Trends and determinants of prelacteal feeding practice in rural Bangladesh from 2004 to 2019: A multivariate decomposition analysis

1 1 Trends and determinants of prelacteal feeding practice in rural Bangladesh from 2004 to 2019: A 2 multivariate decomposition analysis. 3 4 Short title: Trends and determinants of prelacteal feeding practice 5 6 Ya Gao 1, Amanda C. Palmer1*, Andrew L. Thorne-Lyman1, Saijuddin Shaikh2, Hasmot Ali2, Hannah Tong1, 7 Monica M. Pasqualino 1, Lee S. Wu1, Kelsey Alland1, Kerry J. Schulze1, Alain B. Labrique1, Rolf D. Klemm1, 8 Parul Christian 1, Keith P. West Jr1 9 10 1Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, 11 USA 12 2 JiVitA Project, Johns Hopkins University, Bangladesh (JHU,B) Keranipara, Rangpur, Bangladesh 13 14 *Corresponding author 15 E-mail: [email protected] . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 2 16 Abstract 17 Prelacteal feeding (PLF)—giving infants food or liquid other than breastmilk within the first 3 days of 18 life—remains common and hinders optimal breastfeeding in Bangladesh. This study assessed changes in 19 PLF practices in rural Bangladesh from 2004 to 2019 and examined associate household, maternal, and 20 infant factors. We analyzed data from two cluster-randomized trials in rural northwest Bangladesh 21 (n=16,551; n=4,401). Trained staff collected sociodemographic and birth data through household visits. 22 We used multivariable logistic regression to examine associations between household, maternal, and 23 infant characteristics and PLF and a non-linear approximation of the Oaxaca-Blinder regression 24 decomposition to understand the factors associated with the changing prevalence of PLF. The 25 prevalence of PLF declined from 89% in 2004 to 24% in 2019. Factors associated with PLF shifted over 26 time, particularly household wealth, infant sex, and birth weight. Institutional delivery (OR=0.27; 95% CI 27 0.22, 0.32 in 2004; OR=0.78; 95% CI 0.61, 1.00 in 2019) and multigravida status (OR=0.68; 95% CI 0.58, 28 0.79 in 2004; OR=0.73; 95% CI 0.58, 0.93 in 2019) were consistently associated with reduced odds of PLF 29 across cohorts in the multivariable analysis. The decomposition analysis based on the two trials 30 indicated that changes in prevalence of the covariates explained 15% of the decrease in prevalence of 31 PLF, primarily accounted for by increases in health facility deliveries (86%), increases in infant birth 32 weight (13%), and increasing gravidity (12%). 85% of the change remains unexplained by the measured 33 variables. The prevalence of PLF declined considerably in rural Bangladesh over the 15-year period. 34 There are shifts in factors associated with PLF overtime. Improvements in socio-demographic factors 35 played a modest but meaningful role in reducing PLF. However the majority of the reduction remains 36 unexplained by the measured variables. Further research is needed to identify other potential drivers for 37 changes in the prevalence of PLF. 38 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 3 39 Introduction 40 Breastfeeding provides benefits to the immediate health and survival of infants and young 41 children, particularly through its protection against infectious diseases (1, 2). Breastfeeding may 42 also have longer-term benefits, including improved cognitive development (3), as well as 43 reduced risks of overweight and obesity (4-7), hypertension, elevated serum cholesterol, and 44 type 2 diabetes (3, 4) later in life. Given the benefits of breastfeeding, the World Health 45 Organization and United Nations Children’s Fund recommend early initiation of breastfeeding 46 within one hour of birth and exclusive breastfeeding, defined as feeding only breast milk 47 without other liquids or solids, for infants less than 6 months (8). There have been 48 improvements in breastfeeding practices over time, but the latest prevalence of exclusive 49 breastfeeding in Bangladesh (62.6% in 2019) is still short of the World Health Assembly goal of 50 at least 70% by 2030 (9). 51 One of the barriers for optimal breastfeeding practices is prelacteal feeds, i.e., any fluid or solid 52 other than breastmilk fed before the establishment of breastfeeding, usually within the first 3 53 days of life. Prelacteal feeding (PLF) is a common practice in South Asia due to ethnic and 54 cultural beliefs (10). The use of honey or sugar-sweetened water has been reported in 55 Bangladesh, as people believe that these sweet prelacteal feeds can clear the voice, prevent 56 infant from catching a common cold, and bless the child with a charming personality in the 57 future (11). Besides cultural beliefs, the perception of insufficient milk production, sickness or 58 unconsciousness after delivery, and stopping infants from crying were also reported reasons for 59 administering PLF from Bangladeshi mothers (11). PLF may delay the initiation of breastfeeding, 60 potentially reducing exposure to colostrum (12, 13), or cause the early cessation of . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 4 61 breastfeeding (14-16). Prelacteal foods may also expose vulnerable newborns to pathogens or 62 chemicals, but the longer-term impacts on health and nutritional status of these practices are 63 not well studied(17). 64 According to the Bangladesh Demographic and Health Survey (DHS), there was a decreasing 65 trend of PLF prevalence at the national level from 62% in 2007 (18) to 24% in 2019 (19). 66 However, the factors that contributed to this trend have not been explored to date. 67 Furthermore, data collection on PLF in the DHS and other large surveys often requires two to 68 five years of recall to get a sufficient sample size. This may introduce recall bias, particularly 69 because PLF often occurs over a period of just a few days (20, 21). This problem can be avoided 70 by collecting data on PLF soon after delivery. Identifying factors associated with changes in the 71 prevalence of PLF overtime may also lead to the development of sustainable interventions and 72 effective recommendations on breastfeeding policies. 