The economic cost consequences of suboptimal infant and young child feeding practices: a scoping review

Abstract

Breastfeeding is important for women and children’s health, but less than half of infants worldwide begin life with optimal breastfeeding. A growing literature shows consistently large economic costs of not breastfeeding, with global studies showing economic losses of around US$300 billion globally. However, existing studies are highly diverse in approaches, methods, data sources and country results. Building on a landmark 2012 UNICEF UK review focused on high-income countries, we conducted a scoping review to map and characterize the expanding literature and identify future research directions in this research area. We included studies (n = 36) in diverse country settings and outcomes for women and children. We used PubMed, Web of Science, EMBASE, MEDLINE, ProQuest and manual searches of cost of not breastfeeding studies published between 1996 and 2023. Articles were excluded if they were macroeconomic evaluations, did not assign monetary values or only evaluated breastfeeding or formula feeding costs and not outcomes or were cost of programs studies. We found considerable diversity in disciplinary approaches and differences in methodologies. Though there were different cost measurement perspectives (societal, institutional/payer and individual), all but two excluded the costs of unpaid care. Studies typically measured costs of medical treatment, with more recent studies using dynamic simulation models. The largest economic costs were derived from lifetime estimates of human capital losses, namely cost of premature death and loss of intelligence quotient points. Medical and death costs varied widely depending on method of calculation, but total costs consistently exceeded $US100 billion annually for the USA, and around $US300 billion in global studies. Our findings suggest that greater interdisciplinary collaboration is needed particularly to better define infant feeding exposures, and advance comprehensive measurement of costs and outcomes across lifetimes, in order to prioritize breastfeeding as a public health strategy of economic importance.

Suboptimal early feeding, not breastfeeding, cost consequence

Key Messages

• In studies that examine the cost of suboptimal infant feeding, there is considerable diversity in disciplinary approaches and methodologies. Greater interdisciplinary collaboration may strengthen the prioritization of the economic importance of this topic.

• The optimal infant feeding practice of breastfeeding requires maternal time. Yet almost all studies, even those from a ‘societal’ perspective neglect the time costs to mothers.

• The cost of suboptimal infant feeding practices is consistently high, around US$100 billion annually for the USA alone. Costs from child cognitive losses make up the bulk of the costs of suboptimal infant feeding practices, and costs from premature maternal deaths are larger than medical costs in the identified studies.

Introduction

Globally, less than half of infants begin life with optimal breastfeeding (UNICEF, 2019) and this has economic and financial consequences. Breastfeeding is important for public health in all country settings because its premature cessation generates avoidable health risks and costs related to the health and development of children and the health of women. Specifically, breastfeeding is associated with decreased maternal risk for breast and ovarian cancer, myocardial infarction, hypertension, diabetes and stroke and with decreased infant/child risk for otitis media, gastrointestinal illnesses, obesity, leukaemia, necrotizing enterocolitis (NEC) and sudden infant death syndrome (Rollins et al., 2016; Victora et al., 2016; Meek and Noble, 2022).

These conditions have significant economic costs to families, employers and society. The 2016 Lancet series demonstrated the global loss of life attributable to suboptimal breastfeeding for 96 high-, middle- and low-income countries was 823 000 child and 20 000 maternal deaths and resulted in in cognitive losses of US$302 billion annually (Rollins et al., 2016; Victora et al., 2016). The economic cost of lower cognition was modelled using estimates that any breastfeeding duration to 6 months is associated with a ∼3 point increase in intelligence quotient (IQ) score, with subsequent effects on educational attainment and adult labour force earnings. Rollins et al. also demonstrated substantial direct treatment costs of childhood illnesses in high- and middle-income countries.

This accumulation of evidence for health and economic losses attributable to suboptimal breastfeeding results both from improved access to methodological tools involving complex economic modelling software, and to better epidemiological evidence delineating impacts of breastfeeding on health and development outcomes. Further, from 2015 the World Bank acknowledged evidence of inadequate global investment in breastfeeding (Holla-Bhar et al., 2015), with strong statements from its leadership (Hansen, 2016) and publication of its Investment Framework for Nutrition in 2017 (Shekar et al., 2017) showing potential economic gains (net present value of ∼ US$298 billion annually) of reaching the 2025 global nutrition targets for exclusive breastfeeding, by preventing child cognitive losses and child mortality, with a return of $35 on each dollar invested in increasing breastfeeding (Kakietek et al., 2017). The World Bank analysis also indicated that the global nutrition targets for exclusive breastfeeding were unambitious, signalling that even greater economic gains were feasible if targets were revised (Shekar et al., 2017).

