Study design and setting
A school based cross-sectional study was conducted in Hawa Galan District of Kellem Wollega Zone, Oromia Regional state, south west Ethiopia in February 2016. Hawa Gelan is one of the districts in the Kelem Welega zone of Oromia Regional state. Rob Gebeya is the administrative town of the district. According to the projection from the 2007 census the wereda’s total population for the year 2016 is estimated to be 122, 686, of whom 60,265 were male and 62,421 were female. The district has 30 rural kebeles (the smallest administrative structure) and two urban kebeles. There are six health centers and 30 functional health posts in the district. The district has 21 primary schools and four high schools. Hawa Gelan district is one of the food insecure districts in Oromia region. Up to 30 undernutrition cases are reported weekly from all health facilities of the district. .
Participants of the study
All randomly selected students attending primary schools in the district paired with their parents were participated in to this study. The parents accompany their children to provide socio demographic data and parental involvement in their children schooling. However, children with edema and body swellings were excluded from the study as it can affect their weight. Students attending grade one were also excluded from the study because we used the average annual score of two consecutive semesters to determine academic performance (for grade one students this could not be obtained).
Sample size calculation and sampling technique
Sample size was calculated using single population proportion formula. The result of previous study of 30.7% stunting among primary school in Ethiopia , 95% confidence interval (CI) and 5% margin of error were used to calculate the sample size. Therefore, sample was calculated as: n = [(1.96) 2x 0.307 (1–0.307)]/(0.05) 2 = 327. Design effect of 2 was used and the final sample size was calculated to be 654.
A multi stage sampling technique was employed to select the study participants. There were 21 primary schools in the district. These primary schools were initially stratified in to urban (fifteen) and rural (six). Then seven primary schools (two from urban and five from rural) were proportionally selected using simple random sampling (lottery) method. Then the sample size was proportionally allocated for the seven randomly selected schools taking the grades attended in to consideration (124 from Akako, 103 from Lalistu Sombo, 94 from Boni, 89 from Gecho, 84 from Haro Mechara, 82 from Hawa Fincho and 78 from Hawa Babo). Finally, students in each class were alphabetically listed and participants of the study were proportionally selected from the lists of the students in each class using systematic sampling technique.
Socio demographic data and information about parental involvement in their children’s schooling were collected by face to face parental/guardian interview using structured questionnaire. Anthropometric data for the students were gathered through anthropometric measurement. The questionnaire used for interview was prepared in English language after different relevant literature review and translated into Afaan Oromo, a local language, and then re-retranslated to English for analysis to maintain its coherence. Four clinical nurses who can speak local language were recruited as data collectors. Two BSc nurses were hired from the district to supervise the data collection process. To ensure data quality, data collectors and supervisors were given training before commencement of data collection. During the training, the objective of the study, data collection procedures and techniques including how to conduct anthropometric measurements were discussed in detail. During the training, practical exercise of data collection was done through peer interviewer and anthropometric measurement was conducted on students of the same age at a nearby school that was not selected for the real study.
Anthropometric measurements were performed to obtain height and weight of the students. Weight was measured using seca digital weigh scale to the nearest 0.1 kg while the students were dressed their school uniform and without their shoes. Calibration for the weight scale was done upon every case examination. Measurement of height was obtained in a standing position to the nearest1cm using a wooden height board. All measurements were taken twice and the mean value was used for data analysis. Age of the students was collected from student’s records in case there was no confirmed age record.
Weight-for age Z-scores (WAZ) and height-for-age Z-scores (HAZ) were generated by WHO AnthroPlus Version 3.2.2 from anthropometric measures of weight, height and age; taking sex into consideration. Nutritional status of the children was determined in reference to the age and sex-specific growth charts. HAZ below -2SD of the reference population indicates stunting. WHZ below -2SD of the reference population indicates wasting. WAZ below -2SD of the reference population is underweight .
Measurement of academic achievements
The overall subjects the students were given in the academic year 2015/16 were considered to determine the academic achievements of the students. Annual average score was calculated by taking the result of two consecutive semesters of the year. To examine the effect of nutritional status on educational performance, average marks of the overall subjects the students received were divided into two categories, poor score and good score, based on a cut-off mark of 50%. This cut off point was determined by considering the pass mark set by federal ministry of education, Ethiopian.
Data entry was performed using EpiData version 3.1 and exported to SPSS version 20 statistical package for analysis. We employed three data analysis stages. First of all, descriptive statistics including means, ratios, standard deviations, frequency tables and percentages were used to present the results. Secondly, Pearson’s correlation test was performed to assess the linear relationship between undernutrition and the students’ academic performance. Lastly, logistic regression analysis was conducted to estimate predictors of academic achievement of students. Multivariate logistic regression analysis with crude odds ratio at 95% CI was used to determine presence of association between independent and dependent variables. The degree of association between variables was measured using adjusted odds ratio with 95% confidence interval and association was declared significant at P value ≤ 0.05.
The dependent variable in this study was academic performance of students. The independent variables for this study include participants’ nutritional status (weight for age and height for age); socio-demographic variables (sex, age, ethnicity, religion); family characteristics (marital status of the family, family support for the students, educational status of the family, monthly income of family, family size, occupation of family).
Data quality control
The questionnaire was pretested on 32 (5%) of the sample size of similar population in non-selected school in the district. Data collectors and supervisors were selected based on their experience in the field of data collection and supervision. Field staffs were given training before the commencement of data collection. During training, the objective of the study, procedures of data collection and supervision were discussed in detail. Furthermore, each question included in the questionnaire was discussed deeply and any ambiguity was made clear. Each day, the collected data were checked for its completeness and consistence by supervisors and investigators. Data were also cleaned and rechecked after double data entry was performed.
To conduct this study, ethical approval was obtained from Institutional Review Board (IRB) of school of public health, Jimma University. Official letters of cooperation were written to each selected schools from Jimma University. Parents/guardians of the participating students were informed about the objective of the study, risks and benefit, confidentiality of data and privacy of the information. Parents/Guardians were asked for their willingness to participate in the study and they were told that they have the right to refuse to participate at all or can interrupt the question at any time if they feel discomfort to respond for the question. Written and signed informed consent to participate in this study was obtained from the parents/guardians since our study population was children younger than 16 years. Confidentiality issue was insured by using codes and students’ identifiers were not recorded.