Study design and population
This study utilized secondary data from the Nepal Demographic and Health Survey (NDHS) 2016, a nationally representative cross-sectional survey, to explore the prevalence of double and triple burden of malnutrition and associated factors among mother-child pairs. This survey was carried out as part of the DHS program by New ERA under the guidance of the Ministry of Health, Government of Nepal and supported by ICF international and United States Agency for International Development (USAID). The study population for this study were mother-child pairs from the Nepal Demographic and Health Survey 2016.
Sampling strategy
The NDHS 2016 utilized a stratified, two-stage cluster sampling design to provide representatives estimates for seven provinces, three ecological zones, and urban and rural areas. The survey used enumeration areas (EAs) which is a primary sampling unit (PSU) and was selected from 383 wards in both rural (n = 199) and urban (n = 184) areas with probability proportional to size method. In the second stage, 30 households on average within EAs were selected using a systematic sampling technique. A more detailed methodology of the NDHS has been published in the most recent NDHS report [15]. The details of the sample size and exclusion criteria for the selection of the mother-child pairs are presented in Fig. 1.
Fig. 1 Flow chart for sample size selection
Data collection techniques
In this study, we used anthropometric and biochemical indices such as height-for-age, weight-for-height, and weight-for-age and hemoglobin levels to evaluate the nutritional status of 0–59 months child. The WHO Multicenter Growth Reference Study Group, 2006 was used to calculate the anthropometric indicators to evaluate the nutritional status of the child [25]. Children suffering from stunting, wasting and underweight were defined as children with Z-scores below –2 standard deviation (more than 2 standard deviations below the reference median), for height-for-age (HAZ), weight-for-height (WHZ) and weight-for-age (WAZ) respectively. We categorized the blood hemoglobin level as anemic (<11g/dl) and not anemic (≥11gm/dl) for the purpose of analysis. Similarly, we used body mass index (BMI) classification according to WHO for mothers aged 15–49 years. The standard WHO cut-off value was used to determine the normal BMI (18.5 to <24.99kg/m2) and overweight/obesity (>25.0 kg/m2) [26].
Study variables
The detailed plan for data coding and description of the study variables is given in Table 1.
Outcome variables
In order to simplify the analysis of the outcome variables, we dichotomized all dependent variables into presence (coded 1) versus absence (coded 0). We created four different categories of malnutrition such as overweight/obese mother and stunted child (OM/SC), overweight/obese mother and wasted child (OM/WC), overweight/obese mother and underweight child (OM/UC), overweight/obese mother and anemic child (OM/AC) at the same household level. Four different categories were further combined to formed two categories: overweight/obesity mother and undernourished child (stunting or wasting or underweight) which was considered as the double burden of malnutrition (DBM)[12] and the double burden of malnutrition plus anemic child (DBM + anemia) was regarded as the triple burden of malnutrition (TBM) [4, 5].
Independent variable
In this study, we included maternal socio-demographic factors (mother’s age, age at first birth, ethnicity, place of residence, province, education level, occupation, household wealth status, height, iron/folate intake, antenatal care (ANC) visits, parity, delivery by caesarean section), fathers occupation, education and child factors (age of child, child sex, vitamin A consumption, deworming tablet consumption, breastfeeding status, child weight at birth, and total number of children ever born from single mother) as independent variables.
Table 1 Plan for data coding and description of the study variables.
Data analysis
Data were analyzed using STATA/MP version 14.1 (StataCorp LP, College Station, Texas). The ‘svy’ command was used to adjust for EAs and disproportionate sampling weight and non-response. The datasets for women and child files were merged. The prevalence of overweight/obese mother and stunted child (OM/SC), overweight/obese mother and wasted child (OM/WC), overweight/obese mother underweight child (OM/UC), overweight/obese mother and anemic child (OM/AC), double and triple burden of malnutrition were presented as weighted percentage and 95% confidence intervals. The bivariate and multivariable logistic regression model were performed to assess the factors associated with the double and triple burden of malnutrition. To prevent statistical bias in the multivariable logistic regression model, we examined and reported multicollinearity among the predictor variables using variation inflation factors (VIF). In this study, we used “10” as a cut-off value for the maximum level of VIF [27]. Bivariate analysis was performed to assess the association of socio-demographic factors with outcome variables. All variables with statistically significant associations (p<0.05) in bivariate analysis were included in the multivariable regression model. Results were presented as crude odds ratio (COR) and adjusted odds ratio (AOR) with 95% confidence intervals (CI). P-value <0.05 was considered as statistically significant.
Ethical considerations
This study was a secondary analysis of the NDHS 2016 data thus no separate ethical approval was required. However, ethical clearance for the NDHS was obtained from the ethical review board of Nepal Health Research Council and the written informed consent was obtained from each participant as per the standard ethical guidelines of the DHS program. We registered and requested for access to data from the DHS website (URL: https://www.dhsprogram.com/data/available-datasets.cfm) and received an approval to access and download the DHS data file.