Description of the study area
This study was conducted in Dessie town, northeastern Ethiopia. The town is located in Amhara Regional State, northeastern Ethiopia. It is 401 Km away from the capital city, Addis Ababa and 475 Km away from the regional city, Bahirdar.
Study design and study period
A community-based cross-sectional study was conducted in July 2017 G.C.
Source population and study population
The source population of this study was pregnant women who were resided for at least 6 months in the town during the study period. The study population was pregnant women who were randomly selected from studied kebeles and included in this study.
Sample size determination
The sample size was determined using a single population proportion formula indicated below. Prevalence of poor dietary practice among pregnant (P = 60.7%) in northwestern Ethiopia was taken to estimate sample size [9]. Margin of error (0.05), critical value at 95% confidence interval (Z1-ᾳ/2 = 1.96) and design effect (DE) = 1.5 was used. 10 percent (10%) non-response rate was added. Finally, 604 pregnant women were included in this study.
$$ n=\frac{{\left({Z}_{1-\frac{a}{2}}\right)}^2\ast p\ast \left(1-P\right)}{d^2} $$
Sampling procedure and technique
A two-stage sampling technique was used. At the first stage, four sub-cities among 10 were selected using the lottery method. The lists of pregnant women (sampling frame) living in these four sub-cities were obtained. In the second stage, participants from each kebeles were selected proportionally by using a simple random sampling technique.
Inclusion and exclusion criteria
All pregnant women who were lived for at least 6 months in the town and healthy (self-reported) were included. Pregnant women with disability (unable to speak) at the time of data collection were excluded.
Variables
Dependent variables
Dietary practice and nutritional status of pregnant women.
Independent variables
Socio-economic and socio-demographic factors, obstetric and pregnancy-related factors, stage of pregnancy, nutrition knowledge, morbidity, dietary diversity, antenatal care (ANC) attendance and health belief model constructs (perceived severity, susceptibility, benefits, barriers, and self-efficacy).
Data collection tools and procedures
Data were collected using a structured interviewer-administered questionnaire by eight trained nutrition professionals through the house to house visiting. Socio-economic and socio-demographic factors, health belief model constructs, nutrition knowledge, dietary practice, and anthropometric data were collected.
Sociodemographic and economic data were collected using a structured questionnaire adapted from the Ethiopian demographic and health survey EDHS (2016) [8]. Items of Health Belief Model constructs were measured using a five-point Likert scale (5 = strongly agree to 1 = strongly disagree). Then, the value of each Likert scale scored by participants for each question was summed and the mean was calculated.
Dietary diversity was collected using 10 food groups recommended by the food and agriculture organization of the United Nations (FAO). Based on FAO cut of points, dietary diversity is poor if less than five food groups consumed 24-h before the date of data collection; and good if a woman ate at least five food groups in the past 24-h before the date of data collection [17].
Nutrition knowledge was collected using 15 nutrition knowledge questions. Nutrition knowledge score was calculated by conducting factor analysis to reduce data and identify nutrition knowledge that explained most of the variance. Then, the nutrition knowledge responses that showed high variation in factor analysis were summed and the average (mean) was calculated.
The dietary practice was assessed using 13 dietary habit questions. The dietary practice score was obtained by summing each response given by participants. Participants were given score 1 if they correctly answer the question, favorable or healthy for dietary practice, and score 0, if they did not correctly answer the question, not favorable or healthy for dietary practice.
Nutritional status was determined by measuring middle upper arm circumference (MUAC). MUAC of the left arm was measured triplicate using a nonstretchable standard MUAC tape to the nearest 0.1 cm with no clothing on the arm. The average of triplicate measurement was taken. Pregnant women having MUAC< 23 cm were considered undernourished and ≥ 23 cm normal [18].
Data quality assurance
A questionnaire was first prepared in English and then translated to the local language called Amharic. The training was given for data collectors and supervisors on methods of obtaining consent, study objectives, contents of the questionnaire, interviewing technique and MUAC measurement procedures. A pretest was conducted on 10% of the total sample size on pregnant women living in the area other than the study site. MUAC measurement was taken in triplicate to ensure accuracy. Overall data collection was monitored daily and the questionnaire was checked for completeness and consistency at the end of the data collection date.
Data processing and analysis
The data were coded and entered into Epi-Info version 7. Then, it was exported to SPSS version 20.0 and checked for a missing value. The normality of data was checked by the Kolmogorov–Smirnov test. Multicollinearity was checked by variance inflation factors (VIF) test. Factor analysis was conducted to identify variables that explained high variability among nutrition knowledge responses. The result was summarized using frequency, mean, standard deviation and percentage. Bivariate logistic regression analysis was conducted to identify variables associated with dietary practice and undernutrition of pregnant women. One independent variable at a time entered to check associated with the dependent variable in bivariate analysis. Variables with a p-value of less than 0.2 in the bivariate logistic regression analysis were entered into a multiple logistic regression model to control confounders. The forward model selection method was employed in multiple logistic regression. At 95% confidence, variable with probability value (p-value) less than 0.05 was considered statistically significantly associated with dietary practice and nutritional status. The strength and direction of association were described using a crude odds ratio (COR) and adjusted odds ratio (AOR).
Operational definitions
Good nutrition knowledge: women had good nutritional knowledge if she scored greater than or equal to mean.
Poor nutrition knowledge: women had poor nutritional knowledge if she scored less than mean.
Poor dietary practice: women had poor dietary practice if she scored less than 75% for dietary practice questions.
Good dietary practice: women had good dietary practice if she scored at least 75% for dietary practice questions [19, 20].
Poor perceived health belief (poorly perceived severity, susceptibility, benefits, and self-efficacy): a woman had poorly perceived health belief if she scored below mean.
Good perceived health belief (good perceived severity, susceptibility, exhibitions, and self-efficacy): a woman had good perceived health belief if she scored at least mean.
Poorly perceived barrier: a woman had poorly perceived barrier if she scored greater or equal to mean.
Good perceived barrier: a woman had good perceived barrier if she scored below the mean.