Data sources
The study used data from the 2015-16 India Demographic and Health Survey (also called the National Family Health Survey, NFHS-4), conducted by International Institute for Population Sciences, Mumbai through the Ministry of Health and Family Welfare (MoHFW), Government of India. Data collection was technically supported by the Inner City Fund (ICF) International, Maryland, USA. Information on socio-demographic and household characteristics and IYCF practices was collected from a nationally representative sample of eligible women aged 15–49 years. Eligible women were either permanent residents of the surveyed households or visitors who stayed in the households the night before the survey. The response rate in the survey ranged from 94.0% (Andhra Pradesh or West Bengal) to 99.6% in Bihar [24].
In the NFHS-4, a total sample of 249,454,252 households were surveyed based on the 2011 census frame. The households were selected using a two-staged sampling design both for rural and urban areas. Within each rural area, the probability proportional to size was used initially to select villages from a sampling frame and were designated as the Primary Sampling Units (PSUs). The second stage involved the random selection of households from each PSU. In the urban areas, Census Enumeration Blocks (CEBs) were selected in the first stage. The second stage involved the random selection of households from each CEB. Further information on the sampling methodology and data collection has been provided in the final NFHS-4 report [24]. In this study, a total weighted sample of 90,596 maternal responses was used, and the analyses were restricted to the young child aged 0–23 months, living with the respondent, to reduce the potential impact of recall bias, and this approach was consistent with previous studies [4, 5].
Study setting
India is a federation that comprises of 29 states and 7 union territories. The states and union territories are categorised into six administrative zones to facilitate improve financial allocation, economic integration, and inter-state cooperation [25, 26]. The six zonal regions include North, South, East, West, Central and North-Eastern India. The Northern region (n = 11,200) consist of the states and union territories of Jammu and Kashmir, Himachal Pradesh, Haryana, Delhi, Chandigarh, Punjab and Rajasthan. The Southern region (n = 16,469) consist of the states and union territories of Andhra Pradesh, Karnataka, Kerala, Tamil Nadu, Telangana, Andaman and Nicobar Islands, Lakshadweep Islands and the Union Territory of Puducherry. The Eastern region (n = 23,317) consist of states of Bihar, Jharkhand, Odisha and West Bengal. The Western region (n = 11,512) consist of the states and union territories of Gujarat, Maharashtra, Goa, Daman and Diu as well as Dadra and Nagar Haveli. The Central region (n = 24,870) consist of the states of Chhattisgarh, Madhya Pradesh, Uttar Pradesh and Uttarakhand. The North–Eastern region (n = 3,228) consist of the states of Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim and Tripura. In 2019, the Indian government announced that the state of Jammu and Kashmir and Ladakh had been administratively re-organised into union territories [27]. However, we considered Jammu and Kashmir and Ladakh as a part of the state of Jammu and Kashmir due to the nature of the data.
Study outcome
Diarrhoea was the main outcome variable in this study, defined as the passage of three or more loose or liquid stools per day [24]. Mothers were asked whether the child under the age of five years had experienced symptoms of diarrhoea in the 2 weeks prior to the survey. In our study, the measurement of diarrhoea was specific to the child age group for each IYCF indicators, and this method was consistent with previously published studies [28-30] .
Exposure factors
The exposure variables were the IYCF indicators, defined in accordance with the World Health Organisation (WHO) definition for assessing IYCF practices in populations [31]. These IYCF indicators were selected based on past studies from LMICs, which showed that these indicators were associated with diarrhoea [28-30, 32-34].
- Early or timely initiation of breastfeeding was defined as the proportion of children within 0–23 months of age who were breastfed within one hour of birth.
- EBF was defined as the proportion of infants 0–5 months of age who received breast milk as the only source of nourishment but allowed oral rehydration solution, drops or syrups of vitamins and medicines.
- Predominant breastfeeding was defined as the proportion of infants 0–5 months of age who received breast milk as the main source of nourishment but allowed water, water-based drinks, fruit juice, oral rehydration solution, drops or syrups of vitamins and medicines.
- Bottle feeding was defined as the proportion of children 0–23 months of age who were fed with a bottle during the previous day.
- Continued breastfeeding at 1 year was defined as the proportion of children 12–15 months of age who were fed breast milk.
- Continued breastfeeding at 2 years was defined as the proportion of children 20–23 months of age who were fed breast milk.
