Intestinal parasites among rural school children in southern Ethiopia: A cross-sectional multilevel and zero-inflated regression model

Background Over 28 million school-aged children are at risk of intestinal parasite infection in Ethiopia. Few studies have investigated household-level risk factors or applied multilevel analysis to account for the nested data structure. This study aimed to assess the prevalence, intensity, and risk factors of parasite infection among schoolchildren in rural South Ethiopia. Methodology/Principal Findings Using multistage random sampling, we recruited 864 students in the Wonago district. We applied multilevel-logistic and zero-inflated negative binomial regression models (ZINB). Risk factors were concentrated at the individual level; school-level and class-level variables explained less than 5% of the variance. The overall intestinal parasite prevalence was 56% (479/850); Trichuris trichiura prevalence was 75.2% (360/479); and Ascaris lumbricoides prevalence was 33.2% (159/479). The rate of infection increased among children with anemia (AOR: 1.45 [95% CI: 1.04, 2.03]), wasting (AOR: 1.73 [95% CI: (1.04, 2.90]), mothers who had no formal education (AOR: 1.08 [95% CI: 1.25, 3.47]), and those in households using open containers for water storage (AOR: 2.06 [95% CI: 1.07, 3.99]). In the ZINB model, A. lumbricoides infection intensity increased with increasing age (AOR: 1.08 [95% CI: 1.01, 1.16]) and unclean fingernails (AOR: 1.47 [95% CI: 1.07, 2.03]). Handwashing with soap (AOR: 0.68 [95% CI: 0.48, 0.95]), de-worming treatment [AOR: 0.57 (95% CI: 0.33, 0.98)], and using water from protected sources [AOR: 0.46 (95% CI: 0.28, 0.77)] were found to be protective against parasitic infection. Conclusions/Significance After controlling for clustering effects at the school and class levels and accounting for excess zeros in fecal egg counts, we found an association between parasite infections and the following variables: age, wasting, anemia, unclean fingernails, handwashing, de-worming treatment, mother’s education, household water source, and water storage protection. Improving hygiene behavior, providing safe water at school and home, and strengthening de-worming programs is required to improve the health of schoolchildren in rural Gedeo. Author summary Intestinal parasite infections are common among school-aged children in Ethiopia. Several cross-sectional studies have investigated the prevalence and risk factors of these intestinal parasite infections. However, most were conducted in an urban setting in northern Ethiopia; they collected household-level risk factor information from the children, not the parents; and they restricted intestinal parasite infection data to binary outcomes. Therefore, we aimed to assess the prevalence and intensity of intestinal parasite infections and the related individual-, household-, and school-level risk factors among rural schoolchildren in southern Ethiopia. Using a multivariate, multilevel, regression model, we found minimal variation across class- and school-level factors for intestinal parasite infection prevalence. We found associations between intestinal parasite infections and most individual-level factors and some household-level factors. Therefore, interventions focusing on the individual, household, and school should be implemented to reduce the prevalence of infection and parasite load among schoolchildren.


74
Study design and setting 75 We conducted this cross-sectional study from February 2017 to June 2017 in the Wonago district of the Gedeo Zone in the southern  Participants 83 We recruited students aged 5-14 years who gave consent and whose parents or guardians gave consent to participate. Using a three-84 stage cluster sampling method, we assigned schools to level one, classes to level two, and students to level three. We replaced 85 participants who dropped out of school after the selection process with participants of similar class, sex, and age. We selected only one 86 child per household and collected household information from the parent or guardian. Figure 1 shows the recruitment process and  Sample size 91 We calculated the sample size using OpenEpi software (26), assuming a 95% confidence interval ( skin infection [50%]), 5% precision, and design effect 2 due to multistage sampling. We also calculated the sample size using 94 outcome-associated variables, such as undernutrition (32.2% prevalence of stunting among female participants), intestinal parasites 95 (52.6% among children who did not wash their hands before meals), and skin infection (21.8% among children aged 6-10 years). To 96 yield the maximum sample size, we calculated 50% for skin infection (12, 27-29). After adding a 10% non-response rate, we reached 97 a final sample size of 845, the minimum required sample size. We then randomly recruited 864 students.

