Impact of prelacteal feeds and neonatal introduction of breast milk substitutes on breastfeeding outcomes: A systematic review and meta‐analysis

Abstract The introduction of fluids other than breast milk during the first few days of life or later neonatal period has been identified as a risk factor for suboptimal breastfeeding (BF) outcomes in numerous studies using varying study designs. However, the relationship between early introduction of fluids other than breast milk and BF outcomes has not been systematically assessed using only prospective studies that can establish temporality, which is critical for determining whether observed associations are causal. We conducted a systematic review and meta‐analysis of prospective studies to assess if there is a difference in BF outcomes as a result of the introduction of: (a) milk‐based prelacteals, (b) water‐based prelacteals and (c) breast milk substitutes (BMS) between 4 days and 4 weeks postpartum. We searched PubMed, Lilacs, Web of Science and other repositories for original research investigating the relationship between early introduction of prelacteals and/or BMS and BF outcomes. Forty‐eight studies met the inclusion criteria for the systematic review. Of the 39 prelacteal feeding studies, 27 had the prerequisite statistical information for inclusion in the meta‐analysis. Findings from the meta‐analysis showed a relationship between prelacteals and exclusive BF cessation (RR 1.44; 1.29–1.60) and any BF cessation (2.23; 1.63–3.06) among infants under 6 months old. Nine studies focusing on the introduction of BMS during the neonatal period identified this practice as a statistically significant risk factor for a shorter BF duration. Effective interventions are needed to prevent the introduction of unnecessary milk‐based prelacteals and BMS during the perinatal and neonatal periods to improve BF outcomes.


| INTRODUCTION
Timely initiation of breastfeeding (BF) and preventing the unnecessary introduction of prelacteal feeds, defined as fluids other than breast milk offered during the first 3 days after birth and breast milk substitutes (BMS) during the neonatal period is key for the subsequent success of BF and the reduction of neonatal mortality (Boccolini et al., 2013). For example, prelacteal feeding and early introduction of BMS have been consistently associated with suboptimal BF practices, such as delayed initiation of BF, shorter exclusive breastfeeding (EBF) and any BF durations, mainly through ecological, cross-sectional or retrospective studies (Boccolini et al., 2015;Neves et al., 2022;Pérez-Escamilla et al., 1996;Segura-Pérez et al., 2022) and infant morbidity and mortality (Nguyen et al., 2020). Yet, about one in three children in low-and middleincome countries (LMICs) receive unnecessary milk-and/or waterbased prelacteal feeds .
A recent analysis of 76 LMICs  found that milkbased prelacteals were more common than water-based prelacteals in higher-income countries while the opposite was true in lower-income countries. By contrast, water-based prelacteals were relatively more common in lower-income countries. More specifically, milk-based prelacteals were most commonly used in Eastern Europe and Central Asia, East Asia and Pacific and in Latin America and Caribbean regions.
Yet, in the 3 African regions, water-based prelacteal foods were the most prevalent although milk-based prelacteals were also used. In addition to the introduction of prelacteals during the first 3 days of life, BMS are commonly introduced during the neonatal period across the globe .
Given that ecological, cross-sectional and retrospective studies can lead to spurious associations because of recall bias, it is key to find out if there is a relationship between prelacteals and BF outcomes focusing only on prospective studies. This study needs to address the use of milk-and water-based prelacteals together and separately as there are contrasting regional prelacteal feeding patterns across the globe . Therefore, the objectives of this study, as listed in the original protocol, were to conduct a systematic review and meta-analysis of prospective studies to answer the following questions: (a) Is there a difference in BF outcomes when milk-based prelacteal feeds are introduced compared to when they are not? (b) Is there a difference in BF outcomes when water-based prelacteal feeds are introduced compared to when they are not? and (c) Is there a difference in BF outcomes when BMS are introduced between 4 days and 2 weeks and >2 weeks and 4 weeks postpartum compared to when they are not? We hypothesize that prelacteal feeds may undermine BF success by delaying BF initiation, reducing nursing frequency, delaying the onset of lactation, increasing the risk of very early introduction of BMS, reducing milk supply and increasing the frequency of BMS feeding beyond the neonatal period further undermining breast milk production .
Given that previous studies have already identified modifiable risk factors for prelacteal feeding and introduction of BMS during the neonatal period (Akuse & Obinya, 2002;Boccolini et al., 2015;Kavle et al., 2017;Neves et al., 2022;Segura-Pérez et al., 2022), we anticipated that the findings from this review could help advance infant feeding and maternity care policies that are more supportive of BF, especially during the crucial period when the milk supply starts to get established (Boss et al., 2018).

