Risk factors for self‐reported insufficient milk during the first 6 months of life: A systematic review

Abstract The objective of this systematic review was to identify multifactorial risk factors for self‐reported insufficient milk (SRIM) and delayed onset of lactation (DOL). The review protocol was registered a priori in PROSPERO (ID# CDR42021240413). Of the 120 studies included (98 on SRIM, 18 on DOL, and 4 both), 37 (31%) studies were conducted in North America, followed by 26 (21.6%) in Europe, 25 (21%) in East Asia, and Pacific, 15 (12.5%) in Latin America and the Caribbean, 7 (6%) in the Middle East and North Africa, 5 (4%) in South Asia, 3 (2.5%) in Sub‐Saharan Africa, and 2 (1.7%) included multiple countries. A total of 79 studies were from high‐income countries, 30 from upper‐middle‐income, 10 from low‐middle‐income countries, and one study was conducted in a high‐income and an upper‐middle‐income country. Findings indicated that DOL increased the risk of SRIM. Protective factors identified for DOL and SRIM were hospital practices, such as timely breastfeeding (BF) initiation, avoiding in‐hospital commercial milk formula supplementation, and BF counselling/support. By contrast, maternal overweight/obesity, caesarean section, and poor maternal physical and mental health were risk factors for DOL and SRIM. SRIM was associated with primiparity, the mother's interpretation of the baby's fussiness or crying, and low maternal BF self‐efficacy. Biomedical factors including epidural anaesthesia and prolonged stage II labour were associated with DOL. Thus, to protect against SRIM and DOL it is key to prevent unnecessary caesarean sections, implement the Baby‐Friendly Ten Steps at maternity facilities, and provide BF counselling that includes baby behaviours.