- Children ever breastfed was defined as the proportion of children born in the last 24 months who were ever breastfed.
- Introduction of solid, semi–solid or soft foods was defined as the proportion of infants 6–8 months of age who received solid, semi–solid or soft foods.
Potential confounding factors
The potential confounding factors included child, maternal, family, media, health service, environmental and community level factors, selected based on evidence from past studies [28-30, 35-37]. Child factors included sex, immunization status, birth order and perceived size of the baby at birth. Maternal factors such as age, education and literacy level, employment status, type of caste or tribe and religion were considered. Respondents’ marital status and household wealth index were considered as household characteristics. In addition, media factors such as exposure to television, newspaper and radio were also considered. Health service factors included frequency of antenatal clinic (ANC) visit, place of delivery, delivery assistance and mode of delivery. Environmental factors included source of drinking water and sanitation. Source of drinking water and the type of toilet facilities were classified as improved and unimproved based on the taxonomy of the WHO/UNICEF Joint Monitoring programme (JMP) for estimating progress on WASH [38]. Improved sources of water were defined as a piped water into dwelling, piped water to yard/plot, public tap or standpipe, tube-well or borehole, protected dug well, protected spring or rainwater; while unimproved water sources consisted of unprotected spring, unprotected dug well, cart with small tank/drum, tanker-truck, surface water or bottled water. Improved sanitation facility included a flush toilet, piped sewer system, septic tank, flush/pour flush to pit latrine, ventilated improved pit latrine (VIP), pit latrine with slab, composting toilet, and a special case (i.e., flush/pour flush of excreta to a place unknown the respondent place). Unimproved sanitation facility was defined as a flush/pour flush to elsewhere (such as street, yard/plot, open sewer or a ditch), pit latrine without slab, bucket, hanging toilet or hanging latrine, shared sanitation, no facilities, bush or field. Community level factors included designated areas of residence as urban or rural.
Statistical analysis
The first step in the analysis involved the tabulation of frequencies (and corresponding percentages) for each of the potential confounding factors (i.e., child, maternal, family, media, health service, environmental and community level factors) by Indian regions. The prevalence of diarrhoea by each of the exposure variables (i.e., early initiation of breastfeeding, EBF, predominant breastfeeding, bottle feeding, continued breastfeeding at one year, continued breastfeeding at two years, children ever breastfed and the introduction of solid, semi-solid or soft foods) was estimated for each of the designated geographical regions in India. This was followed by univariate and multivariate logistic regression using Generalized Linear Latent and Mixed Models (GLLAMM) with a logit link and binomial family to adjust for clustering and sampling weights and investigate the association between the exposures and diarrhoea for each region in India. The univariate binary logistic regression analysis was performed to examine the unadjusted odds ratios.
In the multivariate analysis, a staged modelling technique was employed. In the first stage, the community and family/household factors were entered into the baseline survey multiple logistic regression model to examine their association with the study outcome (diarrhoea). After performing a manual elimination process, variables that were associated with the diarrhoea were retained in the model (Model 1). Second, child characteristics were added into significant model retained in the first stage (Model 2). In the third stage, maternal characteristics were added to the significant variable retained in the second stage (Model 3). As before, those factors with p-values < 0.05 were retained. Similar modelling processes were carried out accordingly for the fourth and fifth stages, environmental and health services characteristics, respectively (Model 4 and 5). In the final stage of the analysis (Model 6), the exposure factors (IYCF indicators) were added to the significant variables obtained from the fifth stage, and variables with a p-value < 0.05 were retained in the final. The odds ratios (ORs) and their 95% confidence intervals (CIs) obtained from the adjusted multiple logistics model was reported as the measure of association between IYCF indicators and diarrhoea. We also estimated and will report the measure of association using the national level data to allow for comparability of the evidence. Data analyses were performed in Stata (version 14.0, Stata Corp, College Station, TX, USA) using the ‘svy’ command to allow for adjustments for the cluster sampling design used in the survey.
Ethics
Measure DHS project ethical approvals were granted by the Ethics Review Board at the International Institute for Population Sciences, Mumbai, India before the surveys were conducted, with written informed consent obtained from participants during the surveys. Participants were given information about the rationale behind the surveys and reassured about the confidentiality of their responses. Measure DHS granted permission for the usage of this information in this study.