Study design and period
We employed a community based cross-sectional study design. The study was representative of adolescent girls aged 10–19 years. The study was conducted between October and December 2015.
Study setting
The WHO country office is supporting the Accelerating Nutrition Improvement (ANI) project. The WHO ANI activities aimed, among others, to improve dietary habits and reducing iron deficiency anaemia among adolescent girls. The project was implemented in ten districts in the regions of Amhara, Oromia, and Southern Nations, Nationalities, and.
Peoples’ Region (SNNPR), in partnership with John Snow, Incorporated and under the overall coordination of the Federal Ministry of Health (FMoH) of Ethiopia. The ten implementation districts were Beyeda, West Belesa, Laygaynt, Wondogent, Gedebasasa, Derra, Debrelibanos, Angacha, Damotegale, and Boloso bombie. The baseline study was conducted in three districts (namely Debrelibanos, Damotegale and Laygaynt) of the ten WHO Ethiopia country office supported districts.
Sample size determination
In this baseline study, sample size determination considered any form of anaemia (13.4% of adolescent girls aged 10–19 years are anaemic; ([5], a 95% confidence level, 4% precision, design effect of 1.5, and contingency for non-response at 5%. This gave a sample size of 440 adolescent girls aged 10–19 years per district. Thus, a total of 1320 adolescent girls age 10–19 years in the 3 selected districts were included.
Sampling procedures
The 3 study districts (Debrelibanos, Damotegale and Laygaynt districts) were selected randomly among the 10 WHO ANI project districts. From each district, we randomly selected 6 rural and 1 urban kebeles (smallest administrative units in Ethiopia). The selection of kebeles for the survey ensured that the sampled households had a balanced geographical spread (lowlands, highlands) within the larger category of rural and urban. The total sample size in each district (n = 440) was allocated for the selected kebeles proportionally, based on the number of household in the kebeles.
In the selected kebele, a central point was identified. Data collectors worked in teams of two. Each team of two started out in different directions (the direction decided in a way that can transect the villages within a kebele).
The first household was randomly selected from among the first 3 households. From then onwards, data collection teams visited every fourth household. If the selected household did not contain an adolescent girl age 10–19 years, the data collection team moved to the next household (the direction is decided a priori) and resumed their sampling once they identified an eligible household. In households in which there was more than one eligible adolescent girl, the team randomly chose one girl to interview.
Data collection and instrument
A questionnaire, adapted from the Ethiopian Demographic and Health Survey (EDHS) and relevant literature was developed. The questionnaire was translated into local languages (Amharic and afan oromo languages). The questionnaire included socio demographic characteristics of the adolescent girls and their respective parents such as education, schooling status, religion, marital status, parental occupation, and family size. Data on household characteristics such as ownership and size of land, type of house and construction materials; availability of fixed assets such as radio, television, phone, bed, and chair and other household items; possession of domestic animals; and access to utilities and infrastructure (sanitation facility and source of water) were collected.
Household food security was measured using the household food insecurity access scale (HFIAS) developed by Food and Nutrition Technical Assistance (FANTA) Project through the Academy for Educational Development [20]. The geographic locations and elevations of visited households were determined using a hand-held global positioning system (GPS) device (Garmin GPSMAP®).
Due to the lack of vital registration systems, we developed a local events calendar to estimate adolescent girls’ year of birth. All enumerators were required to apply the local events calendar to estimate adolescent girls’ age.
The interviews were conducted during weekends i.e. on Saturdays and Sundays in order to capture both in school and out of school adolescent girls in the sample. Adults, preferably mothers or heads of household as appropriate, were interviewed about general household demographic and economic characteristics.
Following the interviews, experienced nurses and lab personnel tested adolescent girls for anaemia using HemoCue B-Haemoglobin analyser. The HemoCue B-Haemoglobin analyser is a portable, rapid and accurate method of measuring haemoglobin. Results are displayed after 45 to 60 s in g/dl on an LCD display. Bio-safety measures such as use of sterile gloves; alcohol/clean water during collection of specimen as well as safe disposal system were employed (used gloves and other materials was collected using safety boxes).
The enumerators and supervisors were trained for three days on general techniques of interviewing and supervision, administration of each item in the questionnaire, hemoglobin measurements, and instruction on ethical treatment of participants. In addition, the questionnaire was pretested in a village, not selected for the study, before the final study began to assess the performance of the study tools. Some revisions were made on the study instruments based on the feedback obtained from the pretest. Interviews were administered by 20 enumerators (experienced nurses and lab personnel) and supervised by three supervisors. The enumerators had a minimum of diploma education (experience in data collection preferable), fluently spoke the local languages, and were residents in the local area or vicinity. The supervisors had a minimum of a bachelor education and previous experience in supervising community based data collection. The supervisors addressed questions and queries of interviewers and corresponded with the investigators whenever necessary. In addition, 2–3 local residents were recruited from each of the districts to guide the data collectors through the villages and ease communication with the villagers. A field guide manual was also developed for use by the interviewers and supervisors.