| INTRODUCTION
The World Health Organization's (WHO) recommendation of exclusive breastfeeding (EBF) during the first 6 months, the introduction of complementary foods at this age, and breastfeeding (BF) continuation for at least the first 2 years of life continues to be supported by scientific evidence (Bartick et al., 2017;Chowdhury et al., 2015;Li et al., 2022;Pérez-Escamilla et al., 2019;Victora et al., 2016). The fact that only 48% of children under 6 months old living in low-income and middle-income countries are exclusively breastfed and less than 70% of infants in low-and middle-income countries are breastfed during their second year of life, with some regions having less than half of infants continuing BF beyond 1 year of age (UNICEF, 2022) is of public health concern. This is because of the numerous well-known health, nutrition, and cognitive benefits that BF provides to children and women (Tschiderer et al., 2022;Victora et al., 2016). It has been estimated that over 800,000 annual deaths can be prevented among children under 5 years old by following this EBF recommendation Victora et al., 2016;Walters et al., 2016). This benefit also applies to highincome countries; for example, a recent study using national data found BF initiation associated with a lower risk of post-perinatal infant deaths across racial groups within the US population (Li et al., 2022). In addition, BF is friendly to the environment as most CMFs are made from cow's milk, use enormous amounts of water in their manufacturing, and leave large quantities of nonbiodegradable waste behind (Joffe et al., 2019;Smith, 2019).
Although several countries have been able to improve BF outcomes over the past two decades (Bhattacharjee et al., 2021;Neves et al., 2021), these improvements are not happening fast enough to achieve the 70% EBF goal by 2030 set by the Global Breastfeeding Collective led by WHO and UNICEF (Bhattacharjee et al., 2021). Hence, it is key to further understand how to address these breastfeeding challenges since birth.
For decades, BF problems have been commonly cited as one reason for early BF discontinuation, with the mother's complaint of not having enough milk, herein referred to as self-reported insufficient milk (SRIM), being identified as the most common problem for not initiating or stopping BF. Indeed, SRIM is a major public health concern. It is frequently reported from the neonatal period and remains the most frequently cited reason by women all over the world for introducing commercial milk formulas (CMF, oftentimes misleadingly referred to as breast milk substitutes) (Gatti, 2008;Hill & Humenick, 1989;Huang et al., 2021). CMF introduction, in turn, is a strong risk factor for shorter EBF and BF durations (Pérez-Escamilla et al., 2019;Segura-Millán et al., 1994).
There have been several attempts to define and identify factors influencing SRIM. In 1979, Butz (1979) claimed that SRIM, referring to it as 'Insufficient Milk Syndrome', was simply a culturally acceptable reason for stopping breastfeeding and thus a socially acceptable excuse (Butz, 1979). Gussler and Briesemeister disputed this as the sole explanation, describing mother's insufficient milk as a 'transcultural phenomenon' since it affected mothers of different cultures and backgrounds (Gussler & Briesemeister, 1980). They also noted that SRIM was documented even among mothers motivated to BF, and among those with both good or poor nutritional status. They proposed that modernisation and urbanisation disrupted traditional feeding patterns through the separation of mother and child, which was not conducive to supporting breastmilk production and led to mother's interpretation of real or perceived insufficient milk. Thus, 'Syndrome' was described as 'characterized by the lack of "constant contact" between mother and infant in modern urban settings' (Gussler & Briesemeister, 1980). The following year posited that decreased sucking stimulation of the nipples as a result of supplemental feedings was the most likely explanation for SRIM (Greiner et al., 1981).
In 1985, researchers agreed that SRIM was unlikely to be explained by a single factor and conceptualised SRIM as an outcome determined by a complex combination of factors, including maternal-child biological factors, sociocultural factors, health care practices and breastfeeding knowledge (Tully & Dewey, 1985). Consistent with this comprehensive approach to SRIM, findings from a literature review were used to propose an insufficient milk supply conceptual framework based on multifactorial determinants and mediators of milk production (Hill & Humenick, 1989).
The multifactorial determinants fell into four categories (maternal time constraints, sociocultural factors, maternal comfort factors and infant factors) and the mediating factors included three categories (breastfeeding behaviour, maternal psychologic factors, maternal physiological factors). SRIM has been conceptualised as 'a state in which a mother has or perceives that she has an inadequate supply of breastmilk to either satisfy her infant's hunger and/or support the infant's adequate weight gain' (Hill & Humenick, 1989). Consistent with prior studies (Hill & Humenick, 1989;Huang et al., 2021;Mohebati et al., 2021;Segura-Millán et al., 1994;Tully & Dewey, 1985), a subsequent SRIM literature review

