The fifth in the series, NFHS-5, was conducted by the International Institute for Population Sciences (IIPS) in Mumbai under the guidance of the Ministry of Health and Family Welfare (MoHFW), Government of India. It furnishes comprehensive data on demographics, health, and nutrition for India and its individual states, union territories, and 707 districts as of March 31, 2017[12]. To acquire access to the cross-sectional data from NFHS, the registration was done on the DHS (Demographic and Health Surveys) program website [17], and the data was extracted subsequently.

The NFHS-5 survey gathered data from a wide-ranging sample, including 636,699 households, 724,115 women, and 101,839 men, spanning 707 districts, 28 states, and 8 union territories. A consistent sample design was used to ensure accurate national, state, or union territory and district representation. Each district was further categorized into urban and rural areas. The NFHS-5 survey used a stratified 2-stage sampling approach, using the 2011 census data to select Primary Sampling Units (PSUs), with villages in rural areas and Census Enumeration Blocks (CEBs) in urban areas serving as these units. The detailed study design and sampling strategy are mentioned in the NFHS-5 survey report [12]. The survey used 4 distinct schedules or questionnaires (Household, woman, man, and biomarker) and was conducted in 18 local languages, aided by computer-assisted personal interviewing (CAPI). This study specifically included 106,569 breastfeeding mothers in its analysis [12].

We examined and excluded missing data for all variables of interest, including both outcome and explanatory variables, through a complete case analysis approach [20]. Following the removal of missing values in the dependent variable and the exclusion of cases recorded as “others” in the place of the delivery variable, the total number of participants included in the study amounted to 106,569. The participant selection process is presented in Figure 1.

Given the nonproportional allocation of the sample across various survey domains and their urban and rural areas, the use of sampling weights is essential for any analysis using the survey data. Consequently, all analyses were weighted using NFHS-5 sample weights [21].

Moderation refers to a scenario in which the association between 2 variables is not constant but somewhat influenced by the values of a third variable known as a moderator variable. The initial conceptual framework for moderation analysis was provided by Baron and Kenny [22]. A moderator variable can exert 3 possible effects on the relationship between independent and dependent variables. Particularly, a moderator can (1) weaken the relationship, (2) strengthen the relationship, or (3) reverse the relationship [23]. Thus, this moderator variable, or construct, has the ability to alter the strength or even the direction of the relationship between the 2 entities within a model [24]. In moderation analysis, first, the direct effect of a moderator variable on the outcome or dependent variable was estimated using the unadjusted binary logistic regression model framework, followed by the moderating effect, which is studied by including the interaction term between the independent and moderator variables using the adjusted logistic regression modeling [25,26]. In this study, moderation analysis was used to delve deeper into the moderating effect of the socioeconomic variables on the strength of the relationship between the initiation of breastfeeding and the place of delivery (main association; Figure 2).

Separate binary logistic regression models were fitted to study the effect of each one of the moderating variables. The statistical analysis was performed using IBM SPSS Statistics 26.

The regression models can be expressed through the following equations:

This study used preexisting secondary data with no identifiable information extracted from the NFHS-5 study. The DHS Program’s website hosts NFHS-5, an anonymous data set freely available to the public upon reasonable data request, and does not allow for the identification of survey respondents. The original NFHS-5 study adhered to research ethics, with ethical approval from the IIPS, Mumbai, for NFHS-5 (2019-21). In addition, the informed consent form and International Review Board (IRB) also reviewed the survey and granted ethical approval. Respondents granted their signed consent after receiving complete information about the survey’s purpose and procedures. The interviews were conducted after obtaining explicit consent from each participant [27].

The study included 106,569 breastfeeding mothers and Table 1 provides an overview of their sociodemographic characteristics. Among mothers aged 15-25 years and those aged 26-35 years, the rates of delayed initiation were quite similar at 56.3% (28,963/51,442) and 56.6% (27,121/47,912), respectively. In the 36 years and above age group, the delayed initiation rate was reported to be 58.3% (2805/4812). Notably, a majority of mothers (60.9%, 12,804/21,023) without formal education experienced delayed breastfeeding initiation, while more than half of those with higher education (57.6%, 9512/16,501) practiced delayed initiation. The study documented that 56.5% (58,363/103,215) of married women and 55.2% (525/951) of divorced, separated, widowed, or unmarried women delayed the initiation of breastfeeding.

