Page 98 - IJPS-2-1

Cross-sectional study of child malnutrition and associated risk factors among children aged under five in West Bengal, India

with three or more categories. Multivariate binary logistic regression models were subsequently
performed in order to identify the significant risk predictors of underweight status, wasting, and
stunting among the studied children. We coded all outcome variables used in this study as binary
responses,whereas independent variables were presented in categorical or dichotomous forms. We
presented the results of the logit models in terms of odds ratios (OR) with 95% confidence interval
(CI). Furthermore, to evaluate the multiple occurrences of child malnutrition, the children were cat-
egorized into three groups: non-malnourished (coded as 0), single-state malnourished (i.e., malnour-
ished in terms of either underweight status, stunting, or wasting, coded as 1), and multi-state mal-
nourished (i.e., malnourished in any two or all three forms, coded as 2). We applied multinomial re-
gression model to estimate the regression coefficients for all three malnutrition forms.
The interpretation of coefficients in multinomial models is not as straightforward as in the binary
logistic regressions. In order to simplify the interpretation of results, we converted multinomial re-
gression coefficients into adjusted percentage by using multiple classification analysis (MCA) con-
version tables. First, we computed the predicted probabilities by using regression coefficients for the
independent variables, mean values, and regression coefficients for other independent variables
and constant coefficient which were then converted to percentages.
3. Results

3.1 Summary of Child Anthropometric Indices
Table 1 shows a summary of nutritional outcomes among the studied children (N=485). As can be
seen from the reported data, majority of the children exhibited stunted growth. More specifical-
ly, 51% of children under the age of five were classified as stunted (<−2S D) and 25%were severely
stunted(<−3SD). Nearly 41%of children were categorized as underweight (<−2SD) and 13% were
severely underweight. Finally, 20% of children were classified as wasting (<−2SD) and almost 8%
were categorized as severely wasting. A small proportion of sampled children (4%) were classified
as overweight.
Kernel density plots indicate that the Z-scores of the three anthropometric indices were distributed
normally for both genders. It is also evident that the Z-score distribution pertaining to weight-
for-height data resembles that of the reference population more closely compared to the weight-
for-age and height-for-age distributions, which shifted downward. This observation was also con-
firmed by the data provided in Table 1, which indicates that the mean Z-score of weight-for-height
was close to zero (M =−0.86) compared to weight-for-age (M = −1.70) and height-for-age (M = −
1.92). Although the curve distributions were fairly symmetrical, their degree of flatness differs sig-
nificantly for different indices. For example, Figure 1A shows that the Z-score distribution for female
weight-for-age was more symmetrical than that corresponding to males. It is also evident that the
mean Z-score value for females shifted slightly to the left relative to that pertaining to males, where-
as the curves were nearly mesokurtic for both genders. On the other hand, the Z-score distribution for
height-for-age reveals that, while the mean score values were nearly identical for both genders;

Table 1. Summary of nutritional indices among children aged 6−59 months in West Bengal, India
Percent
Nutritional Indices Mean Z score SD
< –2SD < –3SD > +2SD
Underweight 41.44 13.4 – –1.70 1.19
Stunted 51.34 25.57 – –1.92 1.96
Wasted 22.47 7.84 4.12 –0.86 1.75
Note: >+2SD refers high weight-for-height or overweight status among the children.
Underweight = Weight-for-Age; Stunting = Height-for-Age; Wasting = Weight-for-Height;
<−2SD = moderately malnourished; <−3SD = severely malnourished

92 International Journal of Population Studies | 2016, Volume 2, Issue 1