Key messages
• Socioeconomic and demographic factors indicative of lower economic status increased the risk of self-reported insufficient milk (SRIM) and delayed onset of lactation (DOL).
• Timely breastfeeding (BF) initiation and avoidance of inhospital commercial milk formula (CMF) supplementation are likely to reduce the prevalence of SRIM and DOL.
• BF counselling designed to prevent SRIM and DOL needs to strengthen maternal BF self-efficacy, maternal understanding of baby behaviours such as fussiness and maternal wellbeing.
• Research is needed to better understand how the risk of DOL and SRIM increases with primiparity, caesarean sections, maternal overweight/obesity, and poor overall maternal health.
• Intervention studies specifically designed to reduce the risk of SRIM and DOL are urgently needed in low-and middle-income countries. (Gatti, 2008) found that mother's self-assessment of her milk supply was often associated with her perception of infant satiety or satisfaction mainly based on her interpretation of infant behaviours, especially crying or fussiness.
As mentioned above, while some researchers have suggested that SRIM is a sociocultural phenomenon, others have interpreted it as having physiological or biological causes or argued that it needs to be understood from a biocultural and behavioural perspective (Hill & Humenick, 1989). Yet, to date there has not been any systematic global analysis of SRIM risk factors. Hence, there is a need to systematically review the multiple factors contributing to SRIM, understand how they map across different socioeconomic, demographic, bicultural, psychobehavioural and health care systems domains, and identify pragmatic recommendations on how to address those factors that are modifiable.
Building on previous frameworks and empirical evidence (Chapman & Pérez-Escamilla, 1999b;Dewey et al., 2003;Matias et al., 2010;Nommsen-Rivers et al., 2010;Segura-Millán et al., 1994), our research team recently suggested that in many instances, SRIM starts very early as a result of lack of information on what to expect during the colostrum phase or actually delayed onset of lactation (DOL), defined as milk 'coming in >72 h post-partum' (Chapman & Pérez-Escamilla, 1999b. The introduction of CMF products can then delay the onset of lactation even further, interfering with the establishment of the milk supply and increasing the risk of SRIM (Pérez-Escamilla et al., 2019). Furthermore, others argue that the lack of access to qualified lactation counselling and stress management skills during the first days after birth, together with lack of knowledge among caregivers and/or family members on infant behaviours such as crying, push women into a vicious cycle that can lead to actual insufficient milk production (Karall et al., 2015). Researchers have acknowledged that maternal obesity should now also be considered a risk factor for DOL, SRIM, and shorter breastfeeding duration. This is because of consistent epidemiological evidence and strong biological plausibility indicating that maternal obesity can disrupt human lactation as a result of endocrinological alterations, mechanical barriers (large breasts preventing effective infant latch) and/or psychoemotional challenges such as low self-esteem (Amir & Donath, 2007;Chang et al., 2020;Chapman et al., 2013;Pérez-Escamilla et al., 2019). In short, DOL is a special case of SRIM as it happens during the period of time before ample milk secretion begins, known as stage II lactogenesis or lactation secretory activation stage (Boss et al., 2018). DOL is of concern because it has been associated with shorter EBF and BF durations (Chapman & Pérez-Escamilla, 1999a;Huang et al., 2020).
To date, there are no reviews that have comprehensively synthesised the vast literature on DOL and SRIM. The main overall aim of this review is to increase the understanding of factors affecting SRIM and DOL to support the development and testing of interventions to improve BF exclusivity and duration. Thus, the objective of this systematic review is to answer the following questions: (1) Which socioeconomic, demographic, and/or cultural factors increase the risk for SRIM, including DOL.
(2) Which behavioural and biomedical factors increase the risk for SRIM and DOL.

| METHODS
The study protocol was developed and registered a priori in PROSPERO (ID#CDR42021240413). This review focus on studies with mothers and infants with no serious conditions that impede BF. The two main outcomes for the review were: (1) SRIM, defined as maternal report of not having 'enough' or 'sufficient' milk (e.g., not producing enough milk, milk dried up, baby hungry after feeding, not enough to satisfy the needs of the infant or DOL), as a reason for not initiating BF, stopping BF or introducing CMF; and (2) DOL, defined as perception of initiation of ample milk secretion beyond 72 h post-partum. Review findings were reported following the preferred reporting systematic review and metaanalysis protocols (Page et al., 2021).

| Inclusion and exclusion criteria
Studies reporting SRIM or DOL were included only if they met the following criteria: (a) absence of serious maternal complications due to childbirth that might impede a timely initiation of breastfeeding such as severe post-partum haemorrhaging; (b) post-partum women delivering a singleton full-term healthy baby infant or with no more than 10% of data coming from low-birthweight or pre-term infants; (c) quantitative studies with no design restrictions but with a comparison group or exposure; (d) only studies in English, Spanish or Portuguese; (e) studies conducted in high-, middle-or low-income countries; and (f) analysis of the association between mothers reporting DOL or SRIM during the first 6 months post-partum and one or more of the following type of risk factors: sociocultural, economic, behavioural, knowledge or biomedical.
Studies were excluded from the systematic review if they were qualitative studies, reviews, systematic reviews, meta-analyses or quantitative studies with no comparison group. Studies that focused only on premature or low-birthweight infants, included mothers or infants with conditions that might preclude breastfeeding, or were in a language outside of those mentioned in the inclusion criteria were also excluded.

