Using SPSS version 19, the independent variables were crosstabulated with the dependent variable for stunting, followed by the chi square test of association. The variables found to be significantly associated with stunting (p < 0.01) were then entered into a multiple regression analysis with stunting as the dependent variable and the resulting odds ratios were interpreted with a confidence interval of 99% and significant p value of <0.01.
The results were interpreted based on the UNICEF framework for child under-nutrition and divided determining factors into basic, underlying and immediate.
Descriptive analysis
Basic determinants Around 71% (n = 3877) of the children in this study resided in rural areas. The highest number of them fell in the 0-11 months age group (22%, n = 2694) and the lowest number were in the 48 to 59 months age group (16%, n = 2118). Of the 13,587 children involved in the study, 49% (n = 6691) were girls and 51% (n = 6896) were boys.
Underlying determinants Only 6% (n = 798) lived in homes with toilets that flushed, while more than half had pit latrines, and 40% either had no toilet facility or another type. Almost 68% (n = 7553) brought their drinking water from outside the house. Most children lived in the 3 poorest quintiles, and only around 12% (n = 1718) lived in the richest. The majority of children lived in households whose head and mothers had no education (53.2%, n = 7234 and 57%, n = 7721, respectively).
Immediate determinants Twenty-six percent (n = 3577) of children had suffered from diarrhea in the preceding 2 weeks, 32.4% (n = 4408) had had a cough and 8.4% (n = 1142) had had a fever.
Thirty-three point four percent of the children in this study were stunted (n = 3734).
Crosstabulation and chi square test of association
Of the 11 independent variables included in our study, nine were found to be significantly associated with the dependent variable when conducting the chi square test of association and crosstabulation (see table 2).
Immediate determinants: twenty-nine point nine percent (n = 1189) of stunted children had had diarrhoea in the last two weeks, 34.3% (n = 1364) had a cough in the last two weeks and 9.2% (n = 366) had a fever in the last two weeks. Only having diarrhea in the last two weeks was found to be significantly associated with stunting (p <0.01).
Underlying determinants: Around 26.7% (n = 1062) of stunted children lived in the poorest quintile, 27.7% (n = 1104) in the second poorest, 25.3% (n = 1007) in the middle quintile, 14.4% (n = 575) in the fourth and 5.9% (n = 235) in the richest income quintile. The wealth index was found to be significantly associated with stunting (p < 0.01).
Sixty-five point three percent (n = 2580) of stunted children were born to mothers with no education, 26.4% (n = 1045) had primary education and 8.3% (n = 329) had a secondary or higher education. The household head of stunted children had no education in 60% (n = 2231) of cases, a primary education in 27.1% (n = 1003) and secondary or higher in 12.7% (n = 471) of cases. The mother’s and household head’s education level were significantly associated with stunting (p < 0.01).
Basic determinants: Twenty five point six percentage of stunted children (n = 1021) fell in the 36 – 47 months age group and 25% (n = 995) in the 24 – 35 months age group, followed by 23.3% (n = 928) in the 12 – 23 months age group, 16.3% (n = 649) in the 48 – 59 months age group and 9.8% (n = 390) in the 0 – 11 months age group. Age was found to be significantly associated with stunting (p < 0.01)
Fifty-three point three percent (n = 2124) of stunted children were male and 46.7% (n = 1859) were female. Sex was significantly associated with stunting (p < 0.01).
Seventy-eight point seven percent (n = 3133) lived in rural areas and 21.3% (n = 850) in urban areas. Area was significantly associated with stunting (0.01).
Multiple regression analysis
The nine independent variables found to be significantly associated with stunting were entered into a multiple regression analysis, which showed that the wealth index, mother’s education and household head’s education levels, age, gender and rurality were significantly associated with children being stunted (see table 3).
Immediate determinants
Diarrhoea in the last two weeks The odds of children who had been suffering from diarrhoea in the last 2 weeks being stunted were 1.22 times as likely as those who had not had diarrhoea (95% CI 1.11, 1.34, p = 0.000).
Underlying determinants
Type of toilet facility Children living in homes with no toilet facilities or a type other than pit latrine and flushed were 1.34 times as likely to be stunted than those with flushed toilets (95% CI 1.04, 1.72, p = 0.02). Having a pit latrine was not a significant predictor of stunting (OR 1.08, 95% 0.85, 1.36, p = 0.51).
Main source of drinking water The main source of water was not a significant predictor of stunting: brought from outside compared to piped (OR 0.90, 95% CI 0.78, 1.05, p = 0.18), other sources compared to piped (OR 0.93, 95% CI 0.76, 1.14, p = 0.52).
Wealth index The odds of a child living in a household whose income falls in the middle wealth index quintile being stunted compared to the richest group was 2.89 times as likely (95% CI 2.33, 3.60, p = 0.000), closely followed by the second poorest quintile (OR 2.72, 95% CI 2.14, 3.45, p = 0.000) and the poorest quintile (OR 2.51, 95% CI 1.96, 3.22, p = 0.000). The odds of those in the fourth quintile were 1.99 times as likely as being stunted than the richest (95% CI 1.63, 2.42, p = 0.000).
Mother’s education The odds of a child born to a mother with no education being stunted was 1.43 as likely as that for a child born to a mother with secondary school education (95% CI 1.22, 1.68, p = 0.000). The odds of a child born to mother with primary school education was 1.13 times as likely as secondary school education (95% CI 0.96, 1.32, p = 0.13), but this category was not significant.
Household head’s education Children born in households whose head has no education were 1.27 times as likely to be stunted compared to those whose head had a secondary education or higher (95% CI 1.11, 1.45, p = 0.000). Having a primary education was not a significant predictor of stunting (OR 1.11, 95% CI 0.96, 1.32, p = 0.13).
Basic determinants
Area of residence Children living in rural areas were 1.23 times as likely as those living in urban areas of being stunted (95% CI 1.10, 1.37, p =0.000).
Age The highest odds of being stunted compared to the youngest age group was in both the 24 – 35 months age group (OR 4.09, 95% CI 3.54, 4.73, p = 0.000), and the 36 – 47m group (OR 4.08, 95% CI 3.53, 4.71, p = 0.000). Children aged 12 – 23 months had an odds of 3.67 times as likely as being stunted than the oldest group (95% CI 3.71, 4.24, p = 0.000). The odds then dropped in the 48 – 59m group to 2.92 (95% CI 2.50, 3.41, p = 0.000).
Sex The odds of a male child being stunted were 1.26 times as likely as females (95% CI 1.16, 1.37, p = 0.000).