73 To the best of our knowledge, no studies have explored the determinants of the trend in 74 prevalence of PLF over time in Bangladesh. We had the opportunity to analyze data collected in 75 northwest Bangladesh at a site that was considered largely regionally representative and has 76 hosted a variety of trials to assess pregnancy outcomes, such that the questions about PLF 77 could be posed to women shortly after birth. The primary objective of this study is to examine 78 trends in prelacteal feeding (PLF) practices in northwest Bangladesh between 2004 and 2019. 79 Specifically, we aim to (1) quantify changes in the prevalence of PLF over time, (2) identify the 80 socio-demographic factors associated with these changes, and (3) use decomposition analysis 81 to determine the contributions of various explanatory factors, such as maternal education, 82 place of delivery, and birth order, to the observed trends. By addressing these objectives, this . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 5 83 study seeks to inform future interventions aimed at reducing harmful feeding practices and 84 promoting optimal breastfeeding 85 Materials and Methods 86 Study Site Description 87 The JiVitA project operates in a rural area, spanning 435 km2 across 19 unions of the northwest 88 districts of Gaibandha and Rangpur with a population of 650,000 (22). The site has hosted 89 several cluster-randomized controlled trials and observational studies related to maternal and 90 child health (23-25). 91 Study Design and Population 92 The first trial, hereafter referred to as JiVitA-1, was a double-masked, cluster-randomized, 93 placebo-controlled trial assessing the efficacy of maternal vitamin A or beta carotene 94 supplementation in reducing pregnancy-related and infant mortality (23). A cohort of the JiVitA- 95 1 trial (January 2004 to December 2006) that contributed to assessing the effect of 96 supplementing newborns with 50,000 IU of vitamin in reducing all-cause infant mortality 97 through 24 weeks of age was included in the analysis (24). The second trial, hereafter referred 98 to as mCARE-II, was a cluster-randomized controlled trial testing a digital health intervention to 99 improve coverage of antenatal and postnatal care. A cohort of the mCARE-II trial (September 100 2018 to July 2019) that contributed to future infant growth assessment was included in the 101 analysis (26). All studies identified and recruited participants through a five-weekly pregnancy 102 surveillance system that has been in place across the study areas since its establishment. All 103 women within the study areas who self-reported their pregnancy based on a documented 104 missed menstrual cycle and a positive pregnancy urine test were consented for enrollment and . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 6 105 then received follow-up visits from JiVitA research workers throughout their pregnancy and the 106 postpartum period. The details of these trials have been published elsewhere (22-24). 107 Data Collection 108 In both studies, trained field workers collected data on household socioeconomic status and 109 maternal demographic characteristics at the time of enrollment. Upon receiving notification of 110 an infant’s birth, field workers visited the home within 72 hours postpartum, collecting detailed 111 data on breastfeeding initiation and any foods or liquids other than own mother’s breast milk 112 provided to the infant, as well as characteristics surrounding the birth environment, gestational 113 age at birth, infant sex, and other demographic characteristics. 114 Information on PLF practices in both studies was collected through structured interviews by 115 field interviewers. In the first trial, field interviewers administered the interview as soon as 116 possible after birth. Mothers were asked “What food, other than the mother’s breast milk has 117 the baby been fed? Note: Only include foods given within the first 3 days after birth.” The 118 recorded responses included nothing offered or one or more of the common feeds 119 (cow/goat/sheep/buffalo milk; water; drops; honeys; other mother’s milk; sugar water/misri 120 water; oil, or other). In the second trial, field interviewers visited the households of consented 121 mothers within 72 hours postpartum to conduct the interview. Mothers were asked about time 122 intervals following birth. Specifically, mothers were asked: “Was the baby fed other mother’s 123 breast milk or anything than own mother’s breastmilk in the [first 30 minutes, second 30 124 minutes, 2nd hour, 3rd hour, 4th hour, 5th hour, 6th hour, 7th-12th hours, 13th-24th hours, 125 remaining hours until upcoming 6 am after completion of 24-hour, entire day 3 and night]?” For 126 each time interval, if the mother responded yes, data on feeding from other mother’s milk or . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 7 127 specific foods from a list of common non-breast milk feeds (honey, water, animal milk, formula, 128 sugar/sugar candy water, any types of drops, powdered/condensed milk, and others) were also 129 collected. “Drops” in Bangladesh consisted of a wide range of items. Previous studies found 130 drops to include homeopathic supplements and broad-spectrum antibiotics, concentrated 131 vitamin supplements, and purportedly sterile solutions of sucrose, 132 glucose, dextrose, or saline (27). 133 Statistical Analysis 134 For this analysis, we included consented women with singleton live births. To reduce the 135 likelihood of recall bias, we also restricted our analysis to interviews conducted within 30 days 136 postpartum. PLF was dichotomized as “yes” or “no”. For the first trial, yes was defined as any 137 responses other than “nothing offered”. For the second trial, yes was defined as one or more 138 yes responses to the question within the first 3 days after birth. 