| Search strategy
We used a comprehensive search strategy developed by a medical librarian (K.N.), tested against validation articles previously identified by the authors, using both controlled vocabulary and free-text queries. An independent medical librarian peer-reviewed our electronic strategy using the Peer Review Electronic Search Strategies (PRESS) guidelines (McGowan et al., 2016). To prevent the omission of relevant studies, we used backward citation chaining, which involved reviewing the reference lists of articles identified and those from relevant literature reviews. An initial exploratory search was conducted using a list of terms under the following concepts: (a) reasons for weaning or mixed feeding, or (b) SRIM (e.g., not enough milk, milk dried up, baby not full) or (c) DOL.
The full search was conducted in the following databases in April 2021: 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), the Virtual Health Library Regional Portal (including LILACS), Scielo and Global Index Medicus. There were no time limits specified for this search. The complete final search strategy for MEDLINE is presented in Table 1, and all the searches are available at https://osf.io/jkx6s/.

| Study selection process and data extraction
Covidence online software was used to conduct the screening process.
Two of the researchers independently screened the first 200 titles and abstracts and compared their inclusion or exclusion assessments.
Differences were resolved through a consensus process facilitated by the senior author (R. P. E.). Following this, three researchers (S. S. P., M. A., and R. P. E.) proceeded with independently reviewing the remaining titles and abstracts of each publication in Covidence (n = 8562), identifying 984 studies for full-text review that were reviewed by all three authors.
Discrepancies were resolved until consensus was reached among authors on the final list of included articles (Figure 1). Two researchers (S. S. P. and R. P. E.) extracted the following data from the included articles: study design, main outcomes, population and setting, main independent variable, other control variables, type of analysis, key findings and information required for quality assessment.

| Quality assessment and risk of bias
The Joanna Briggs Institute (JBI) critical appraisal tools appropriate for different study designs were used to assess the quality of observational and experimental studies (Munn et al., 2014). While JBI endorses the GRADE approach for systematic reviews and has similar approaches to assessing risk of bias, it has developed a wider variety of critical appraisal checklists for different research designs. Specifically, the JBI checklist for cross-sectional studies is one of the newest and preferred tools for assessing the quality of evidence in systematic reviews. The critical appraisal checklists use a binary scoring process (i.e., yes/no) to assess quality, which can graphically display assessments of the methodological strengths and weakness of the literature, and it can also be transformed into scores assessing the evidence of the reviewed studies in a similar manner to the GRADE.

| Search outcomes
Our search in seven databases identified 19,187 records, with no additional articles obtained from manual search of references and websites. After removing 6316 duplicates via the Yale deduplicator and 4309 duplicates in Covidence, 8562 records remained for screening.
After reading through titles and abstracts we identified 7578 articles that were not related to SRIM or DOL leaving 984 articles meeting the eligibility criteria for full-text review. A total of 126 studies were initially identified as meeting the inclusion criteria and six more articles were excluded during extraction; three articles were excluded due to the design not qualifying and three were excluded because they only reported SRIM prevalence. Finally, 120 studies qualified for inclusion for this review (Figure 1).
1646 10 (breastfe* or breast fe* or infant feeding).mp. or exp Infant Nutritional Physiological Phenomena/ 82,319 11 (milk adj5 (early or low or insufficient or sufficient or inadequate or adequate or problems or perceived or perception or volume or supply or production)).mp.
19,709 12 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or (10 and 11) 4422 introducing other liquids or solids (n = 9), or as part of a list of maternal BF problems or concerns (n = 33). Among these SRIM studies, it was not possible to tease out when SRIM referred to insufficient milk production, low milk quality, or both. Of the studies reporting maternal perception of onset of lactation, DOL was defined in different ways.
Studies identified DOL (>72 h post-partum) based on maternal selfreport of breast fullness symptoms (N = 8) and perception of when their milk came in (N = 6). The remaining DOL studies used a variety of approaches to determine maternal self-report of DOL.
The overall characteristics of the included studies, including SRIM and DOL prevalence reported, are summarised in Tables 3 and 4 and detailed information including statistical coefficients of associations can be found as online Supporting Information Materials (Appendix SA).