| METHODS
This systematic review and meta-analysis followed the Institute of Medicine guidelines. Before reviewing the literature, the protocol was developed and registered in PROSPERO (ID# CRD4202 1240669). BF outcomes of interest were grouped into short-, medium-and longer-term outcomes. The outcomes of this review were EBF duration or prevalence among infants less than 6 months old, duration/prevalence of EBF reported beyond 1 month postpartum and the duration/prevalence of any BF until 1 year postpartum.
The short term outcome delayed onset of lactation, that is, milk arrival >3 days after birth, was listed in the original protocol, however, it was decided not to include it because it was an outcome in another systematic review published in this supplement . For this review, definitions from the World Health Organization were used for the timely initiation of BF, EBF and any BF (UNICEF, 2016). Timely initiation of BF was defined as BF initiation within 1 h after birth. Exclusive breastfeeding was defined as the infant receiving only breast milk with no other liquids or solids introduced (exceptions were that the infant can receive medicines, vitamins/ minerals in liquid form and oral rehydration solution). Any breastfeeding was defined as the infant receiving breast milk either directly from the breast or expressed breast milk.
The two exposures explored in this systematic review were prelacteals and the introduction of BMS during the neonatal period.
Prelacteal feeds were defined as any fluids other than breast milk Key messages • The practice of prelacteal feeding and introduction of breastmilk substitutes (BMS) during the neonatal period is widespread across the globe.
• In this systematic review and meta-analysis of prospective studies, the introduction of prelacteals was associated with suboptimal breastfeeding practices including a higher risk of not exclusive breastfeeding and any breastfeeding cessation among infants under 6 months of age.
• Prospective studies show that the introduction of BMS during the neonatal period is a risk factor for suboptimal breastfeeding practices.
• Facility and community-based interventions are needed to prevent the introduction of prelacteals perinatally and BMS during the neonatal period.
given during the first 3 days postpartum .
Prelacteal feeds were characterized as being water-or milk-based. It was not possible to classify milk-based prelacteals further as infant formula versus other types of animal milk because this information was not available. The introduction of BMS during the neonatal period was defined as the feeding of any milk-based BMS between 4 days and 4 weeks postpartum.

| Search strategy
We searched the databases MEDLINE All (via Ovid), Web of Science Core Collection (as licensed at Yale University, including SCI-EXPANDED 1900-, SSCI 1900-, A&HCI 1975-, CPCI-S 1991-, CPCI-SSH 1991-, BKCI-S 2005-, BKCI-SSH 2005-, ESCI 2015-and CCR-EXPANDED 1985, PsycINFO (via Ovid), EMBASE (via Ovid), LILACS (via the Virtual Health Library Regional Portal), SciELO and Global Index Medicus databases. The searches were run between March and May 2021. We searched for articles containing one or more controlled vocabulary terms or keywords related to both of the following concepts: infant feeding outcomes and prelacteal feeds. The latter concept was operationalized with text words and adjacency statements addressing in-hospital supplementation and early introduction of BMS, as well as prelacteals per se (Table S1). The search was designed, tested and run by a team that included public health scientists with expertise in BF research (R. P. E., A. H. F., E. C. R.) and a medical librarian (K. N.). A reproducible search strategy from all the database searches is available at https://osf.io/jkx6s/. The search results were combined and de-duplicated in EndNote. To ensure no relevant studies were left out, we conducted backward citation chaining through the identification of studies that were cited in previously published systematic reviews, consulted with experts in the field and reviewed researchers' files to identify additional articles to be included. This use of both bibliographic databases and citation networks helped ensure the comprehensive retrieval of relevant papers.