139 The potential determinants of PLF used for analyses were identified after consulting similar 140 literature (27, 28). The selected variables were infant gestational age at birth, infant sex, type of 141 delivery, birth location, infant birth weight, maternal age, maternal literacy, maternal 142 education, maternal gravidity, participation in any micro-credit program, and household 143 socioeconomic status. A Living Standards Index (LSI) was constructed using principal component 144 analysis of data on 20 socioeconomic factors, including household assets and house 145 construction materials (29). LSI was then categorized into five wealth quintiles, with the highest 146 quintile corresponding to the wealthiest group and the lowest quintile representing the 147 poorest. Five categories of birth locations were combined into two categories. One category . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 8 148 included home and enroute/other. The other category, institutional deliveries, included family 149 welfare visitor’s house, health or welfare center, and hospital/clinic/medical college. 150 Bivariate and multivariable logistic regression models were used to examine the associations 151 between different demographic characteristics and PLF status. A value of P < 0.05 was 152 considered statistically significant. None of the covariates included in the multivariate models 153 had variance inflation factors higher than 3 or tolerance values less than 0.1, suggesting that 154 there was no multicollinearity among covariates in the final models. 155 To examine the extent to which the differences in the prevalence of PLF over the 15-year 156 period from 2004 to 2019 were due to changes in maternal and infant demographic 157 characteristics, a non-linear approximation of the Oaxaca-Blinder regression decomposition 158 technique was used (30). We used Stata’s mvdcmp command and included a full set of 159 maternal and infant demographic variables related to prelacteal feeding in the decomposition 160 model. Data management and statistical analyses were conducted in Stata Version 15.1.21. 161 Ethics Statement 162 The protocols for the JiVitA-1 and mCARE-II trial (IRB No. 00006469, approved on August 27, 163 2015) were reviewed and approved by institutional review board at the Bloomberg School of 164 Public Health at Johns Hopkins University and Bangladesh Medical Research Council, Dhaka, 165 Bangladesh. Individual participants provided written informed consent. 166 Results 167 A total of 16,551 infants enrolled in the JiVitA-1 cluster-randomized trial from 2004-2006, and 168 4,401 infants enrolled in the mCARE-II cluster-randomized trial from 2018-2019, were included 169 in the final analysis (Supplement Figure 1). Maternal and infant characteristics of the two study . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 9 170 cohorts are summarized in Table 1. The percentage of women giving birth before the age of 19 171 years decreased from 41.5% to 26.8% over this period while the literacy rate among women 172 increased from 48.6% to 81.1%. The percentage of premature births dropped from 26.8% to 173 17.7%. Health Facility deliveries increased from 6.1% to 40.1%, including an increase in delivery 174 at hospital, clinic, or medical college (2.5% to 28.1%). Cesarean delivery rose markedly from 175 1.9% to 25.1% in the study area. 176 Table 1: Maternal and infant characteristics of the two study cohorts in rural Bangladesh. 2004-2006 (n=16,551) Mean +/- SD or n (%) 2018-2019 (n=4,401) Mean +/- SD or n (%) Maternal characteristics Age, year ≤19 6,862 (41.5) 1,140 (26.9) 20-34 9,168 (55.5) 2,938 (69.2) ≥35 505 (3.1) 166 (3.9) Literate Yes 8,025 (48.6) 3,449 (81.1) No 8,502 (51.4) 806 (18.9) Education No schooling 6,846 (41.4) 493 (11.7) Class 1-9 8,537 (51.7) 3,045 (71.8) SSC passed 443 (2.7) 242 (5.7) 11 year or above 693 (4.2) 458 (10.8) Gravidity Primigravid 6,866 (41.6) 1,253 (28.6) Multigravid 9,656 (58.4) 3,133 (71.4) Participation to micro-credit program Yes 4,517 (27.4) 1,843 (47.0) No 11,999 (72.6) 2,079 (53.0) Wealth quintile 1 3,317 (20.1) 867 (20.4) 2 3,296 (19.9) 837 (19.6) 3 3,304 (20.0) 850 (20.0) 4 3,306 (20.0) 850 (20.0) 5 3,305 (20.0) 851 (20.0) Infant characteristics Preterm Yes 4,420 (26.8) 733 (17.7) No 12,054 (73.2) 3,399 (82.3) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 10 Sex Male 8,410 (50.8) 2,294 (52.1) Female 8,141 (49.2) 2,107 (47.8) Birth location Home 15,207 (93.4) 2,580 (58.6) Family welfare visitor’s house 221 (1.4) 47 (1.1) Health or welfare center 359 (2.2) 478 (10.9) Hospital/clinic/medical college 410 (2.5) 1,240 (28.1) Enroute/other 84 (0.5) 55 (1.3) Type of delivery Vaginal 15,963 (98.1) 3,155 (74.9) Cesarean 314 (1.9) 1,059 (25.1) Birth Weight, kg 2.49 ± 0.47 2.89 ± 0.51 Received prelateal feeding Yes 14,736 (89.1) 1,056 (24.0) No 1,805 (10.9) 3,345 (76.0) 177 178 The prevalence of any form of PLF was 89.1% during the period 2004-2006 (Table 2). Among 179 the infants who received PLF, sugar-sweetened water was the most common prelacteal feed 180 (47.2%), followed by animal milk (45.6%), honey (40.9%), and drops (13.4%) (Figure 1). In 2018- 181 2019, the prevalence of PLF had dropped to 24.0% and the most common prelacteal feed was 182 animal milk (20.6%), followed by sugar-sweetened water (20.5%), honey (20.5%), and any type 183 of drops (14.4%). 184 Table 2. Maternal and infant characteristics of the two study cohorts in rural Bangladesh by prelacteal 185 feeding status 2004 to 2006 cohort 2018 to 2019 cohort Feeding prelacteals (n=14,736) Not feeding prelacteals (n=1,805) Feeding prelacteals (n=3,345) Not feeding prelacteals (n=1,056) Mean ± SD or n (%) Mean ± SD or n (%) Mean ± SD or n (%) Mean ± SD or n (%) Maternal characteristics Age, year ≤19 6,267 (42.6) 590 (32.7) 281 (27.7) 859 (26.6) 20-34 8,003 (54.4) 1,160 (64.3) 687 (67.8) 2,251 (69.7) ≥35 450 (3.1) 55 (3.1) 46 (4.5) 120 (3.7) Literate Yes 7,135 (48.5) 887 (49.2) 825 (81.3) 2,624 (81.0) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 11 No 7,580 (51.5) 915 (50.8) 190 (18.7) 616 (19.0) Education No schooling 6,085 (41.4) 754 (41.9) 124 (12.3) 374 (11.6) Class 1-9 7,675 (52.2) 860 (47.7) 722 (71.4) 2,323 (71.9) SSC passed 378 (2.5) 65 (3.5) 59 (5.8) 83 (5.6) 11 year or above 569 (3.9) 123 (6.8) 106 (10.5) 352 (10.9) Gravidity Primigravid 6,252 (42.5) 608 (33.8) 334 (31.7) 919 (27.6) Multigravid 8,460 (57.5) 1,192 (66.2) 720 (68.3) 2,413 (72.4) Participation to