, Kenya. 2Faculty of Infectious and Tropical Diseases, London School 23. Njenga SM, Mwandawiro CS, Muniu E, Mwanje MT, Haji FM, Bockarie MJ: of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Adult population as potential reservoir of NTD infections in rural villages Kingdom. 3Division of Vector-borne and Neglected Tropical Diseases, Ministry of Kwale district, Coastal Kenya: implications for preventive of Public Health and Sanitation, P.O. Box 19982–00202, Nairobi, Kenya. chemotherapy interventions policy. Parasit Vectors 2011, 4:175. 4 24. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1983. Development Research Group, The World Bank, 1818 H Street NW, Washington, D.C 20433, United States of America. 5Children’s Investment 25. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1984. Fund Foundation, London, United Kingdom. 6KEMRI-Wellcome Trust 26. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1985. Research Programme, P.O. Box 43640–00100, Nairobi, Kenya. 27. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1986. 28. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1987. Received: 21 February 2013 Accepted: 27 June 2013 29. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1988. Published: 5 July 2013 30. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1995. 31. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1996. References 32. Division of Vector Borne Diseases: Annual Report. Nairobi: Ministry of Health; 1997. 1. Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud C, Ezzati M, 33. Brooker S, Kabatereine NB, Tukahebwa EM, Kazibwe F: Spatial analysis of Shibuya K, Salomon JA, Abdalla S, et al: Disability-adjusted life years the distribution of intestinal nematode infections in Uganda. Epidemiol (DALYs) for 291 diseases and injuries in 21 regions, 1990–2010: a Infect 2004, 132:1065–1071. systematic analysis for the Global Burden of Disease Study 2010. Lancet 34. Sturrock HJ, Gething PW, Clements AC, Brooker S: Optimal survey designs 2013, 380:2197–2223. for targeting chemotherapy against soil-transmitted helminths: effect of 2. World Health Organization Expert Committee: Prevention and control of spatial heterogeneity and cost-efficiency of sampling. AmJTrop Med Hyg schistosomiasis and soil-transmitted helminthiasis. World Health Organ 2010, 82:1079–1087. Tech Rep Ser 2002, 912:1–57. 35. Soares Magalhaes RJ, Barnett AG, Clements AC: Geographical analysis of 3. King CH, Lombardi G, Lombardi C, Greenblatt R, Hodder S, Kinyanjui H, the role of water supply and sanitation in the risk of helminth infections Ouma J, Odiambo O, Bryan PJ, Muruka J, et al: Chemotherapy-based of children in West Africa. Proc Natl Acad Sci USA 2011, 108:20084–20089. Mwandawiro et al. Parasites & Vectors 2013, 6:198 Page 14 of 14 http://www.parasitesandvectors.com/content/6/1/198 36. Ziegelbauer K, Speich B, Mausezahl D, Bos R, Keiser J, Utzinger J: Effect of sanitation on soil-transmitted helminth infection: systematic review and meta-analysis. PLoS Med 2012, 9:e1001162. 37. Odiere MR, Rawago FO, Ombok M, Secor WE, Karanja DM, Mwinzi PN, Lammie PJ, Won K: High prevalence of schistosomiasis in Mbita and its adjacent islands of Lake Victoria, western Kenya. Parasit Vectors 2012, 5:278. 38. Brooker S: Spatial epidemiology of human schistosomiasis in Africa: risk models, transmission dynamics and control. Trans R Soc Trop Med Hyg 2007, 101:1–8. 39. Hodges MH, Soares Magalhaes RJ, Paye J, Koroma JB, Sonnie M, Clements A, Zhang Y: Combined spatial prediction of schistosomiasis and soil-transmitted helminthiasis in Sierra Leone: a tool for integrated disease control. PLoS Negl Trop Dis 2012, 6:1694. 40. Clements AC, Lwambo NJ, Blair L, Nyandindi U, Kaatano G, Kinung‘hi S, Webster JP, Fenwick A, Brooker S: Bayesian spatial analysis and disease mapping: tools to enhance planning and implementation of a schistosomiasis control programme in Tanzania. Trop Med Int Health 2006, 11:490–503. 41. Glinz D, Silue KD, Knopp S, Lohourignon LK, Yao KP, Steinmann P, Rinaldi L, Cringoli G, N‘Goran EK, Utzinger J: Comparing diagnostic accuracy of Kato-Katz, Koga agar plate, ether-concentration, and FLOTAC for Schistosoma mansoni and soil-transmitted helminths. PLoS Negl Trop Dis 2010, 4:e754. doi:10.1186/1756-3305-6-198 Cite this article as: Mwandawiro et al.: Monitoring and evaluating the impact of national school-based deworming in Kenya: study design and baseline results. Parasites & Vectors 2013 6:198. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit