Study setting and period
The study was conducted in Abuna Gindeberet district governmental schools. This is located in the West Shoa zone, Oromia regional state 192 km to the West of Addis Ababa capital city of Ethiopia and 134 km to North of Ambo town zonal capital city. There are eleven 1–4 grades, forty 1–8 grades and seven high schools (9–12 grades), and two 1–4 grades of private schools. The total adolescent girls in Abuna Gindeberet district school were 19,150. The study was conducted from January 1–30, 2021.
Study design and population
An institution based cross-sectional quantitative study design was conducted. All adolescent girls (10–19 years) attending schools in Abuna Gindeberet district were source population and all randomly selected adolescent girls from randomly selected schools were study populations. All 10–19 years age girls in the schools of Abuna Gindeberet district, and self-reported pregnant and lactating girls were included and excluded from the study respectively.
Sample size determination and sampling technique
Sample size determination
The sample size was calculated by using a single and double population proportion formula for first and second objective respectively. For first objective by considering proportion of undernutrition among adolescent girls from previous study in the Hawzen district schools (thinness = 32.2%) and (stunting = 33.2%) (42), 95% CI, 4% marginal error. From this prevalence of stunting = 33.2% that yield the highest from the two proportions, sample size calculated as:
$$\mathrm{n}=\frac{{({\mathrm{z\alpha }/}_{2})}^{2}\times \mathrm{pq}}{{\mathrm{d}}^{2}}=\mathrm{n}=\frac{{(1.96)}^{2}\times 0.332\times (1-0.332)}{{(0.04)}^{2}}=533$$
where Z α/2 = critical value for normal distribution at 95% confidence level which equals to 1.96; P = prevalence of stunting from study in Hawzen district school, q = 1-p; d = margin of error.
Then by adding 10% non-response rates sample size for first objective became 587.
For the second objective, by using assumption of 80% power, 95% C.I, then by using Epi info statcalc for the most significant variable from previous study.
Sample size calculation for second objectives
|
S/N
|
Selected variable
|
Prevalence among exposed
|
Prevalence among unexposed
|
Sample size
|
With 10% non-response rate
|
|---|
|
1
|
Grade level (42)
|
Prevalence of thinness among grade 4–8 = 16.7%
AOR = 2.95
|
Prevalence of thinness among grade 9–12 = 15.5%
|
440
|
484
|
|
2
|
DDS (23)
|
Prevalence of thinness among low DDS = 15.6%
AOR = 2.1
|
Prevalence thinness among adequate DDS = 5.7%
|
342
|
377
|
From the calculated sample size, sample size for the first objective gave highest which were 587.
Sampling technique and sampling procedures
First schools were stratified into primary school (4–8 grades) and secondary school (9–12 grades). From forty primary schools, 12 schools (4–8 grades), and seven secondary schools (9–12 grade) three schools were selected by lottery method. The numbers of adolescent girls were obtained from the respective school director's office. From registration girl's age, 10–19 years were screened before actual data collections. The sample size was allocated to each selected school by using probability proportional allocation to total adolescent girls. Then k (constant interval was determined by dividing total number of study population (N) by required sample size(n)). k = N/n = 6101/587 = 10. Then, from 1 to 10 random start were selected by lottery methods. The random start 4 were selected, then every 10th adolescent girls were selected from school registers until fulfilling the required sample size (Fig. 2.)
Operational and term definitions
Undernutrition is regarded as the presence of stunting and/ or thinness.
Thinness is defined as the proportion of adolescent girls with value BAZ < -2 SDs [33].
Stunting is defined as the proportion of adolescent girls with value HAZ < -2SDs [33].
Dietary Diversity Score is the sum of food groups eaten by adolescent girls over the last 24 h [34].
Meal pattern: the measure of whether they consumed their meal regularly or skipped some times.
Primary school: In this study, institutional schools that have students from grades 4–8.
Secondary school: Institutional schools that have students from grades 9–12.