micro- credit program Yes 3,992 (27.2) 525 (29.2) 428 (46.1) 1,415 (47.3) No 10,714 (72.8) 1,275 (70.8) 500 (53.9) 1,579 (52.7) Wealth quintile 1 2,940 (20.0) 375 (20.8) 193 (19.0) 674 (20.8) 2 2,964 (20.1) 329 (18.2) 199 (19.6) 638 (19.7) 3 2,973 (20.2) 331 (18.4) 190 (18.7) 660 (20.4) 4 2,958 (20.1) 347 (19.2) 213 (21.0) 637 (19.6) 5 2,881 (19.6) 421 (23.4) 220 (21.7) 631(19.5) Infant characteristics Preterm Yes 3,944 (26.9) 476 (26.5) 174 (17.5) 559 (17.8) No 10,725 (73.1) 1,319 (73.5) 819 (82.5) 2,580 (82.2) Sex Male 7,508 (51.0) 898 (49.8) 585 (55.4) 1,710 (51.1) Female 7,228 (49.0) 907 (50.2) 471 (44.6) 1,635 (48.9) Birth location Home† 13,793 (95.2) 1,491 (84.0) 660 (62.6) 1,975 (59.0) Health Facility‡ 702 (4.8) 285 (11.5) 395 (55.0) 1,370 (41.0) Type of delivery Vaginal 14,279 (98.5) 1,675 (94.3) 773 (76.3) 2,382 (74.4) Cesarean 211 (1.5) 102 (5.7) 240 (23.7) 819 (25.6) Birth Weight, kg 2.48 ± 0.46 2.56 ± 0.48 2.88 ± 0.52 2.89 ± 0.51 186 †Includes home and enroute/other, number of births happened enroute/other is much smaller than 187 number of births happened at home, See Table 1. 188 ‡Includes family welfare visitor’s houses; or health or welfare center; or Hospital/clinic/medical college. 189 190 Figure 1. Change in frequency of prelacteal feeding by types in rural Bangladesh from 2004 to 2019. 191 Figure 1 legend: The frequencies are nonexclusive, meaning that it was possible for one woman to feed 192 multiple types of foods. Data about formula was not collected in 2004-2006 cohort and data about oil 193 was not collected in 2018-2019 cohort. 194 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 12 195 The bivariate and multivariate analyses of the determinants of PLF among mothers and infants 196 enrolled in the 2004-2006 cohort are shown in Table 3. In bivariate analyses, PLF was 197 associated with younger maternal age, lower maternal education, maternal primigravity, 198 vaginal delivery, home delivery, and lower infant birth weight. In the multivariate regression 199 model, maternal multigravidity (OR=0.68; 95% CI 0.58, 0.79); health facility deliveries (OR=0.27; 200 95% CI 0.22, 0.32); and higher infant birth weight (OR=0.81; 95% CI 0.72, 0.91) reduced the 201 odds of PLF. 202 Table 3. Odds of prelacteal feeding based on characteristics of women and infants of the two study 203 cohorts 2004 to 2006 cohort 2018 to 2019 cohort Feeding prelacteals (crude) Feeding prelacteals (adjusted) Feeding prelacteals (crude) Feeding prelacteals (adjusted) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) Maternal characteristics Age, year ≤19 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 20-34 0.65 (0.59- 0.72)* 0.89 (0.76-1.03) 0.93 (0.80- 1.09) 1.08 (0.86, 1.36) ≥35 0.77 (0.57-1.03) 1.04 (0.75-1.44) 1.17 (0.81- 1.69) 1.40 (0.91, 2.15) Literate Yes 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) No 1.03 (0.93-1.14) 1.06 (0.88, 1.27) 0.98 (0.82- 1.43) 0.92 (0.68, 1.24) Education No schooling 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) Class 1-9 1.11 (1.00-1.23) 1.06 (0.89-1.27) 0.94 (0.75- 1.17) 0.83 (0.59, 1.18) SSC passed 0.72 (0.55- 0.95)* 0.98 (0.69-1.38) 0.97 (0.68- 1.39) 0.79 (0.47, 1.30) 11 year or above 0.57 (0.47- 0.71)* 0.94 (0.70-1.27) 0.91 (0.67- 1.22) 0.72 (0.45, 1.13) Gravidity Primigravid 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) Multigravid 0.69 (0.62- 0.77)* 0.68 (0.58-0.79)* 0.82 (0.71- 0.95)* 0.73 (0.58-0.93)* . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 13 Participation to micro- credit program Yes 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) No 1.11 (0.99-1.23) 1.06 (0.95-1.19) 1.05 (0.90- 1.21) 0.96 (0.81, 1.13) Wealth quintile 1 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 2 1.15 (0.98-1.34) 1.17 (1.00-1.38) 1.09 (0.87- 1.37) 1.11 (0.87-1.41) 3 1.15 (0.98-1.34) 1.17 (0.99-1.37) 1.01 (0.80- 1.26) 1.03 (0.81-1.33) 4 1.09 (0.93-1.27) 1.12 (0.95-1.33) 1.17 (0.93- 1.46) 1.09 (0.84-1.41) 5 0.87 (0.75-1.01) 1.01 (0.84-1.21) 1.22 (0.98- 1.52) 1.36 (1.03-1.81)* Infant characteristics Preterm Yes 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) No 0.98 (0.88-1.10) 1.06 (0.94, 1.19) 1.01 (0.85- 1.23) 1.00 (0.82, 1.23) Sex Male 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) Female 0.95 (0.86-1.05) 0.92 (0.83, 1.02) 0.84 (0.73- 0.97)* 0.83 (0.71-0.97)* Birth location Home† 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) Health Facility‡ 0.27 (0.23, 0.31)* 0.27 (0.22, 0.32)* 0.86 (0.75, 0.99)* 0.78 (0.61, 1.00)* Type of delivery Vaginal 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) 1.0 (ref.) Cesarean 0.24 (0.19- 0.31)* 0.89 (0.65-1.19) 0.90 (0.77- 1.07) 1.05 (0.79, 1.39) Birth Weight, kg 0.69 (0.62- 0.76)* 0.81 (0.72-0.91)* 0.97 (0.84- 1.10) 1.01 (0.86, 1.18) 204 *Factor significantly associated with prelacteal feeding compared to its reference group, P<0.05. 205 †Includes home and enroute/other, number of births happened enroute/other is much smaller than 206 number of births happened at home, See Table 1. 207 ‡Includes family welfare visitor’s houses; or health or welfare center; or Hospital/clinic/medical college. 208 209 The bivariate and multivariate analyses of the determinants of PLF among mothers and infants 210 enrolled in the 2018-2019 cohort are shown in Table 3. Maternal primigravity, male infant sex, 211 and home delivery were associated with higher odds of PLF in bivariate analyses. In the . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 14 212 multivariate regression model, maternal multigravidity (OR=0.73; 95% CI 0.58, 0.93); female 213 infant sex (OR=0.83; 95% CI 0.71, 0.97); and health facility deliveries (OR=0.78; 95% CI 0.61, 214 1.00) reduced the odds of PLF. When compared to the lowest wealth quintile, the children from 215 wealthiest quintile household had higher odds of PLF (OR=1.36; 95% CI 1.03, 1.51). 