98
Variables and measurements 99 We used a multilevel logistic regression model to analyze three separate binary outcome variables: the presence or absence of any 100 intestinal parasite infections, the presence or absence of T. trichiura, and the presence or absence of A. lumbricoides. Two separate 101 count models also were constructed for the T. trichiura and A. lumbricoides fecal egg counts. The association between the over-102 dispersed count outcome with excess zeros and the potential predictors of T. trichiura and A. lumbricoides infection was determined 103 using a zero-inflated negative binomial (ZINB) regression model (30).

104
The intensity of parasite infection was measured according to density of eggs in stool samples. loss of appetite in the past month, de-worming treatment in the past 6 months, anemia (< 11.5g/dl for those aged 5-11 and < 12g/dl for 110 those aged 12-14), stunting (height-for-age Z scores below -2 SD), and wasting (body-mass-index-for-age Z scores below -2 SD).
111 Parent factors included the educational level of the mother and father. Household factors included the wealth index, which was 112 constructed using principal component analysis of 15 household assets (electricity, radio, television, mobile phone,

177
To examine potential factors associated with infection intensity, a count model was applied using the fecal egg counts for T. trichiura 178 and A. lumbricoides infections. A Poisson model was appropriate for count data, but the assumption of equal variance and mean did 179 not fit to our data, because the mean of A. lumbricoides and T. trichiura eggs was higher than the variance. Moreover, one-part models

203
Demographic and socio-economic status of schoolchildren and their parents 204 The mean age of the 861 schoolchildren (483 boys and 378 girls) was 11.4 (95% CI: 11.3-11.5) years, ranging from 7 to 14 years.

205
About 88.4% (761/861) of mothers and 48.8% (420/861) of fathers never attended school. S1 undercooked vegetables, wealth, source of drinking water, using treated water at home, or participation in a school feeding program.
238 Table 1 and 2 shows the details 239

Risk factors for T. trichiura and A. lumbricoides infections 248
The intra-cluster correlation value calculated in Model V for T. trichiura was low and insignificant, indicated that the variability in 249 this infection prevalence was not attributable to class or school factors. The S5 Table of   Intestinal parasite infections were found to be a public health problem among schoolchildren aged 7 to 11 years in the Gedeo zone of 294 southern Ethiopia. Controlling for clustering effects at the school and class levels and accounting for excess zeros of fecal egg counts, 295 we found that intestinal parasite infections were associated with increased age, girls, wasting, anemia, loss of appetite in past month, 296 unclean fingernails, lack of nail trimming, lack of hand washing with soap after using the latrine, de-worming treatment, mothers' 297 education levels, water source, and using uncovered water storage container at home. Variations attributable to both class and school-298 level factors for intestinal parasite infection prevalence were less than 5%, indicating minor influence. 299 We used a large and representative sample of schoolchildren and applied a multilevel, mixed-effect model and a ZINB model to parasites. The dependence of clustered data within the school and class levels was measured and indicated using intra-cluster 304 correlation. Unlike previous studies (8, 11-17), we also used a ZINB model to model infection prevalence and intensity. factors, such as sanitation and hygiene, due to homogeneity issues.

313
The prevalence of intestinal parasite infections that we found was higher than previously found in southern Ethiopia (12,24)  Intestinal parasite infection prevalence increased among children whose mothers had no formal education, similar to previous studies 347 in Ethiopia (8,12) and rural Mexico (69). This could be due to lack of knowledge about poor home sanitation and hygiene. Using 348 piped water has been shown to influence the prevalence of A. lumbricoides infection (9). In this study, low rates of A. lumbricoides 349 infection were observed among children living in households using a protected water source, indicating a possibility for contamination 350 when water is not protected from soil-transmitted helminths eggs during transport and storage (70). We also observed a high risk of infection. This finding suggests unsafe or unhygienic food preparation and poor sanitary facilities at schools in the study area. School 357 sanitation and hygiene could affect this finding, though we were unable to show a link due to similarity of this potential exposure 358 variable. Furthermore, some schools had no access to safe water, putting those children at higher risk. Schools with feeding programs 359 thus may be area at high risk of food insecurity and vulnerability to infection.