Heightened awareness of the economic costs of not breastfeeding according to WHO/UNICEF global recommendations has spurred development of accessible online tools to measure global and/or country level impacts (Stuebe et al., 2017; Walters et al., 2019; Smith et al., 2023). The most recent tool for investigating both global and country-level economic impacts of not breastfeeding is a collaboration led by Walters with the international not-for-profit, Alive & Thrive (Walters et al., 2019). This study reported aggregate global economic losses of US$341.3 billion, arising from health system treatment costs (US$1.1 billion annually), premature deaths of women and children (US$53.7 billion) and child cognitive losses (US$285 billion), using an online tool. Stuebe and colleagues offer an interactive online calculator to estimate the effects of changes in breastfeeding rates on healthcare costs of infectious illnesses for the first year of life. The online tool developed by Smith et al. (2023) identifies the global and country level quantities of human milk production that are lost at current breastfeeding rates, alongside estimates of quantities of milk produced by women at current breastfeeding rates, and uses market prices to indicate monetary values of these indicators.

These new summative works and tools are a boon to researchers and policymakers as they help assemble the case for global agencies and countries to invest in breastfeeding protection, support and promotion. However, because of differences in approaches, methods and data sources, and how costs are identified, measured and valued, this literature can generate inconsistent or conflicting economic cost estimates at the country level, and confusion about their conceptual and empirical basis. No existing review has drawn together the literature on the economic costs of not breastfeeding to help navigate the growing number of disparate and complex cross-disciplinary studies. Thus, this study aimed to bridge this critical gap by comprehensively mapping and summarizing the important features of the existing research and identifying the main implications for future research.

Methods

We chose to conduct a scoping review because our purpose was to characterize the literature and identify research gaps, which best aligns with the definition of a scoping review (Grant and Booth, 2009; Munn et al., 2018). Additionally, there are a relatively small number of very heterogeneous and cross-disciplinary studies, making strict adherence to the PRISMA checklist problematic. Our review follows the ethos of PRISMA by using a documented and structured approach in our review as far as practicable to ensure our mapping and synthesis of the literature is replicable and robust.

We took a multifaceted approach to identifying the recent evidence on the economic costs of not breastfeeding. The basis of our approach was a simplified update and expansion of the first major systematic review of cost of illness studies and breastfeeding performed by Renfrew and colleagues for UNICEF UK (Renfrew et al., 2012). Renfrew’s team used standard economic evaluation techniques identified by Drummond et al. to review cost of illness for ‘the UK and comparable industrial countries’ published between 1950 and 2010. Renfrew’s study was the most comprehensive analysis at the time that included a thorough review of the relevant economic literature. Since its publication, there have been a number of additional studies—some with newer and more sophisticated economic modelling. Our study built on that report by expanding the scope of the eligible literature, to include a more diverse geography [non-Organisation for Economic Cooperation and Development (OECD) countries] more child health and development outcomes (such as for obesity and chronic disease and cognitive development), a wider range of study types (including grey literature), and a general updating to include publications since 2010. We also include studies of chronic disease and maternal health outcomes which have utilized advances in computer software and techniques for dynamic modelling. One study used by Renfrew et al. was excluded from this analysis because it was limited to donor milk costs (Wight, 2001).

We surveyed the published peer reviewed literature and grey literature from 1996 to 2023 for all analyses which had an abstract in English that put a monetary value on cost savings of breastfeeding, the cost of not breastfeeding, the cost of suboptimal breastfeeding or the cost of artificial feeding. The year 1996 was selected as the starting point for our study because that is when the landmark paper by Horton et al. (1996) was published. This was the first health economic study to call attention to the cost-effectiveness of maternity services supporting breastfeeding, identifying both the costs and the impacts of breastfeeding promotion programmes.

Our search strategy for relevant literature was as follows. For peer-reviewed literature, we searched PubMed, Web of Science, EMBASE and MEDLINE, using the following keywords: suboptimal breastfeeding, inadequate breastfeeding, formula, artificial feeding, cost and cost analysis. For the grey literature, we searched breastfeeding and economic reports produced by national and international agencies including UNICEF, WHO, International Monetary Fund and the World Bank and we searched dissertation and theses abstracts available through the ProQuest service using their subject categories ‘breastfeeding & lactation’ and ‘costs’. We also examined references for both the peer-reviewed and grey literature to identify additional potential articles and reports.

Figure 1 summarizes our search process.

Figure 1.