| Prospective studies
Of the 11 items included in the assessment of the prospective studies (n = 39), most studies (37 out of 39) met five of the criteria. Twelve studies were classified as not having performed adequate statistical analyses, and 20 did not address missing data due to incomplete follow-up ( Figure 2b).

| Quasi-experimental studies
Among the nine items that were used to assess the quality of the nine quasi-experimental studies, a significant proportion had deficiencies in statistical analyses mainly due to not properly accounting for potential confounding (Figure 2c). H&H scale (Hill & Humenick, 1996) 20 items addressing 3 constructs: maternal BF confidence/commitment; perceived infant satiety (5 items); maternal/infant breastfeeding satisfaction (5 items Based on insufficient milk syndrome framework (Hill & Humenick, 1989) SRIM problems score (Henly et al., 1995) Mother's rating of the adequacy of her milk supply; constructs: insufficient letdown, not enough milk, baby weight gain, frequency of nursing, and baby crying or dissatisfied; 7 point Likert-scale (0 [no problem] to 7 [major problem]); higher score = more insufficient milk perception Duckett et al. (1998) BF perception questionnaire 5 items; 1 on mother's level of BF confidence (1 [strongly agree]-4 [strongly disagree]), 4 rating mother's perception of: infant feeding frequency (not often enough, normal, too often); amount of baby takes each time (too little, normal, too much); time infant takes at each feed (too slow, normal, too fast); mother's perception of her milk production (too low, normal, too high) Kent et al. (2015) BF perception questionnaire Insufficient milk perception SRIM as a BF concern, assessed with 6-point

| Experimental studies
Five of the 10 RCTs met the criteria for at least 9 items out of 12 items. However only two met the outcomes assessors; four the intervention; five met the participant allocation blindness criteria; seven met the trial design, randomisation, and follow-up criteria, and eight met the statistical analysis criteria (Figure 2d).

| SRIM
There were 63 risk factors identified and categorised into seven domains: socioeconomic and cultural; demographic; psychosocial and behavioural; health care systems; biomedical; breastfeeding knowledge, styles, and problems, and maternal lifestyles (Supporting Information Appendix SA). The following section summarises individual or groups of risk factors that were found to be consistently associated with SRIM, meaning that over half of studies examining a specific risk factor documented the significant association (Tables 3 and 5).

Socioeconomic and demographic
The SRIM risk factors very consistently identified were low maternal All studies (n = 8) comparing ethnic/racial groups within-countries or mothers across countries, and three of four studies comparing area of residence characteristics including urban/rural, found significant associations of these demographic variables with SRIM tending to be comparatively higher among women of relatively disadvantaged groups, at higher SRIM risk were: Jewish versus Arab women (Heldenberg et al., 1993); Japanese versus Caucasian women (Hla et al., 2003); Hispanic (vs. African American, White) women (Hurley et al., 2008;Rozga et al., 2015); Hispanic (vs. White) women (Li et al., 2008); non-Caucasian (vs. Caucasian) women (Kent et al., 2021;Williams et al., 1999); rural versus urban (Mosha et al., 1998), Japanese versus French/US (Negayama et al., 2012), outside United States Midwest versus other parts of the United States (Kirkland & Fein, 2003). Findings from 19 studies did not show clear patterns of associations between SRIM and infant age (Table 3).