216 Maternal age, literacy, gravidity, participation in micro-credit program, infant sex, birth weight, 217 birth location, and type of delivery were included in the Oaxaca-Blinder decomposition model 218 to examine the extent to which changes in these factors contributed to the change in PLF 219 prevalence overtime in rural Bangladesh. About 15% of the decrease in PLF prevalence from 220 2004 to 2019 was explained by the changes in the maternal and infant demographic 221 characteristics (Table 4). Among the explained components, the increase in health facility 222 deliveries contributed most to the reduction in PLF (62%), followed by increase in multigravida 223 (12%), increased average infant birth weight (13%), increase in cesarean delivery (5%), 224 increased maternal age at birth (5%), and improved maternal literacy (2%) (Figure 2). The 225 disaggregated results for the Oaxaca-Blinder decomposition of change in PLF between 2004 and 226 2019 were reported in Supplementary Table 1. 227 Table 4. Summary Results for the Oaxaca-Blinder decomposition of change in PLF prevalence between 228 2004 and 2019 in Bangladesh1 PLF at baseline, % 89.1 PLF at endline, % 24.0 PLF change from baseline to endline, % 65.1 PLF change, %, accounted for by explanatory variables (explained)2 9.5 PLF change, %, accounted for by coefficients (unexplained)3 55.6 Share of PLF change explained by the model, % 14.7 229 1 The decomposition is based on regression models in Table 3, which includes maternal age, education, 230 gravidity, participation in micro-credit program, infant sex, birth weight, birth location, and type of 231 delivery. PLF, Prelacteal feeding. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 15 232 2 The explained component refers to changes in PLF prevalence accounted for by changes in the means 233 of the explanatory variables multiplied by their corresponding regression coefficients from Table 3. 234 3 The unexplained component consists of two parts: variations in regression coefficients between 235 baseline and endline; and the interaction between changes in coefficients and changes in explanatory 236 variables. 237 238 Figure 2. Contributions to PLF prevalence reduction by maternal and infant characteristics 239 Figure 2 legend: *P value <0.05. 240 241 Discussion 242 The prevalence of PLF declined over the 15-year observation period from 89% in 2004 to 24% in 243 2019 in our study area. Institutional delivery and multigravida status were consistently 244 associated with reduced odds of PLF across cohorts; however, the importance of institutional 245 delivery decreased in the more recent cohort (2018-2019). Infant weight was no longer a 246 significant determinant, while the highest wealth category became a significant predictor of PLF 247 in the more recent cohort. The sex of the infant played an increasingly important role in the 248 more recent cohort, with female infants being less likely to receive PLF in 2018-2019. The 249 finding suggests shifts in factors associated with PLF over time, particularly related to 250 household wealth, infant sex, and birth weight. Over the 15-year period, the changes in 251 prevalence of the covariates explained 15% of the decrease in prevalence of PLF, primarily 252 accounted for by increases in health facility deliveries, increasing gravidity, an increase in infant 253 birth weight, and increasing maternal age. 254 Sugar water, animal milk, honey, and drops remain the most commonly fed prelacteals in this 255 rural Bangladeshi setting over the 15-year period. Our results are in alignment with a cross- 256 sectional study carried out in the Tangail district in rural Bangladesh, where they found infants . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 16 257 were most commonly fed sugar water, followed by drops, infant formula, honey, and other milk 258 than breastmilk (31). The preference of sugar water and honey conveys the cultural beliefs and 259 ritual practices around giving something sweet to newborns (32). 260 Despite the considerable decline in PLF prevalence over the past 15 years, 1 in 4 women still fed 261 their newborns prelacteals in this rural Bangladeshi setting. The provision of prelacteal feeds, 262 by definition, disrupts exclusive breastfeeding and may have displaced colostrum. As the “first 263 milk” produced between birth through the first 5 days of lactation, colostrum has been found 264 to have a variety of nutritional and immunological benefits to neonates (33). Moreover, in rural 265 settings with poor hygiene, the preparation of prelacteal feeds may potentially introduce 266 harmful substances, such as heavy metals and pathogens in contaminated water. The provision 267 of these feeds to neonates may increase the risk of infectious diseases such as diarrhea and 268 pneumonia, as well as other acute infections and allergies, compromising infant growth and 269 development. However, there are no conclusive findings on the impact of PLF on infant growth 270 (17, 34, 35). 271 In both study cohorts, women giving birth at home were at a higher risk of feeding their 272 newborns prelacteals than women who gave birth at health centers of medical institutions. 273 Similar associations between home delivery and PLF have also been reported in India and 274 Pakistan (36, 37). In addition, the results of decomposition analysis indicated the increase in 275 health facility delivery, from 6.6% to 41.4%, to be the primary driver of the change in PLF, 276 explaining 9.1% of the reduction in PLF prevalence over the past 15 years. Another 277 decomposition analysis study in Ethiopia found that increase in health facility delivery, from 278 6.4% to 35.6%, explained 7.8% of the reduction in PLF (38). Mothers giving birth at home were . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 17 279 more likely to be influenced by their family, friends, and unskilled birth attendants who based 280 their advice on cultural beliefs or personal experience, which may facilitate the practice of PLF. 281 Attending an institutional delivery would have exposed mothers to infant and young child 282 feeding education and immediate postnatal care, such as the encouragement of early initiation 283 of breastfeeding, which may reduce their tendency to feed prelacteals (39, 40). The 284 government of Bangladesh has taken steps to encourage institutional delivery (41). A pilot 285 maternity voucher scheme, providing monetary incentive for attending antenatal care and 286 delivery at public or private facility, or at home with a skilled birth attendant, reached more 287 than 10 million people (42). 