In 2015, the Ethiopia’s population was estimated at 100 million. The country has three administrative levels. The first level includes nine Regional States: Afar, Amhara, Benishangul-Gumuz, Gambela, Harari, Oromia, Southern Nations, Nationalities and Peoples' Region (SNNPR), Somali and Tigray; and two chartered City Administrations: Addis Ababa and Dire Dawa. The second and the third administrative levels are zones and woredas (districts), respectively. Woredas are the implementation units for NTD control programs in Ethiopia.
The distribution of STHs and SCH was mapped over two surveys. In the first survey (November 2013 to March 2014), eight of the nine Regional States and one of the two City Administrations were included. In the second survey (February and April 2015), Amhara and Addis Ababa were mapped. In addition, the second survey included fine-scale mapping survey in Somali Regional State with the aim to cover woredas that were not mapped in the first survey.
For each woreda (or sub-city in the case of City Administrations), ten schools were randomly selected from the list of elementary schools provided by the Federal Ministry of Education (FMoE). From these, five schools were purposively selected by the Woreda Health Office, biased towards schools thought to be at-risk for SCH infections (due to proximity to water bodies, reports on SCH infections, irrigation and fishing practices of the community). An exception to this rule were those woredas where the safety of the field teams during the survey could not be secured. This was of particular impact in Somali Region and was compounded by the relatively small numbers of children enrolled per school. This is evidenced in the number of people surveyed. For the later determination of treatment approach, in these these areas the nearest adjacent woreda was used to decide MDA.
Once the school selection was completed, the field team made the necessary pre-visit arrangements with the school directors. On the day of visit, all students of grade 5 (children around 12 years of age) were arranged in two lines, one for girls and one for boys. A random selection was made of 25 girls and 25 boys, resulting in a total maximum of 50 students per school. In schools with fewer than 25 boys or girls in the appropriate grade, children from lower grades (grade four: children around 11 years of age) or higher grade (grade six: children around 13 years of age) were included. The selected students were asked to provide both a stool and a urine sample.
Stool samples were screened for the presence of STHs and S. mansoni eggs, applying a single Kato-Katz thick smear.The number of eggs of STHs (A. lumbricoides, T. trichiura, and hookworms) and S. mansoni were multiplied by 24 to obtain the faecal egg counts (FECs) expressed in eggs per gram of stool (EPG) for each of the four helminth species.
Urine specimens were screened for S. haematobium in two consecutive steps. First, the presence of haematuria was assessed using Haemastix® (Bayer HealthCare LLC,Elkhart, Indiana, USA). The results of this test were recorded as either negative, trace, +, ++ or +++. Subsequently, urine samples in which at least a trace of haematuria was detected were subjected to urine filtration to assess the number of S. haematobium eggs in 10 ml of urine.
To ensure the quality of the parasitological results the field team were instructed to read slides within 30 minutes to avoid over-clearing of hookworm eggs. For each team, we assigned a team leader to read 10% of the Kato-Katz slides and give feedback on a timely basis to the field team. Except for hookworm eggs, any of the STHs and SCH eggs was used for quality control purpose.
In addition, a questionnaire was conducted to collect information on water, sanitation and hygiene (WASH) at the school level. The results of this questionnaire and parasite infection for the first round of survey were reported elsewhere by Grimes and colleagues.
Training of the field teams
In total, 54 field teams from the different Regional States and City Administrations were involved in this survey. Each team consisted of one health officer (for the questionnaires and treatment) and three laboratory technicians (for the examination of stool and urine samples). The training was provided at the Ethiopian Management Institute (EMI) in Bishoftu, Oromia regional state. During these trainings, both theoretical and practical sessions were given to 164 health officers and laboratory technicians. The theoretical sessions focussed on a variety of aspects of the diseases (life cycles, pathogenesis, laboratory diagnosis, prevention and control strategies), while the practical sessions focussed on the diagnostic methods (the use of Haemastix®, urine filtration and Kato-Katz thick smear), archiving Kato-Katz slides for quality control, completing questionnaires on WASH and using LINKS® (a smart phone-based application to collect data).
Mapping coordination and supervision
Supplementary Information (SI) 1 provides an overview of the coordination and supervision of this study. The Ethiopian Public Health Institute (EPHI), the technical arm of FMoH, was responsible for the overall coordination of the survey. For this coordination, four central supervisors were assigned. At the Regional States and City Administrations, at least one supervisor from either the Regional Health or Education Office for each Regional State (and City Administration) was assigned throughout the mapping period. In addition, external supervisors from the Ugandan Vector Control Division (Uganda) and the Kenyan Medical Research Institute (Kenya) conducted monitoring visits. These supervisors independently evaluated the survey coordination and communication flow between the central level (EPHI), the regional supervisors (Regional States and City Administration) and the field teams. They also monitored the operational procedures at the schools, providing feedback to the field teams, correcting those that were not adhering to the mapping protocol.