Social support, psychosocial and behavioural
SRIM protective factors related to social support and psychosocial indicators were identified.
Excessive newborn weight loss or perception that infant was not growing well were also identified as SRIM risk factors (five out of five studies) (Flaherman et al., 2016;Hill & Aldag, 1991;Hillerviklindquist et al., 1991;Kent et al., 2021;Moll Pons et al., 2012;O'Sullivan et al., 2015). Three of four studies identified low-birth-weight or prematurity as SRIM risk factors.
Additional risk factors for SRIM included early introduction of CMF (Segura-Millán et al., 1994), displacement of breast milk by solids between BF episodes (Hillerviklindquist et al., 1991) or mixed feeding Kent et al., 2021) (four out of four studies).
In sum, a graphic synthesis of our findings suggests that SRIM is determined by distal (socioeconomic and demographic), intermediate (social support, psychoemotional and baby behaviours, maternity ward practices, biomedical) and proximal (CMF supplementation, and BF challenges) factors (see conceptual framework in Figure 3).

Socioeconomic and demographic
Studies identified household poverty (Brownell et al., 2012;Haile et al., 2017) maternal employment (Brownell et al., 2012) were socioeconomic indicators associated with DOL in our review.
Demographic factors such as primiparity was identified as a risk factor in three studies (Brownell et al., 2012;Huang, Li et al., 2020;Scott et al., 2007), and in two more were linked to an interaction, only among primiparous who pumped (Chapman & Perez-Escamilla, 2000) or among primiparas who delivered a large infant (Dewey et al., 2003).

Maternal health and lifestyles
Two studies found that maternal anxiety and depression were risk factors for DOL (Flaherman et al., 2016;Rocha et al., 2020). Studies also showed that poor maternal sleep ( (Brownell et al., 2012) and tobacco (Brownell et al., 2012) use were risk factors for DOL.
In sum, a graphic synthesis of our findings suggests that DOL is determined by distal (socioeconomic and demographic), intermediate (social support, maternal lifestyles, maternity ward practices, and breastfeeding counseling, biomedical) and proximal (CMF) supplementation, and factors (see conceptual framework in Figure 4).

| DISCUSSION
Our highly comprehensive systematic SRIM review, the first of its kind as far as we know, indicates that DOL and SRIM continue to be highly prevalent and that both SRIM and DOL are associated with multiple factors distributed across socioeconomic, demographic, support systems, health care systems, psychosocial and behavioural, and breastfeeding and human lactation domains.
Our findings showed that delayed BF initiation, separation rather than rooming-in, in-hospital CMF supplementation, lack of BF counseling, and other maternity practices inconsistent with the Ten Steps were risk factors for both DOL and SRIM. Since not following these practices was associated with an increase in the risk for DOL and SRIM, we strongly recommend that future studies examining the impact of the Baby-Friendly Hospital Initiative Ten Steps also include DOL and SRIM as primary outcomes.
Our review strongly affirmed the multifactorial nature of SRIM and the importance of designing interventions to address the constellation of risk factors identified. We documented a relationship of maternal breastfeeding self-efficacy with a reduced risk of SRIM and longer breastfeeding duration. Hence, it is important for breastfeeding counseling programmes to prepare women from pregnancy and the early post-partum period by building their confidence toward establishing an ample milk supply to nourish their infants. Counseling may be especially important for primiparous women, as they were consistently found to be at higher risk of SRIM.
Our review strongly suggests that to prevent SRIM it is important to improve caregivers' and healthcare professionals' understanding and management of baby behaviours perceived to be 'difficult' prevent SRIM. It is likely that psychoemotional support from counsellors, family, friends, and healthcare providers as part of breastfeeding programmes will lead to improving maternal BF selfefficacy, which was consistently found in our review to be a protective factor against SRIM.
Regarding biomedical factors including obstetric practices, our review consistently showed that maternal pre-pregnancy obesity was consistently identified as a risk factor of DOL. Furthermore, C-sections and maternal overweight or obesity were risk factors for both DOL and SRIM. Therefore, interventions are needed to both prevent unnecessary C-sections and maternal excessive body fat and weight gain during pregnancy and to provide additional needed lactation support to women exposed to these highly prevalent risk factors (Pérez-Escamilla et al., 2019).
Consistent with a previous review  we found that women of lower socioeconomic status, as reflected by household income and education level, are more likely to report DOL and SRIM.
Likewise, there were differences in risk between ethnic/racial groups or urban versus rural areas within countries and when comparing women across countries. Hence, addressing this global public health concern should be done through an equity lens focusing on the social determinants of health (Pérez-Escamilla, 2020; Pérez-Escamilla & Sellen, 2015) mediating the relationship between poverty and SRIM.
Overall, our review supports that SRIM may be a concern that starts since the colostrum stage when there is very little milk  (Karall et al., 2015;Pérez-Escamilla et al., 2019)  In the previous literature, PIM was used to refer to 'perceived insufficient milk'. We chose not to use this term in our review because 'perception' has been taken to imply that the milk insufficiency mothers are reporting is often times not real and simply given as a socially acceptable excuse by women. Instead, we coined the term SRIM to describe more accurately the phenomena of interest in this article and it is non-judgmental.
As stated above, it should not be assumed that SRIM does not reflect real milk insufficiency (Pérez-Escamilla et al., 2019;Stuebe, 2021). It is important for the hydration status of all newborns to be closely monitored. It is also important to empower women to work together with their health care providers to ensure that their infants are growing and developing well. It is crucial for all women and their infant feeding support networks to receive anticipatory guidance and counseling starting at the beginning of pregnancy and continuing through the perinatally and post-natally periods on the different stages of lactation (lactation secretory activation, establishment, and maintenance; Boss et al., 2018) and what to expect with regard to milk production during each of them.
Specifically, counsellors should prepare mothers what to expect with regard to their milk production before, during, and after the onset of lactation; allay fears about not having enough milk; counsel them on how to establish if their infants are getting enough nourishment through breastfeeding; and support them in learning and properly applying breastfeeding techniques that are key to promoting sufficient milk supply and managing SRIM and DOL as necessary-for example, through increased nursing frequency during infant growth spurts (Galipeau et al., 2018). All considered, this is the only way caregivers can be reassured that the breastfed infant is receiving adequate nourishment (Pérez-Escamilla et al., 2019).