| Study selection criteria
The records identified from the database searches were screened using Covidence. Studies were included in this review if they: (a) were published in English, Spanish, or Portuguese; (b) used a prospective cohort, quasi-experimental, or randomized controlled trial (RCT) design; (c) reported BF duration and other BF outcomes.
Studies were excluded if they: (a) were cross-sectional studies, case-control studies, clinical studies, qualitative studies, reviews, systematic reviews, or meta-analyses; (b) included women who had a delivery involving serious maternal complications such as severe post-partum haemorrhaging that prevented them from BF; (c) included women with contraindications to BF such as cancer chemotherapy and taking lithium medications; (d) included newborns without serious medical complications such as asphyxia at birth; (e) focused on premature or very low birth weight babies; (f) did not report prelacteal feeds or early introduction of BMS; (g) did not report any longer-term BF outcomes (i.e., duration/ prevalence of any BF) and (h) were not published in the peerreviewed literature-for example, technical reports, dissertations, conference abstracts.
Titles and abstracts of studies were independently screened for inclusion by two reviewers (A. H. F., E. C. R.) and those that did not meet the inclusion criteria were excluded. For standardization and consistency, the reviewers independently reviewed the first 100 titles and abstracts and then met to review the level of agreement and discuss any differences that arose in the operationalization of the inclusion criteria.
As the reviewers had a strong agreement in the inclusion/exclusion classification of the first 100 titles and abstracts, the rest of the titles and abstracts were divided between the two reviewers so each of the remaining records could be excluded based on only one screener's opinion. Both reviewers screened all the full texts of studies identified as potentially meeting the inclusion criteria based on the title and abstract.
Full texts deemed to be included in the review had to be identified as fulfilling all inclusion criteria by both reviewers. The screening results were compared and any discrepancies were discussed until a consensus was reached. If consensus could not be reached, a third reviewer (R. P. E.) resolved those conflicts.
Titles and abstracts identified through citation chaining and experts were searched for in Covidence and any documents that had already been screened in Covidence were removed. The remaining titles, abstracts and full texts of studies were screened by the same reviewers (A. H. F., E. C. R., R. P. E.) using the same method as described above.

| Statistical analyses
The results of the search and article selection are presented using the the prelacteals meta-analyses. However, 12 prelacteal feeding studies were not included in the meta-analysis due to a lack of needed information on statistical parameters. Meta-analyses were conducted on prelacteals regardless of type (milk-and waterbased combined) and then milk-based prelacteals only. Waterbased prelacteals by themselves could not be analysed as only one study was available (Qiu et al., 2007). It was not possible to conduct meta-analyses for neonatal BMS introduction (between 4 days and 1 month postpartum) because in most studies the precise time of the introduction of BMS could not be ascertained.
As the length of follow-up and BF practices assessed varied greatly across studies, outcomes were grouped into five categories: (1) any BF cessation among infants under 6 months of age; (2) EBF cessation among infants under 6 months of age; (3) any BF among infants under 6 months of age; (4) EBF among infants under 6 months of age; and (5) any BF cessation among infants up to 1 year of age.
A meta-analysis was then conducted to test the effect of prelacteal feeds on each of these outcomes (Tables 1, S2 and S3, Supporting Information Appendix A). As such, a total of five separate meta-analyses focused on prelacteal feeds and BF outcomes. If a study did not present BF outcomes that fit within these BF outcome categories, the study was not included in a meta-analysis and was only reviewed in the narrative as part of the systematic review (Table S3).
The included studies were carried out in different world regions, and evaluated different BF outcomes among populations of children of different ages. Hence, we pooled the estimates using both random effects and fixed effects models using Stata 17.0. The findings were similar for random and fixed effects models, hence only the results from random effects models are presented here.
The I 2 was used to investigate heterogeneity and the funnel plot and Egger test were used to assess the occurrence of publication bias.
Some studies reported on different effect measures (either hazard ratio or odds ratio). Thus, sensitivity analyses were performed by limiting the models to studies that reported on the same impact measure (Supporting Information Appendix B).

| Study quality assessment
The Joanna Briggs Institute (JBI) critical appraisal tools for observational, quasi-experimental and experimental studies was used to assess the quality of each study included in the review

| RESULTS
Of the 48 studies, 18 were conducted in North America, followed by 7 in East Asia and Pacific, 6 in Europe and Central Asia, 5 in the Middle East and North Africa, 5 in Latin America and the Caribbean, 3 in South Asia and 2 in Sub-Saharan Africa. Thirtytwo studies were conducted in high-income countries, 7 were conducted in upper-middle-income countries and 7 in LMICs.
Finally, 2 studies were conducted in multiple lower-and uppermiddle-income countries that spanned 3 regions, including South Asia, Latin America and the Caribbean, Sub-Saharan Africa.

