Study design, area, and population
This community-based cross-sectional study was conducted on 284 women from Jan 27 to March 7, 2014 in Tahtay Adiyabo woreda Northwest Tigray specifically in two parts of Tahtay Adiyabo (Lemlem and Shimbelina). Pregnant and lactating mothers were excluded from the study for there would be temporary weight gain during these periods. The three largest ethnic groups reported in Tahtay Adiyabo are the
Tigrayan
, the
Kunama
, and foreign residents from Eritrea. There are 15 health posts and six health centers in the Tahtay Adiyabo woreda. There is one health post in Kunama catchment and one health center.
Sample size calculation
The sample size of the study was calculated using the formula for estimation of single population proportion by the assumption of: P = proportion of women (15–49 age group) that are underweight were 40 % in Tigray [3]. With an assumption of margin of error 0.05 at 95 % confidence level and 10 % non-response rate, using sample size reduction formula, the sample size was 284. Finally, using simple random sampling technique, 284 households were selected. From each household, only one eligible woman at the reproductive age group was selected using simple manual lottery method, interviewed, and measured for her weight and height.
Measurements
The nutritional status of the women had been determined by taking weight in kg and height in cm; weight had been measured using standardized digital Seka weight scale calibrated to 0.1 kg, and height had been measured using Seka measuring rod calibrated to 0.5 cm [7]. All measurements had been taken twice and the average had been computed. When the two measurements differ by one unit, the measurement had been repeated.
Food diversity of women was determined using women dietary diversity score (WDDS). DDS had been assessed based on the number of food groups consumed over the immediate past 24 h [8]. Based on a set list of nine food items (starchy staples, dark-green leafy vegetables, oils and fats, other fruits and vegetables, organ meat, meats and fish, eggs, legumes, nuts and seeds, milk and milk products), a score lower than 4 was classified as low dietary diversity.
Food security had been assessed using a complete form of household food insecurity access scale (HFIAS), an 18-item scale developed by USAID, and the HFIAS indicators categorized the set of responses into two levels of household food security status: food secure and food insecure. Based on the responses given to the nine questions and frequency of occurrence over the past 30 days, households are assigned a score that ranges from 0 to 27 [9].
Data collection and quality assurance
Seven 10th grade-completed students who can fluently speak, read, write, and listen kunamigna and two diploma nurses as supervisor were recruited and trained for 2 days by the principal investigator on the study instrument, weight and height measurement, consent form, how to interview, and data collection procedure. The study subjects had been informed about the aim of the study and confidentiality issue, and informed consent had been secured from the study subject. The data were collected using face-to-face interview and anthropometric assessment. Tools used for the anthropometric assessment were calibrated daily before the assessment begins. After checking all questionnaires for consistency and completeness, the supervisor had presented the filled questionnaire to the principal investigator everyday. Additionally, in order to maintain the quality of the data, the principal investigator rechecked the completed questionnaires, and any problem faced at the time of data collection had been discussed and immediate solution had been given.
Data analysis
We analyzed data from 284 women. Descriptive analysis as proportions, percentages, frequency distribution, and measures of central tendency had been used. Both bivariate and multiple logistic regression analyses were used to examine associations of various factors with nutritional status of women of reproductive age. To control the effect of confounding factors, variables that had been found significant in bivariate analysis had entered into multivariate logistic regression in order to identify predictors of undernutrition among women of reproductive age with adjusted OR correspondence to 95 % CI. Linearity, outliers, interaction, and multicollinearity had been checked for dichotomous data, and the p value <0.05 cutoff point was considered as significant for all the independent variables in the model logistic regression employed. Statistical significance was set at 95 % CI and p value <0.05. Analysis was conducted using SPSS Inc. version 20.
Ethical considerations
This study was approved by the Institutional Review Board of the Mekelle University College of Health Sciences. Then, an official letter was gained from the health bureau. Letter of permission was secured from administrative bodies of the area to communicate with relevant bodies in the community. Participation was voluntary, confidentiality ensured, and informed consent secured before the start of each interview.