Article inclusion and exclusion matrix

Articles were excluded if they:

1. were only a commentary or editorial (Langabeer, 2018; Smith, 2019);

2. had economically relevant outcomes but did not have a monetary value assigned to them, such as cost minimization studies only reporting hospital length of stay or the number of potential lives saved (Bhutta et al., 2013);

3. were studies of the cost of programs to increase breastfeeding, often referred to as costing studies (Renfrew et al., 2009; Frick et al., 2011; Holla-Bhar et al., 2015; Carroll et al., 2020);

4. examined only donor human milk (Arnold, 2002; Ganapathy et al., 2012; Hair et al., 2016, ) or only formula feeding (Pennock, 2002) or only infant feeding supplies (Berridge, 2004);

5. had insufficient information included (e.g. were only a conference abstract; Jegier, 2016);

6. could not be retrieved (Riedel, 2000); and

7. were a macroeconomic valuation, for example, Smith (1999), Gupta and Khanna (1999), Oshaug and Botten (1994) and Hatloy and Oshaug (1997), which have a different conceptual basis and measurement approach to cost of illness studies, and are considered elsewhere (Smith et al., 2023).

Definitions

We have provided the definitions used in this literature area for feeding practices, breastfeeding and early nutrition for ease of reference when comparing studies reviewed. Our study inclusion and exclusion criteria did not consider these definitions when determining eligibility. Our detailed data extraction tool included reporting on infant feeding exposures such as ‘any’ or ‘exclusive’ breastfeeding, as these are crucial for interpreting results and potential upward or downward bias in estimates.

‘Suboptimal infant and young child feeding practices’ was defined as a feeding behaviour and practices that did not meet recommended standards. WHO and UNICEF recommend:

• early initiation of breastfeeding within 1 h of birth;

• exclusive breastfeeding—meaning no other foods or liquids are provided, including water for the first 6 months of life; and

• introduction of nutritionally adequate and safe complementary (solid) foods at 6 months together with continued breastfeeding up to 2 years of age or beyond [World Health Organisation (WHO), 2021].

‘Suboptimal breastfeeding’ was defined as no breastfeeding, or breastfeeding for less than the current recommended duration or exclusivity according to medical definitions set by the WHO. The medical definition within individual papers may have been defined according to country specific organizations. For example, most papers conducted by authors from the USA used the American Academy of Paediatrics medical standards at the time of their publication.

‘Early feeding or early nutrition’ refers to the age group 0–36 months.

A longstanding and important issue in epidemiological research on the impacts of not breastfeeding is the precise measurement of ‘breastfeeding’ [see Labbok and Krasovec (1990) and Smith and Harvey (2011)]. ‘Breastfeeding’ in this study was defined both as an infant or young child suckling at the breast of a self-identified woman, and the provision of expressed milk. This definition is the most commonly used definition, although there are others in the published literature [World Health Organization (WHO), 2008; Eidelman et al., 2012].

Table 1 sets out our categorization and definitions of costing approaches evident in the literature. We have included the definition for all four categories although this study excluded macroeconomic papers.

Table 1.

Definition and description of costing approaches

Approach	Common methods of measurement	Limitations to method	Examples of this type of study
Micro-costing (‘bottom up’ approach)	Direct measurement of actual financial costs using cost-accounting data.	Relies on risk ratios which may not accurately define infant feeding exposure. Does not have proscriptive, universally accepted methodological approach to defining or measuring costs. May use insurance payment or provider charge data, but neither may represent actual costs. Rarely encompasses costs of mother’s time.	Ball and Wright, 1999; Jegier et al., 2010
Static modelling (‘top down’ approach)	Costs are extrapolated over a fixed period of time covering finite episodes of illness using epidemiologic risk ratios for illness or disease by type of infant feeding exposure.	Relies on risk ratios which may not accurately define infant feeding exposure. Not tracking cost impacts probabilistically over time so less accurate. Costing of episodes rather than total disease costs over a period. Limited by multiple assumptions on disease incidence in the case of chronic diseases, and wide variability in cost assumptions. Only accounts for health costs, not for mother’s time.	Drane, 1997; Bartick and Reinhold, 2010; Colchero et al., 2015; Rollins et al., 2016; Walters et al., 2019
Dynamic modelling (‘top down’ simulation approach)	Builds on the static modelling, but using a dynamic probabilistic (e.g. Markov chain) computer simulation (e.g. Monte Carlo) model. Usually measures cost impacts of multiple diseases in a large hypothetical population of women and/or their offspring over long periods of time, such as from birth to age 70.	As above for static computation. Relies on probabilistic assumptions about risk ratios by age. Discounting rate choices more influential on cost outcomes.	Bartick et al., 2013; Bartick et al., 2017; Unar-Munguia et al., 2017
Macroeconomic (‘Replacement cost; foregone value’)	Cost to a country of replacing maternal human milk at contemporary breastfeeding rates with breastmilk substitutes or donor human milk. Suboptimal breastfeeding cost may be calculated by subtracting this cost from the cost at an ideal or target breastfeeding rate, which would be the value of ‘lost’ milk due to low breastfeeding rates.	Relies accurately defining infant feeding exposure. Willingness to pay as a measure of value may not capture utility or social value. Prices and valuation based on willingness to pay may understate the economic value of breastmilk due to imperfect scientific knowledge of the health impacts of formula feeding. The cost of women’s time to breastfeed or express donor milk is not accounted for so it measures gross not net value-added.	Aguayo and Ross, 2002; Smith and Forrester, 2013, Lee, 2013, Lee, 2015,