Data quality assurance
Three days long training was given for data collectors and supervisors. The focus of the training was on understanding the instrument and interviewing skills, appropriate use of HemoCue for anaemia testing, and GPS operation. Role plays and pretests were done before the actual data collection. The supervisors checked all filled questionnaires for completeness and consistency each day before turning them to the investigator.
The HemoCue instrument is widely used to measure hemoglobin in anaemia surveys. Although the instrument is excellent on its own, data quality is dependent on good blood sample collection (capillary blood sample). The personnel were trained on the correct handling of the instrument and procedures. In order to get accurate and reliable hemoglobin values using HemoCue, standardization exercises were conducted multiple times during training until the performance standard was met. The performance standard is met when the difference in hemoglobin levels between data collectors and expert is less than 0.5 g/dl.
We also checked the quality of the hemoglobin data in the sample by calculating its standard deviation (SD). A smaller SD (1.1–1.5) of hemoglobin is usually considered to denote a better data quality than a larger SD.
Bio-safety measures such as use of sterile gloves and alcohol/water during collection of specimen as well as safe disposal system were employed. Materials like gloves and lancets were collected using safety boxes and were transported for safe disposal, either to be buried or incinerated. For adolescent girls who were anemic, counseling to take Iron and Folic Acid (IFA) supplement and referral to a nearby health facility was arranged.
Data management and analysis
We used Epi Data Version 3.1 for data entry and Stata 14.0 (Stata Corp, College Station, TX) for cleaning and further analysis.
Descriptive analysis on the general characteristics of the adolescent girls such as age, schooling, marital status, and knowledge on anaemia as well as household characteristics such as household food insecurity, and household dietary diversity was done. In addition, data presentation using tables, graphs and appropriate summary figures were included.
Household wealth: We applied a principal component analysis (PCA) to construct wealth index. In order to construct a relative household’s wealth index, a suite of several socio economic indicators were collected: land ownership, type of house and building materials, availability of fixed domestic assets (i.e. radio, television, bed, chairs and other household items), ownership of domestic animals, source of drinking and cooking water and availability and type of latrine. A relative socio-economic status was constructed by dividing the resulting score into quintiles that indicate poorest, poor, medium, rich and richest households.
Household food insecurity and diversity
The household food insecurity (access) was derived from the HFIAS tool. The frequencies of affirmative responses to the HFIAS questions were used to classify households into one of the four categories of food insecurity i.e. food secured, mild, moderate and severe food insecurity. We generated household-level mean dietary diversity score using the sum of all foods (food groups) eaten in the respective household during the day and night prior to the date of the survey. We classified households into three levels (lowest, medium and high) of dietary diversity; a household with a lowest dietary diversity score consumed three or less food groups, a household with a medium dietary diversity score consumed four or five food groups, while a household high dietary diversity consumed six or more food groups.
Anaemia prevalence
The hemoglobin (Hb) level was adjusted for high altitude and smoking status before defining anaemia. The adjustment was done to account for a reduction in oxygen saturation of blood. We used the following formula for adjustment of hemoglobin for high altitude.
$$ \mathrm{Hb}\kern0.5em \mathrm{adjustment}\kern0.5em =\kern0.5em \hbox{-} 0.032\ast \left(\mathrm{altitude}\kern0.5em +\kern0.5em 0.003280\right)+0.02\ast +{\left(\mathrm{altitude}\kern0.5em +\kern0.5em 0.003280\right)}^2 $$
where the Hb adjustment is the amount subtracted from each individual’s observed hemoglobin level.
Moreover, hemoglobin adjustments for smoking were done by subtracting 0.3 from individual’s observed hemoglobin level.
Adolescent girls who had an Hb values below 12 g/dL were considered as anemic. Adolescent girls with Hemoglobin values of 11–11.9 g/dL, 8–10.9 g/dL, and < 8 g/dL were categorized as having mild, moderate, and severe anaemia, respectively.
A complex survey data analysis was employed to calculate the district level anaemia prevalence, designating the survey’s primary sampling unit (villages) and strata (urban and rural). The variance was adjusted using Taylor linearized variance estimation method.
Anaemia prevalence was also calculated for each of the districts; among the age groups such as early, middle and late adolescence period; urban and rural residence; and within the following categories: household wealth, food insecurity levels, dietary diversity status, schooling status, agro ecology, heard the term anaemia, and IFA and wealth status, as appropriate.
Analysis of the determinants of Anaemia
We run a multivariate logistic regression model using the ‘svy’ command in STATA 14.0 (StataCorp College Station, TX) to ensure that standard errors are adjusted for the complex survey design. This was done to identify factors that could potentially be associated with the occurrence of anaemia among adolescent girls. We selected theoretically relevant variables from the literature for the regression model including household, personal and diet related variable such as household food security, dietary diversity, socio-economic condition, place of residence, adolescent’s age, schooling, smoking and awareness of the term anaemia and IFA tablets.