Data collection and data management
Data was collected using the LINKS® data collection system developed by the Task Force for Global Health (Atlanta, USA). This is an android-based application that allows standardized entry of epidemiological data across the different teams. At the school level, the teams collected the GPS coordinates, the total number of students, the total number of boys and girls in the school, the availability of toilets, water supplies and hand washing facilities. At the individual level, the teams collected the age, sex, the number of Ascaris, Trichuris, hookworm, S. mansoni eggs, the presence of haematuria, and the number of S. haematobium eggs. Raw data were downloaded from the LINKS system server and was subsequently curated using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA).
Statistical data analysis
The prevalence and intensity of infections were calculated for any STH and SCH, and the individual helminth species, separately. Prevalence was estimated by the proportion of children for whom eggs of a particular helminth species were detected. Intensity of infection was measured as the prevalence of MHI infections were calculated for any STH and SCH, and for the different helminth species separately. Unweighted prevalence was calculated at both regional and woreda level. To this end, the proportion of children excreting eggs over the total number screened in the region or woreda was calculated.Subsequently, the point estimates of the different parameters were plotted on a geographical map of Ethiopia using ArcGIS version 10.4 (ESRI, Inc).
We explored the variation in infections of any intensity and MHI by generalized linear mixed models, which were fitted for each of the five helminth species with the presence/absence of infections of any intensity or MHI infections as outcome, the age (in years) as covariate, and sex (2 levels: boy and girl) and Region/City Administration (11 levels: Afar, Amhara, Benishangul-Gumuz, Gambela, Harari, Oromiya, Southern Nations, Nationalities and Peoples' Region (SNNPR), Somali, Tigray (Regions), Addis Ababa and Dire Dawa (City Administrations)) as factors. In the analysis, we accounted for clustering of children from the same school and schools from the same woreda. These infection parameters (prevalence, infection intensity and mixed infections) were calculated at the different administrative levels (national, Regional States/City Administrations, woredas and schools).Subsequently, the point estimates of the different parameters were plotted on a geographical map of Ethiopia using ArcGIS version 10.4 (ESRI, Inc).We explored the variation in infections of any intensity and MHI by generalized linear mixed models, which were fitted for each of the five helminth species with the presence/absence of infections of any intensity or MHI infections as outcome, the age (in years), sex (2 levels: boy and girl) and Region/City Administration (11 levels: Afar, Amhara, Benishangul-Gumuz, Gambela, Harari, Oromiya, Southern Nations, Nationalities and Peoples' Region (SNNPR), Somali, Tigray (Regions), Addis Ababa and Dire Dawa (City Administrations)) as explanatory variables. In the analysis, we accounted for clustering of children from the same school. To facilitate an easy interpretation, we centralized the age around its median (= 12 years) and converted into an binary outcome (boy = 0, girl = 1). The level of significance was set at p <0.05.
In addition, we determined the prevalence of mixed infections (the proportion of children who were excreting eggs of at least two different helminths). Finally, we determined the MDA design for each of the woredas included in this survey. To this end, we classified the wordas into low, moderate and high endemic for STHs and SCHs applying WHO classification criteria  on the prevalence of any STH and SCH infection across the five schools of the same woreda (the proportion of cases over all children screened in that woreda).
Quality control result analysis for the Kato-Katz thick smear
Quality control of the FECs of any STHs and S. mansoni was performed for 6,042 individuals. For this, Kato-Katz thick smears were re-examined by a team leader. The proportions of both false positives and false negatives were assessed. To this end, we assumed that the team leader was correct. In cases where both results indicated presence of eggs, the agreement in egg counts was assessed. Result disagreements were defined when the difference in egg counts was greater than 10 when the team leader counted fewer than 100, or when the difference in egg counts was more than 20% when the team leader counted more than 100 eggs.
The results from the field surveys were used to develop graphical maps of the distribution of SCH and STH infection, and of SCH and STH moderate and high intensity infections. This was done in ArcGIS version 10.4 (ESRI, Inc). The maps are intended as a useful geographical guide of the distribution of infection. There were challenges with representing the woreda-level results from the field surveys onto a graphical map meaning a one-to-one representation was not possible. This was related particularly to the rapid increase in the number of woredas in the country (resulting in splits between ‘mother’ and ‘daughter’ districts and subsequent changes to administrative boundaries) and the lag in availability of updated mapping software to represent these accurately. This means that the graphical maps cannot be a perfect one-to-one representation of the data collected. This is an inherent challenge in countries with changing administrative borders and sub-divisions. The numbers of woredas in the tables for each infection category are most accurate and used to inform the control program.
Informing MDA strategy
In order to determine the MDA strategy in each district, the upper 95% confidence interval of woreda (district) level prevalence was taken of either ‘any STH’ or ‘any SCH’. This was then compared to the WHO-recommended cut-offs [Ref]. Using the upper confidence interval rather than the point prevalence represents a conservative approach to ensure the program reaches as many people in need of treatment as possible.