The definitions we used to characterize the disciplinary approaches used in Table 2 were:

1. ‘Epidemiological’: evaluation derived primarily from the identification of comparative costs of treating cases and non-cases of disease based on an early feeding exposure such as not breastfeeding.

2. ‘Health economics’: evaluation derived primarily from the cost consequence of treating a given condition based on an exposure such as not breastfeeding. Cost consequences included direct and indirect costs of disease or mortality and/or used health loss or health gain measures such as disability-adjusted life years (DALY), quality-adjusted life year (QALY) or value of a statistical life (VSL).

3. ‘Nutritional economics’: evaluation derived primarily from the cost consequence of a feeding exposure, such as not breastfeeding on cognition losses (loss of IQ points) and lifetime work productivity from a societal perspective.

Procedures and analysis

Each paper was independently reviewed by J.S. and M.B. A shared electronic spreadsheet was used for data extraction. Data extraction used an expanded version of the template used by Renfrew et al. in their 2012 review of cost of illness studies (Renfrew et al., 2012). Data extraction for the review by Renfrew et al. included study characteristics, aspects of data and methods, including perspective, treatment cost and outcome measurement, and findings on the economic impact of breastfeeding (Renfrew et al., 2012). We recorded the details of the study purpose, hypothesis, process, disciplinary approach, cost methods used and outcomes.

Once extraction was complete, the entries were assessed for accuracy, completeness and consistency by B.J. who also resolved any discrepancies and finalized data extraction. Table 2 summarizes each of the studies reviewed. An expanded review is provided as a supplementary appendix.

Table 2.