288 Multiparous mothers have been found to be less likely to practice PLF in this rural Bangladeshi 289 setting over the 15-year period. Similar findings have been reported by two studies in Nepal, 290 that first time mothers tended to have a higher likelihood of giving prelacteals (43, 44). It is 291 likely that first time mothers who had no previous child feeding experiences were more likely to 292 be guided by advice from family members who encouraged PLF (45). In addition, prior studies 293 have also revealed that multiparous mothers were more likely to initiate breastfeeding early 294 after delivery, and those who had prior breastfeeding experiences would maintain 295 breastfeeding for a longer duration compared with first time mothers (46). In the 296 decomposition model, the increased gravidity over time also explained 1.8% of the reduction in 297 PLF prevalence. It is important to note that the 2004 cohort recruited more newly married 298 women to the surveillance activities to ensure that enough new pregnancies were captured, 299 thereby the 2004 cohort is not representative of the pregnant women at the JiVitA site. The 300 2019 cohort provides a better representation of pregnant women at the JiVitA site. The . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 18 301 contribution to the change might be smaller had the older cohort provided a better snapshot of 302 the pregnant women in the study site. 303 Only 9.5% of the 65.1% reduction in PLF is explained by observable changes in factors such as 304 birth location, gravidity, birth weight, and maternal age. This suggests that improvements in 305 these socio-demographic factors played a modest by meaningful role in reducing PLF. But the 306 majority (55.6%) of the reduction in PLF remains unexplained by the measured variables 307 included in the analysis. The scaling up of infant and young child feeding education at health 308 care facilities, as well as increased access to various breastfeeding promotion campaigns may 309 have contributed to a greater extent, the decline in PLF, as evidenced by decline in prevalence 310 of PLF at national level from 62% in 2007 to 29% in 2018 (18, 47). A large-scale program to 311 improve infant and young child feeding practices was implemented in Bangladesh, from 2010 to 312 2014, covering 50 rural sub-districts, through the existing national Essential Health Care 313 program (48). The at-scale program lowered the use of PLF through interpersonal counseling, 314 mass media, community mobilization, and policy advocacy (48). Since the program covered 315 JiVitA site, women residing in JiVitA site might be indirectly influenced. However limited data on 316 breastfeeding advocation have been collected in this setting to be able to quantify its effects. 317 Our current analysis adds to the literature on the changes, as well as drivers of changes, in PLF 318 in rural Bangladesh over the past 15 years. We had the advantage of having data on PLF 319 collected at similar time points between the two studies that were conducted 15 years apart, 320 which enabled the comparison of data over time. Meanwhile, information on PLF was captured 321 soon after birth, which substantially reduced the likelihood of recall bias. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 19 322 While this study provides important insights into trends in PLF practices, there are several 323 limitations that must be acknowledged. First, the design of the two studies were slightly 324 different. JiVitA-1 had the aim of determining the efficacy of providing an oral supplement 325 containing the weekly equivalent of an RDA of vitamin A from the first trimester of pregnancy 326 through 12weeks (84thday) after pregnancy termination, in reducing all-cause maternal 327 mortality. Newly married women were added to the surveillance activities to ensure that 328 sufficient sample size was included, thereby boosting the number of primiparous women. 329 mCARE-II only identified and recruited participants through the pregnancy surveillance system, 330 providing a snapshot of all pregnant women at the site. The differences in design elicited a 331 different balance of parity, resulting in increased parity in more recent years despite the overall 332 reduction in total fertility rate within Bangladesh (18, 47). Second, the cross-sectional nature of 333 the data limits our ability to establish causality between the observed socio-demographic 334 changes and PLF. Additionally, shifts in cultural practices, healthcare policies, and public health 335 messaging during the 15-year period could have influenced both the prevalence of PLF and 336 associated factors, but these contextual changes were not explicitly captured in the data. 337 Furthermore, there were differences in survey questions between the two periods. We cannot 338 be certain if the same responses would be elicited if ask two different ways. The differences 339 might be a potential source of bias. Some types of prelacteal feeds were not consistently 340 recorded across both survey periods, limiting our ability to conduct a direct comparison of 341 feeding practices over time. However, we had collected data on six different types of prelacteal 342 foods in both studies, which allowed us to characterize changes in common prelacteal practices 343 over time. But we were not powered to look at the trend in types of PLF, for example milk- . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 20 344 based vs water-based, over time. Milk-based PLF were more common in higher-middle-income 345 countries, whereas water-based PLF were more common in low-income countries (9). 