Author
Year

| Prelacteals and exclusive BF cessation
There was a strong relationship between prelacteal feeds and EBF cessation among infants under 6 months old when the prelacteals were analysed together regardless of type relative risk [RR] 1.44; 1.29-1.60) ( Figure 4a) and the relationship remained when only milkbased prelacteals were included (RR 1.40; 1.24-1.58) (Figure 4b).
F I G U R E 2 Assessment of study quality with Johan Briggs Institute protocol. Prelacteal feeds studies.
F I G U R E 3 Assessment of study quality with Johan Briggs Institute protocol. Neonatal breast milk substitutes studies.
T A B L E 2 Included studies that prospectively examined the association between prelacteals and BF outcomes. • Mothers who did not give prelacteals at birth practiced EBF more than mothers who did give prelacteals (log-rank test χ 2 = 5.6; p = 0.02). • Higher risk for discontinuing EBF was seen with mothers who fed prelacteals (HR = 2.12; 95% CI:

Weisband et al. (2017) United States
The study recruited mothers with GDM and those without, but only mothers without GDM (n = 2139) were included in this review. Mothers were assessed prenatally and 10 times over a 1year period.

Zakarija-Grkovic et al. (2016) Croatia
773 mothers were included in the study and assessed at birth and 3, 6, 12 and 24 months (nurses also recorded hospital feeding data during hospitalization).

| Prelacteals and BF cessation
Prelacteal feeds were also a risk factor for any BF cessation by 6 months post-partum when the prelacteals were analysed together regardless of type ( Figure 5a) and the relationship remained when only milk-based prelacteals were included (RR 2.23; 1.63-3.06) ( Figure 5b). Prelacteal feeds were also a risk factor for any BF cessation by 1 year post-partum (HR 2.02; 1.29-3.17) (Figure 5c).

| Sensitivity analysis
We were able to perform this analysis for any and EBF cessation under 6 months only, as for the other outcomes analysed the same impact measure was reported in the studies (either hazard ratio or odds ratio) (Supporting Information Appendix B). When limiting the analysis to articles that reported on the same effect measure, our findings remained: any BF cessation under 6 months, HR: 1.90, 95% CI: 1.23-2.95; EBF cessation under 6 months, HR: 1.41 95% CI: 1.24-1.61.

| Prelacteal feeds and BF outcomes of studies not included in the meta-analysis
As indicated above, 12 prelacteal feeding studies were not included in the meta-analysis due to a lack of needed information on statistical parameters (Table S3). Ten of those studies found statistically significant inverse associations between prelacteal feeds and suboptimal BF outcomes. Of these, 4 were inversely associated with prevalence or duration of EBF (Demirci & Bogen, 2017;Hayek et al., 2019;Hossain et al., 1992;Vehling et al., 2018), 1 with full BF (Lakati et al., 2010) and 4 with any BF (Gray-Donald et al., 1985;Samuels et al., 1985;Weisband et al., 2017). In 1 US study, the frequency of in-hospital formal supplementation was inversely associated with BF duration (Feinstein et al., 1986). In another US study, prelacteal feeds were identified as the mediators between minority ethnicity/race (Black compared with White mothers) and short BF duration (McKinney et al., 2016). In the remaining 2 studies, an inverse relationship between prelacteals and BF outcomes was not found (Rasheed et al., 2009;Sheehan et al., 1999).

| Risk of publication bias
The analysis of funnel plots suggests a small or no publication bias for most outcomes analysed, except for any BF cessation under 6 months and EBF under 6 months (Supporting Information Appendix C) that were indicative of a significant presence of bias. However, the Egger test was not statistically significant for any outcome investigated (any

| Quality assessment
The quality assessments showed that most studies (prelacteals and BMS) did not clearly describe whether the exposure or outcomes T A B L E 3 Included studies that prospectively examined the association between breast milk substitutes and BF outcomes.

Author (Year) Country Population/methods
Type of exposure BF outcomes Results* Breast milk substitutes Barría et al. (1990) Chile 365 infants were included in the life-table analysis to determine the influence of supplementation on lactation.

Milk-based
Exclusive lactation Median natural exclusive lactation was 10 months among infants not supplemented versus 3 months among infants supplemented with formula by 1 month postpartum. Bunik et al. (2010) United States 341 mothers agreed to participate in the RCT but analyses were conducted only with the intervention group (N = 107).
Mothers were assessed at discharge, daily for the first 2 weeks, then 1, 3 and 6 months postpartum.