Data collection tool and procedure
Data were collected by face-to-face interview using a pre-tested structured questionnaire adapted from previous studies, guidelines prepared by Food and Agricultural Organization, and anthropometric measurements [21, 30, 34, 35]. The adolescent dietary diversity was measured by a qualitative recall of all foods consumed by each adolescent girl during the previous 24 h, which were validated tools prepared by FAO. It is a dichotomous indicator of whether or not to feed ≥ 4 of 9 food groups in the last 24 h. This was categorized as poor dietary diversity score (< 4 food groups) and good dietary diversity score (≥ 4 food groups). The nine food groups considered were starch staples, dark green leafy vegetables, vitamin A-rich fruits and vegetables, other fruits and vegetables, organ meat, flesh meat, eggs, legumes, and nut and milk and milk products (34). Physical activities levels of adolescent girls were assessed by using the WHO STEP wise approach to non-communicable disease risk factor surveillance (STEPS). Which contained four parts related to work-related activities; travel to and from places, recreational activities, and sedentary behavior [36].
Anthropometric measurements
After training, data collectors were recorded height and weight by using a portable non-stretchable plastic height-measuring board with a sliding head bar following standard procedures and portable digital scales respectively. For height measurement, subjects were asked to stand erect with their shoulders level, hands at their sides, thighs and heels comfortably together, the buttocks, scapulae, heels, and head were positioned in contact with the vertical backboard with a sliding head bar. Then height was measured to the nearest 0.1 cm.
For weight measurement, adolescent girls were asked to remove their shoes, wear light clothes (schools' uniforms) and then, trained data collectors have weighed the subjects on a calibrated portable digital scale and were record the value to the nearest 0.1 kg. Each measurement was standardized and calibrated by carefully handling, placing the weight scale on a flat surface, and confirmed reading at zero to ascertain accuracy every time before measurements.
Data quality control
Data were collected by using pretested and structured questionnaires. The one-day training was given to both data collectors and supervisors. Language experts translated questionnaires to Afan Oromo and then back to English to check the consistency. Pretest was performed before actual data collection on 5% of sample size in neighbor district Gindeberet. Height and weight were taken two times to minimize intra and inter observer’s variability of the data collectors, relative technical error measurement (TEM) was calculated. The accepted relative technical measurement error for intra-observer and inter-observer were less than 1.5% and less than 2% respectively. The proper functioning of digital weight scales was checked every time before weight measurement. The data collectors assured the reading scale was exactly at zero before taking weight. Before data entry into the computer, every questionnaire was checked for completeness.
Data processing and analysis
First, questionnaires were checked for completeness and consistency before data entry. Then, data were coded and entered onto the Epi-info version 7.2.2.6 and exported to SPSS version 25 for analysis. Anthropometric data and other essential variables were exported to WHO Anthro-plus software, a computer program that converts anthropometric data into Z-scores of the indices, BAZ and HAZ, by using WHO 2007 population references. Descriptive statistics such as frequency, proportions, mean and standard deviation were used to describe characteristics of the study population. Normality for continuous variables was checked and data were normally distributed. The presence of multicollinearity between independent variables was checked by using the variance inflation factor (VIF). However, there was no identified variable with multicollinearity problems. Bivariable and multivariable logistic regression analysis was carried out to identify predictors of undernutrition among adolescent girls. Model fitness was checked by Hosmer–Lemeshow goodness-of-fit test. An odds ratio with 95% confidence intervals was used to see the strength of association between each independent variable and outcome variable. Level of statistical significance was declared at p-value < 0.05.
Ethical consideration
After the briefing, the purposes of the study Research Ethical Committee (REC) of the Ambo University College of Medicine and Health Sciences have ethically approved it. Upon approval letter of permission was obtained from the colleges. For participants, less than 18 years of consent was obtained from their parents and assent from the students. For participants age greater than/equal to 18 years informed consent was obtained from students themselves. Confidentiality and privacy of the information were maintained.