346 Conclusions 347 The prevalence of PLF declined considerably by 65% in rural Bangladesh over the 15-year period 348 from 2004 to 2019. Among women practicing PLF, sugar sweetened water, animal milk, honey, 349 and drops remained to be the most frequently fed prelacteals over time. In multivariate 350 models, PLF, in this rural Bangladeshi setting, was significantly associated with lower infant 351 birth weight, home delivery, male infant gender, and maternal primigravidity. Approximately 352 15% of the reduction in PLF between 2004-2006 and 2018-2019 can be attributed to changes in 353 socio-demographic characteristics, most notably birth weight and institutional delivery rates. 354 85% of the change remains unexplained by the measured variables. This suggests that other 355 factors, such as shifts in cultural norms, improvements in public health campaigns promoting 356 exclusive breastfeeding, or changes in healthcare delivery, may have contributed to the decline 357 in PLF. Future research should explore these potential influences to gain a more comprehensive 358 understanding of what drives reductions in PLF. Additionally, it is worth considering whether 359 policy changes, such as increased access to skilled birth attendants or the introduction of new 360 breastfeeding promotion programs, may have indirectly influenced mother’s feeding practices. 361 Acknowledgements 362 We are grateful to the mothers, infants, and their household members for participation in the 363 study. We also thank the many JiVitA field workers and the management team for their support 364 and assistance with recruitment and data collection. 365 Reference . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 21 366 1. León-Cava N, Lutter C, Ross J, Martin L. Quantifying the benefits of breastfeeding: a summary of 367 the evidence. Washington, DC: The Food and Nutrition Program; 2002. 368 2. Effect of breastfeeding on infant and child mortality due to infectious diseases in less developed 369 countries: a pooled analysis. The Lancet. 2000;355(9202):451-5. 370 3. Binns C, Lee M, Low WY. The Long-Term Public Health Benefits of Breastfeeding. Asia Pac J 371 Public Health. 2016;28(1):7-14. 372 4. Horta BL, Victora C. Long-term effects of breastfeeding: a systematic review. Geneva, 373 Switzerland: WHO; 2013. 374 5. Owen CG, Martin RM, Whincup PH, Davey-Smith G, Gillman MW, Cook DG. The effect of 375 breastfeeding on mean body mass index throughout life: a quantitative review of published and 376 unpublished observational evidence. Am J Clin Nutr. 2005;82(6):1298-307. 377 6. Grummer-Strawn LM, Mei Z. Does breastfeeding protect against pediatric overweight? Analysis 378 of longitudinal data from the Centers for Disease Control and Prevention Pediatric Nutrition Surveillance 379 System. Pediatrics. 2004;113(2):81-6. 380 7. Weyermann M, Rothenbacher D, Brenner H. Duration of breastfeeding and risk of overweight in 381 childhood: a prospective birth cohort study from Germany. Int J Obes (Lond). 2006;30(8):1281-7. 382 8. WHO. Infant and young child feeding: Model chapter for textbooks for medical students and 383 allied health professionals. Geneva; 2013. 384 9. Perez-Escamilla R, Tomori C, Hernandez-Cordero S, Baker P, Barros AJD, Begin F, et al. 385 Breastfeeding: crucially important, but increasingly challenged in a market-driven world. Lancet. 386 2023;401(10375):472-85. 387 10. Mihrshahi S, Ichikawa N, Shuaib M, Oddy W, Ampon R, Dibley MJ, et al. Prevalence of exclusive 388 breastfeeding in Bangladesh and its association with diarrhoea and acute respiratory infection: results of 389 the multiple indicator cluster survey 2003. J Health Popul Nutr. 2007;25(2):195-204. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 22 390 11. Khatun H, Comins CA, Shah R, Munirul Islam M, Choudhury N, Ahmed T. Uncovering the barriers 391 to exclusive breastfeeding for mothers living in Dhaka's slums: a mixed method study. Int Breastfeed J. 392 2018;13:44. 393 12. Amele EA, Demissie BW, Desta KW, Woldemariam EB. Prelacteal feeding practice and its 394 associated factors among mothers of children age less than 24 months old in Southern Ethiopia. Ital J 395 Pediatr. 2019;45(1):15. 396 13. Rogers NL, Abdi J, Moore D, Nd'iangui S, Smith LJ, Carlson AJ, et al. Colostrum avoidance, 397 prelacteal feeding and late breast-feeding initiation in rural Northern Ethiopia. Public Health Nutr. 398 2011;14(11):2029-36. 399 14. Raihan MJ, Choudhury N, Haque MA, Farzana FD, Ali M, Ahmed T. Feeding during the first 3 days 400 after birth other than breast milk is associated with early cessation of exclusive breastfeeding. Matern 401 Child Nutr. 2020;16(3):e12971. 402 15. Zarshenas M, Zhao Y, Scott JA, Binns CW. Determinants of Breastfeeding Duration in Shiraz, 403 Southwest Iran. Int J Environ Res Public Health. 2020;17(4). 404 16. Perez-Escamilla R, Hromi-Fiedler A, Rhodes EC, Neves PAR, Vaz J, Vilar-Compte M, et al. Impact 405 of prelacteal feeds and neonatal introduction of breast milk substitutes on breastfeeding outcomes: A 406 systematic review and meta-analysis. Matern Child Nutr. 2022;18 Suppl 3(Suppl 3):e13368. 407 17. Tong H, Thorne-Lyman A, Palmer AC, Shaikh S, Ali H, Gao Y, et al. Prelacteal feeding is not 408 associated with infant size at 3 months in rural Bangladesh: a prospective cohort study. Int Breastfeed J. 409 2024;19(1):15. 410 18. Bangladesh Demographic and Health Survey 2007. Dhaka, Bangladesh and Calverton, Maryland, 411 USA: National Institute of Population Research and Training, Mitra and Associates, and Macro 412 International.National Institute of Population Research and Training, Mitra and Associates, and Macro 413 International.; 2009. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 23 414 19. Multiple Indicator Cluster Survey 2019 Survey Findings Report. Dhaka: Bangladesh Bureau of 415 Statistics, United Nations Children's Fund; 2019. 416 20. Agampodi SB, Fernando S, Dharmaratne SD, Agampodi TC. Duration of exclusive breastfeeding; 417 validity of retrospective assessment at nine months of age. BMC Pediatr. 2011;11:80. 418 21. Burnham L, Buczek M, Braun N, Feldman-Winter L, Chen N, Merewood A. Determining length of 419 breastfeeding exclusivity: validity of maternal report 2 years after birth. J Hum Lact. 2014;30(2):190-4. 420 22. Labrique AB, Christian P, Klemm RD, Rashid M, Shamim AA, Massie A, et al. A cluster- 421 randomized, placebo-controlled, maternal vitamin A or beta-carotene supplementation trial in 422 Bangladesh: design and methods. Trials. 