Milk-based
Predominant BF Infants introduced to formula within the first month of delivery were more likely to be predominately BF rather than EBF (AdjOR = 1.54; 95% CI: 1.14-2.09). Grossman et al. (1990) United States 97 mothers were enroled in the intervention or control group and were assessed in the hospital and at 6 weeks. Mothers still nursing at 6 weeks were contacted monthly until weaning occurred.
Any form BF cessation (<6 weeks) • A higher percentage of infants who were weaned at <6 weeks had received supplements ≤2 weeks after birth compared to those who were BF at 6 weeks (63% vs. 38%).
• There was no difference in the percentage of infants weaned at <6 weeks and those who were BF at 6 weeks that had received supplements between 2-4 weeks after birth (34% vs. 33%).
Milk-based BF (4,8,12,16,20 weeks) Across all weeks, a significantly higher proportion of mothers in both samples who EBF were still BF compared with those who supplemented with formula. Marques et al. (2001) Brazil 289 mother-infant pairs complete the full follow-up where mothers were assessed in the hospital within 24 h of delivery after milk was established and 2× a week for 12 months.
Milk-based (with or without cereal)

BF duration
Mothers who introduced other milk in the first month had a shorter median duration than those who did not (65 vs. 165 days, respectively, p < 0.001).
(Continues)  (Qiu et al., 2007). Second, we could not conduct a meta-analysis for neonatal BMS introduction as it was not possible to ascertain the precise time when these were introduced during the neonatal period, as specified in the systematic review protocol.
Another potential limitation of the meta-analyses is that 12 of the 39 articles focusing on prelacteal feeding could not be included because they did not report key statistical information. This may have been in part due to the fact that many of these articles did not examine the relationship between prelacteals and BMS as their primary aim. Milk-based BF (2, 4 months) • Mothers who were fully BF at 1 week postpartum were more likely to BF at 2 months than those partially BF (AdjOR = 4.6, 95% CI: 1.3-15.8).
• Mothers fully BF at 1 week postpartum were more likely to BF at 4 months than those partially BF (AdjOR: 4.1, CI: 1.7-10.0). Qiu et al. (2010) China 1520 mothers were enroled and assessed before hospital discharge and at 1, 3 and 6 months postpartum.
• Infants given water before 1 month of age were at higher risk of stopping any BF than those who were given water after 1 month (AdjHR = 1.713; 95% CI: 1.290-2.274) Abbreviations: BF, breastfeeding; CI, confidence interval; EBF, exclusive breastfeeding; HR, hazard ratio; OR, odds ratio; RCT, randomised controlled trial.
*Results are significant at the p < 0.05 level unless otherwise indicated. There were also notable strengths. Our findings are based only on results from prospective studies, which are much less likely to be affected by recall bias compared with cross-sectional and retrospective studies. The included articles represent different world regions, over one-quarter of them were identified through citation chaining and we reviewed studies in three languages. Furthermore, we conducted sensitivity analyses and found no evidence of publication bias. The findings from the studies not included in the prelacteals meta-analysis were consistent with the directionality of associations found in the meta-analysis; that is, they also identified prelacteals as a risk factor for suboptimal BF outcomes. Indeed, 10 of the 12 articles excluded from the meta-analysis found statistically significant inverse relationships between prelacteal feeds and diverse BF outcomes consistent with the results of the meta-analysis. Moving forward, experimental interdisciplinary research studies are needed to identify the most promising interventions to address the risk factors for prelacteal feeding and the introduction of BMS during the neonatal period. To optimize the successful implementation of these interventions, it is important to codesign them with the target population and the providers serving them, following a health care systems implementation framework that takes the social determinants of health into account (Boccolini et al., 2015;Kavle et al., 2017;Nguyen et al., 2020;Pérez-Escamilla & Sellen, 2015;Tomori et al., 2022). We found no prospective studies conducted in low-income countries. There is an urgent need to conduct prospective studies to understand the effect of prelacteal feeds and neonatal introduction of BMS on BF outcomes in these settings, particularly given that BF is critical for promoting optimal maternal and child health outcomes in low-income countries where women and children are at high risk of adverse outcomes for which BF is protective. More prospective studies are needed to improve our understanding of the association between water-based prelacteals and BF outcomes. Lastly, studies need to clearly explain how they defined and measured prelacteal and later neonatal BMS feeding .

| CONCLUSION
In conclusion, this review found that the introduction of BMS during the early and late neonatal periods is a statistically significant risk factor for shorter BF duration. Effective interventions are needed to