Mothers with a “poor” wealth index displayed elevated rates of delayed initiation, with 58.1% (29,489/50,788) experiencing delays. Those in the middle and rich wealth index categories recorded 55.7% (11,273/20,243) and 54.7% (18,126/33,135), respectively. Rural-dwelling mothers had a higher rate of delayed initiation at 57.4% (45,085/78,524). Among mothers who delivered in institutional settings, 55.7% (52,308/93,898) reported delayed initiation, while those delivering at home showed a higher prevalence of delays at 64.1% (6581/10,269).

The primary outcome of our study, that is, delays in breastfeeding initiation, is significantly and negatively associated with institutional delivery. The odds of delayed initiation (odds ratio [OR] 0.705, 95% CI 0.676-0.735, P<.001) were significantly lower for the mothers who delivered babies at the institutional facility than for those who delivered their child at home. This confirms the hypothesis that institutional delivery results in significantly lesser odds (29.5%) for delayed breastfeeding among Indian mothers.

Table 2 presents the direct effects of moderating variables on delayed initiation among Indian breastfeeding mothers using binary logistic regression modeling. Mothers aged 15-25 years (OR 0.922, 95% CI 0.868-0.979, P=.008) and those aged 26-35 years (OR 0.933, 95% CI 0.879-0.991, P=.02) documented slightly lower odds of delayed initiation of breastfeeding compared with those aged 36 years and older. The odds of delayed initiation were highest among mothers with no education (OR 1.145, 95% CI 1.098-1.193, P<.001) compared with those with higher education. However, these odds significantly decreased with higher education levels, namely primary (OR 0.944, 95% CI 0.901-0.989, P=.01) and secondary education levels (OR 0.882, 95% CI 0.852-0.914, P<.001). The findings suggested a significant association between the wealth index and delayed breastfeeding initiation. Mothers in the poor category are more likely to practice delayed initiation than those in the rich category, as evident by significantly higher odds (OR 1.146, 95% CI 1.115-1.179, P<.001). Mothers in the middle wealth index category also demonstrated significantly higher odds than those in the rich category for delayed initiation of breastfeeding (OR 1.041, 95% CI 1.005-1.078, P=.02). Mothers residing in rural areas displayed significantly higher odds of delayed initiation (OR 1.156, 95% CI 1.124-1.190, P<.001) than those in urban areas. The classification accuracy for each of the moderating variables is above the threshold of 0.5.

Separate multivariable logistic regression models with interactive terms for moderation analyses were fitted (Table 3). The detailed regression models and the procedure of estimation of moderation effects are represented in (Table S1 in Multimedia Appendix 1). The moderation analysis findings indicated that among mothers in the 15-25 years age group who delivered in institutions, the odds of delayed breastfeeding initiation were 24.3% lower (estimated moderating effect in terms of adjusted OR [aOR] 0.757, 95% CI 0.696-1.307, P=.02) as compared with women of higher age groups and those delivering the child at home. These odds were 29.5% lower (OR 0.705, 95% CI 0.676-0.735, P<.001) when women delivered in an institution in the absence of this moderating variable, thereby indicating that the inclusion of age as a moderating variable weakened the effect of institutional delivery on delayed initiation (main association).

Interestingly, education as a moderating variable has significantly strengthened the negative effect of institutional delivery on delayed initiation. It was observed that after delivering in health institutions, mothers with no education had approximately 38% lesser odds of delayed initiation (estimated moderating effect in terms of aOR 0.616, 95% CI 0.429-1.930, P=.005). The odds of delayed initiation were 29.5% lower (OR 0.705, 95% CI 0.676-0.735, P<.001) when women delivered in the institution in the absence of education as a moderating variable. Similarly, mothers with primary education who delivered in institutions exhibited 49% lower odds (aOR 0.510, 95% CI 0.429-1.772, P=.04) than women with higher education and those delivering the child at home. These odds were lower (49%) as compared when women delivered in an institution, in the absence of moderating effect of education (OR 0.705, 95% CI 0.676-0.735, P<.001), thereby indicating that the mothers with no education and lower level of education (primary), when delivered the child at the health institution have further reduced chances of having delayed initiation.