| Data gaps and multidisciplinary research recommendations
We identified the following key gaps in our knowledge of SRIM and DOL and research priorities.
First, there is a strong need for studies in low-and low-middleincome countries. These studies should focus on the potential role of maternal malnutrition, both under and overweight and micronutrient deficiencies, on DOL and SRIM, and use objective measures of breast milk volume and quality. It is particularly important to prioritise SRIM and DOL prevention among infants less than 6 months old to prevent wasting taking into account available BF assessment tools for at-risk and malnourished infants aged under 6 months old (Brugaletta et al., 2020;UNICEF, 2021).
Second, studies are needed to examine the longer-term implications of SRIM on the risk of illness episodes; in addition to its potential impact on the anthropometric status and growth trajectories of infants. Seventh, understanding must be improved on how best to support the onset of milk production and other breastfeeding needs SEGURA-PÉREZ ET AL.
Eighth, it is key to reach a consensus on the best ways to measure SRIM to track trends in prevalence and risk factors through monitoring and surveillance systems.

| Practice implications
Our review has several important public health practice implica-

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS
Sofia Segura-Pérez and Rafael Pérez-Escamilla led the conceptualising and drafting of the protocol for the systematic review, reviewed abstracts, titles and manuscripts, extracted study data, assessed study quality and drafted the full manuscript. Misikir Adnew contributed with titles screening and full-text reviews.
Rafael Pérez-Escamilla provided guidance in decisions to include or exclude specific studies when consensus was not initially reached between Sofia Segura-Pérez and Misikir Adnew. Kate Nyhan developed and tested the search strategy, conducted the search, and contributed to defining quality assessment tools. Linda Richter, Elizabeth C. Rhodes, Amber Hromi-Fiedler, Mireya Vilar-Compte, and Misikir Adnew contributed to the conceptualisation and draft of the protocol for the systematic review, supported manuscript development, and provided a critical review of the full manuscript. All authors read and approved the submitted version of the manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.