2011;12:102. 423 23. West KP, Jr., Christian P, Labrique AB, Rashid M, Shamim AA, Klemm RD, et al. Effects of vitamin 424 A or beta carotene supplementation on pregnancy-related mortality and infant mortality in rural 425 Bangladesh: a cluster randomized trial. Jama. 2011;305(19):1986-95. 426 24. Klemm RD, Labrique AB, Christian P, Rashid M, Shamim AA, Katz J, et al. Newborn vitamin A 427 supplementation reduced infant mortality in rural Bangladesh. Pediatrics. 2008;122(1):e242-50. 428 25. West KP, Jr, Shamim AA, Mehra S, Labrique AB, Ali H, Shaikh S, et al. Effect of Maternal Multiple 429 Micronutrient vs Iron–Folic Acid Supplementation on Infant Mortality and Adverse Birth Outcomes in 430 Rural Bangladesh: The JiVitA-3 Randomized Trial. JAMA. 2014;312(24):2649-58. 431 26. Palmer AC, Hossain MI, Ali H, Ayesha K, Shaikh S, Islam MT, et al. Protein supplementation 432 delivered alone or in combination with presumptive azithromycin treatment for enteric pathogens did 433 not improve linear growth in Bangladeshi infants: results of a cluster-randomized controlled trial. Am J 434 Clin Nutr. 2025;121(3):597-609. 435 27. Sundaram ME, Labrique AB, Mehra S, Ali H, Shamim AA, Klemm RD, et al. Early neonatal feeding 436 is common and associated with subsequent breastfeeding behavior in rural Bangladesh. J Nutr. 437 2013;143(7):1161-7. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 24 438 28. Sundaram ME, Ali H, Mehra S, Shamim AA, Ullah B, Rashid M, et al. Early newborn ritual foods 439 correlate with delayed breastfeeding initiation in rural Bangladesh. Int Breastfeed J. 2016;11:31. 440 29. Gunnsteinsson S, Labrique AB, West KP, Jr., Christian P, Mehra S, Shamim AA, et al. Constructing 441 indices of rural living standards in Northwestern Bangladesh. J Health Popul Nutr. 2010;28(5):509-19. 442 30. Powers D, Yoshioka H, Yun M. Mvdcmp: Multivariate decomposition for nonlinear response 443 models. The Stata Journal. 2011;11(4):556-76. 444 31. Joshi PC, Angdembe MR, Das SK, Ahmed S, Faruque ASG, Ahmed T. Prevalence of exclusive 445 breastfeeding and associated factors among mothers in rural Bangladesh: a cross-sectional study. Int 446 Breastfeed J. 2014;9:7. 447 32. McKenna KM, Shankar RT. The practice of prelacteal feeding to newborns among Hindu and 448 Muslim families. J Midwifery Womens Health. 2009;54(1):78-81. 449 33. Cacho NT, Lawrence RM. Innate Immunity and Breast Milk. Front Immunol. 2017;8:584. 450 34. Meshram, II, Mallikharjun Rao K, Balakrishna N, Harikumar R, Arlappa N, Sreeramakrishna K, et 451 al. Infant and young child feeding practices, sociodemographic factors and their association with 452 nutritional status of children aged <3 years in India: findings of the National Nutrition Monitoring 453 Bureau survey, 2011-2012. Public Health Nutr. 2019;22(1):104-14. 454 35. Harding KL, Aguayo VM, Webb P. Birthweight and feeding practices are associated with child 455 growth outcomes in South Asia. Matern Child Nutr. 2018;14 Suppl 4:e12650. 456 36. Fikree FF, Ali TS, Durocher JM, Rahbar MH. Newborn care practices in low socioeconomic 457 settlements of Karachi, Pakistan. Social Science & Medicine. 2005;60(5):911-21. 458 37. Roy M, Mohan U, Singh S, Singh V, Srivastava A. Determinants of prelacteal feeding in rural 459 Northern India. International Journal of Preventive Medicine. 2014;5(5):658-63. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 25 460 38. Teshale AB, Worku MG, Tesema GA. Spatial distribution and determinants of the change in pre- 461 lacteal feeding practice over time in Ethiopia: A spatial and multivariate decomposition analysis. PLoS 462 One. 2021;16(1):e0244574. 463 39. Legesse M, Demena M, Mesfin F, Haile D. Prelacteal feeding practices and associated factors 464 among mothers of children aged less than 24 months in Raya Kobo district, North Eastern Ethiopia: a 465 cross-sectional study. Int Breastfeed J. 2014;9(1):189. 466 40. Tariku A, Biks GA, Wassie MM, Gebeyehu A, Getie AA. Factors associated with prelacteal feeding 467 in the rural population of northwest Ethiopia: a community cross-sectional study. Int Breastfeed J. 468 2016;11:14. 469 41. Das S, Alcock G, Azad K, Kuddus A, Manandhar DS, Shrestha BP, et al. Institutional delivery in 470 public and private sectors in South Asia: a comparative analysis of prospective data from four 471 demographic surveillance sites. BMC Pregnancy Childbirth. 2016;16:273. 472 42. Schmidt JO, Ensor T, Hossain A, Khan S. Vouchers as demand side financing instruments for 473 health care: a review of the Bangladesh maternal voucher scheme. Health Policy. 2010;96(2):98-107. 474 43. Karkee R, Lee AH, Khanal V, Binns CW. Initiation of Breastfeeding and Factors Associated with 475 Prelacteal Feeds in Central Nepal. J Hum Lact. 2014;30(3):353-7. 476 44. Bhandari S, Thorne-Lyman AL, Shrestha B, Neupane S, Nonyane BAS, Manohar S, et al. 477 Determinants of infant breastfeeding practices in Nepal: a national study. Int Breastfeed J. 2019;14:14. 478 45. Bililign N, Kumsa H, Mulugeta M, Sisay Y. Factors associated with prelacteal feeding in North 479 Eastern Ethiopia: A community based cross-sectional study. Int Breastfeed J. 2016;11:13. 480 46. Hackman NM, Schaefer EW, Beiler JS, Rose CM, Paul IM. Breastfeeding outcome comparison by 481 parity. Breastfeed Med. 2015;10(3):156-62. 482 47. Bangladesh Demographis and Health Survey 2017-18. Dhaka, Bangladesh, and Rockville, 483 maryland, USA: National Institue of Population Research and Training (NIPORT), and ICF; 2020. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint 26 484 48. Menon P, Nguyen PH, Saha KK, Khaled A, Kennedy A, Tran LM, et al. Impacts on Breastfeeding 485 Practices of At-Scale Strategies That Combine Intensive Interpersonal Counseling, Mass Media, and 486 Community Mobilization: Results of Cluster-Randomized Program Evaluations in Bangladesh and Viet 487 Nam. PLoS Med. 2016;13(10):e1002159. 488 Supporting information 489 S1. Supplementary figure and table . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 11, 2025. ; https://doi.org/10.1101/2025.07.11.25331340doi: medRxiv preprint