In addition, mothers from poor socioeconomic status (wealth index) who delivered in the institutions showed approximately 33% lower odds (estimated moderating effect in terms of aOR 0.672, 95% CI 0.528-1.432, P=.004) as compared with those with rich socioeconomic backgrounds. These odds were lower (33%) as compared with when women delivered in an institution in the absence of wealth index as a moderating variable (29.5% lower odds with OR 0.705, 95% CI 0.676-0.735, P<.001), thereby indicating that the mothers from poor wealth index when delivered the child at the health institution have even lesser chances of having delayed initiation. This substantiates that the wealth index as a moderating variable has strengthened the negative effect of institutional delivery on delayed initiation (main association).

All the regression models involving moderation analysis displayed a “good fit.” The classification accuracy of these models in predicting the outcome (Table 3) was adequate and above the requisite threshold of 50% [28,29]. The omnibus test of model coefficients was also statistically significant, indicating that the model (with explanatory variables) has a good fit.

In this research endeavor, we examined the association between institutional childbirth and the delayed initiation of breastfeeding using data sourced from NFHS-5. The moderating influence of diverse socioeconomic parameters on the strength of the relationship between breastfeeding initiation and institutional childbirth was investigated. The primary aim was to identify impediments impacting the timely breastfeeding initiation within institutional childbirth scenarios, with the overarching objective of ameliorating maternal and child health outcomes within the context of India.

The study on 106,569 mothers highlights the prevalence of delayed breastfeeding initiation. Notably, delayed initiation rates were similar between mothers aged 15-25 years and those aged 26-35 years, suggesting similar contributing factors. However, a slight increase was observed in mothers aged 36 years and older, indicating unique challenges for older mothers, such as health issues, complications, previous breastfeeding experiences, societal perceptions, and support systems. Binary logistic regression modeling showed that mothers aged 15-25 years and those aged 26-35 years had slightly lower odds of delayed initiation compared with those aged 36 years and older. A study by Kitano et al [30] among Japanese mothers also found that first-time older mothers faced higher risks of delayed exclusive breastfeeding initiation. Results from a secondary analysis based on the WHO Global survey suggested that maternal age, specifically 35 years or older, is associated with a decreased likelihood of initiating breastfeeding within 1 hour after birth. However, this association becomes nonsignificant after adjusting for potential confounding factors [2].

The majority of mothers without formal education experienced delayed breastfeeding initiation, which might be due to a lack of awareness or access to proper health care information among this demographic. The direct effects of moderating variables indicated that the mothers with no formal education demonstrated significantly higher odds of delayed initiation relative to their counterparts with higher educational attainment. Along similar lines, a study showed that among mothers without formal education, the proportion of early initiation of breastfeeding and exclusive breastfeeding was the smallest [31].

The study’s documentation of delayed breastfeeding initiation indicates that marital status alone may not be a significant determinant of breastfeeding behavior. Furthermore, the mothers of the poor wealth index category exhibited a heightened prevalence of delayed breastfeeding initiation relative to counterparts categorized under middle and high wealth index strata. The regression analysis (direct effects of moderating variables) also underscores a noteworthy association between wealth index and delayed breastfeeding initiation. Mothers categorized as economically disadvantaged (poor) exhibited markedly higher odds of delayed initiation compared with those classified as rich [19]. Furthermore, mothers within the middle wealth index category also displayed significantly elevated odds of delayed initiation compared with their affluent counterparts. Mothers residing in rural areas demonstrated notably higher odds of delayed initiation than those dwelling in urban locales. Similar results were shown in another study in Ethiopia, where mothers residing in urban areas were 29% more inclined to initiate early breastfeeding than those living in rural areas [32].

Mothers who gave birth outside of a health care facility were more likely to have increased odds of delayed initiation [31]. Our research also showed that the mothers opting for home delivery exhibited a greater prevalence of delays. The principal hypothesis (main association) proving the finding of our investigation was that the delay of breastfeeding initiation exhibited a statistically significant negative association with institutional delivery. These results are consistent with the findings of other studies that show that mothers delivering in health care facilities exhibited a reduced likelihood of encountering delayed onset of breastfeeding [1,33]. Along similar lines, a greater incidence of timely breastfeeding initiation was identified among mothers in public health care institutions than those in private facilities or at home [32,34].

The results of the regression-based moderation analyses in our study provided valuable insights into how certain variables interact with the relationship between institutional delivery and delayed breastfeeding initiation. While studying age as a moderating variable, it came to light that for younger mothers, delivering in institutions appears to have a mitigating effect on delayed initiation. The negative effect of institutional delivery on delayed initiation is significantly strengthened when considering education. It was observed that the mothers with no education and a lower level of education (primary), when delivering the child at the health institution, witness further reduced chances of having delayed initiation. This indicates that the impact of institutional delivery on reducing initiation delays is more pronounced for mothers without formal education or lower levels of education, which highlights the importance of breastfeeding and nutritional counseling provided by midwives and obstetricians in health institutions at the time of childbirth [35].

Similar to education, the wealth index as a moderating variable has strengthened the negative effect of institutional delivery on delayed initiation (main association). Mothers from poor wealth index, when delivering the child at a health institution, witness even lesser chances of having delayed breastfeeding initiation. This might be because factors such as poverty-related factors including lack of awareness, knowledge, support, and other psychosocial factors get compensated to some extent when poor women delivering at the health institution gets continuum of antenatal and postnatal care and breastfeeding support services under programs such as MAA [9,35]. Considering the complex moderating effects and interplay of socioeconomic factors, these insights can inform targeted interventions to improve breastfeeding practices among vulnerable populations.

This study’s merits lie in its use of a nationally representative survey with a substantial sample size, affording sufficient power to incorporate diverse potential factors into the analysis. Furthermore, this study ensured greater reliability of its results by applying appropriate statistical adjustments, including using sample weights. This study distinguishes itself through its novel methodology for examining how different socioeconomic factors moderate the strength of association between institutional delivery and breastfeeding initiation among Indian mothers. The study highlights the importance of considering moderating variables such as age, education, and wealth when examining the relationship between institutional delivery and breastfeeding initiation timing. The findings suggest that while institutional delivery results in a reduced probability of delayed breastfeeding initiation for specific demographic groups (such as younger mothers), this effect can be strengthened by education levels and socioeconomic status. The study, thus, offers valuable insights for tailoring interventions to address specific moderating factors associated with delayed initiation and develop effective management strategies by identifying areas needing attention.

This study faces a few limitations. The data set’s cross-sectional design precludes any causal inference. As the study relies on data sourced from nationally representative surveys, certain crucial variables could not be incorporated due to data gaps or missing information. Due to the self-reporting nature, asking mothers about breastfeeding initiation introduces the possibility of recall bias, potentially resulting in either underestimation or overestimation of delayed initiation. Nevertheless, this study minimized recall bias by focusing the analysis on currently breastfeeding mothers and their last-born children.

The research investigated the moderating effects of socioeconomic variables on the association between institutional childbirth and delayed breastfeeding initiation. Given the positive influence of institutional delivery on breastfeeding initiation, it is imperative to enhance the promotion of institutional deliveries through national programs to attain full coverage. The direct unadjusted effects of socioeconomic factors such as age, education level, wealth index, and place of residence play a crucial role in breastfeeding initiation. Mothers of the advanced age group (36 years and older) who lack formal education, belong to low-income households and reside in rural areas are at a higher likelihood of delaying breastfeeding initiation. Institutional delivery significantly lowers the likelihood of delayed breastfeeding initiation, and this negative effect is significantly strengthened when uneducated or less educated women with poor wealth status deliver their children to institutional facilities, underscoring the significance of these moderating factors.

Developing strategies targeting these socioeconomic factors is crucial. Tailored awareness programs designed explicitly for mothers lacking formal education can enhance coverage. Interventions should be crafted to address the needs of rural and uneducated mothers from economically disadvantaged backgrounds. Outreach initiatives aimed at promoting health care service use during pregnancy and delivery, as well as raising awareness about breastfeeding practices, are essential measures to improve the adoption and implementation of early breastfeeding initiation.

In conclusion, our study provides valuable insights into the complex interplay of socioeconomic factors affecting breastfeeding initiation within institutional childbirth settings in India. The moderating factors that weaken and strengthen the association between delayed initiation and institutional, when identified, can guide early interventions aimed at enhancing breastfeeding practices at the health facility level to achieve overall maternal and child health, especially for low- and middle-income countries such as India.