Summary of reviewed articles

Author and date	Country	Disciplinary approach, methodology and perspective	Breastfeeding or formula milk exposure measured	Cost outcomes measured	Outcomes	Main findings	Author stated limitation
Horton et al., 1996	Brazil, Honduras, Mexico	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Static Modelling Perspective Societal	(1) Breastfeeding practices in hospital. (2) Breastfeeding at follow-up (1 month for all countries, 2 months in Brazil, 3 months in Honduras, and 4 months in Mexico).	(1) Annual cost of breastfeeding promotion program. (2) Annual cost per birth. (3) Incremental cost per birth. (4) Net cost per diarrhoeal case averted. (5) Net cost per diarrhoeal death averted. (6) DALY gained from diarrhoeal cases and deaths averted. (7) Summary of comparison costs per case and costs per death for other intervention programs.	Paediatrics Outcomes (1) Diarrhoea Maternal Outcomes None Other Outcomes (1) DALY gained	(1) Cost per birth in the control hospitals ranged from $0.09 to $8.74 and in the program hospitals from $2.70–$11.47. (2) Programs that focuses on reducing formula feeding cost $0.30 to $0.40 per birth and generate reduced cases and deaths, and gain DALY respectively of $0.65–$1.10, $100-$200, and $2–$4. (3) Programs that have already eliminated formula that invest $2–$3 per birth can generate reduced cases and deaths, and gain DALY respectively of $3.50–$6.75, $550–$800 and $12–$19.	Wherever a choice was available, the authors chose to use a conservative measure for estimates of morbidity, mortality, DALY, breastfeeding impact and cost. For example, their estimates were lower for disease and death than other literature estimates. This raises the risk that the findings are an underestimate from the true impact of breastfeeding on diarrhoeal morbidity and mortality.
Drane, 1997	Australia	Disciplinary Approach Primary: Epidemiological Secondary: None Methodological Approach Static Modelling Perspective Societal	Prevalence of exclusive breastfeeding at 3 months.	(1) Cost savings to health and social service system. (2) Measure data of economic significance for breast and formula feeding. (3) Cost utility of breastfeeding for preterm infants.	Paediatrics Outcomes (1) NEC in LBW. (2) Gastrointestinal illness in term. (3) Eczema in LBW. (4) Eczema in term. (5) Insulin-dependent diabetes in children and adolescents. (6) IQ in LBW. Maternal Outcomes None Other Outcomes None	(1) $11.75 million cost savings for health and social service systems with increase from 60% to 80% exclusive breastfeeding at 3 months. (2) $14.79 cost per QALY for the cheap lactation support model to $58.32 cost per QALY for the expensive lactation support model for preterm infants.	The analysis is likely an underestimate of the economic impact because the analysis relied on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration) and the estimates available may underestimate the impact of breastfeeding.
Hoey and Ware, 1997	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Payer	Breastfed or Formula fed exclusively for 6 months of life.	(1) Total utilization and costs on fee for service schedule for office visits by feeding type. (2) Total utilization and costs for hospitalization by feeding type. (3) Total utilization and costs for prescriptions by feeding type.	Paediatrics Outcomes (1) Visits to the office for medical care. (2) Prescriptions. (3) Hospitalizations. Maternal Outcomes None Other Outcomes None	(1) Formula-fed infants cost the health plan $200 more on average per infant for during the first year of life. This difference was not statistically significant.	(1) Small sample size. (2) Unable to control for demographic variance between infant feeding types. (3) Did not measure impact of partial breastfeeding. (4) Clinical severity for office visit was not measured.
Montgomery and Splett, 1997	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Payer	Breastfed exclusively for at least 3 months of life or Formula fed exclusively for the first 6 months of life.	(1) Total food costs by feeding type. (2) Total dollar value of claims paid by Medicaid for medical care for the infant during the first 6 months of life.	Paediatrics Outcomes (1) Health claims paid by Medicaid. (2) Formula and cereal costs incurred by WIC. Maternal Outcomes (1) Food package costs incurred by WIC. Other Outcomes None	(1) Compared to formula-fed infants, breastfed infants saved WIC and Medicaid US$478 per infant before accounting for the formula rebate and US$161 after accounting for the formula rebate. (2) Breastfed infants saved Medicaid is not statistically significant US$112 per infant in the first 6 months of life. (3) The combined cost for WIC and Medicaid to the taxpayer was statistically lower for breastfed infants (US$795) compared to formula-fed infants (US$956).	(1) Medicaid data was incomplete and did not allow for capture of costs from plans that paid in non-fee for service methods like health maintenance organizations. (2) WIC costs were tied to use of the voucher which may not always be captured.
Riordan, 1997	United States	Disciplinary Approach Primary: Epidemiological Secondary: None Methodological Approach Static Modelling Perspective Societal	Not explicitly stated (potentially varied by condition).	(1) Cost savings to health system for 4 paediatric health conditions. (2) Marginal cost to WIC for formula package.	Paediatrics Outcomes (1) Diarrhoeal/gastrointestinal illness in first year of life. (2) RSV in first year of life. (3) Otitis media in first year of life. (4) Insulin-dependent diabetes in childhood Maternal Outcomes None Other Outcomes WIC benefits	(1) $291.3 million for infant diarrhoea in non-breastfed infants. (2) $225 million for RSV in non-breastfed infants. (3) $9.6–$124.8 million insulin-dependent diabetes in non-breastfed children. (4) $660 million for otitis media in non-breastfed infants. (4) $2 665 715 in excess marginal cost for federal WIC benefits for non-breastfed infants.	The analysis is likely an underestimate of the economic impact because the analysis relied on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration) and the estimates available may underestimate the impact of breastfeeding.
Ball and Wright, 1999	Scotland and United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Payer	Breastfeeding exposure during the first 3 months of life (never, partial, exclusive).	(1) Total utilization and costs direct costs from practice records. (2) Total utilization and contractual daily rate for hospitalization from hospital records. (3) Total utilization and costs for prescriptions from pharmacy records. (4) Total utilization and costs for radiographs from hospital records.	Paediatrics Outcomes (1) Lower respiratory tract infection. (2) Otitis media. (3) Gastrointestinal infection. Maternal Outcomes None Other Outcomes None	(1) Compared to infants exclusively breastfed for 3 months, infants not breastfed had excess costs of between $331 and $475 during the first year of life.	The analysis is limited by the use of cost data from a single site.
Barton et al., 2001	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Institutional (Hospital)	Exclusive breastfeeding or formula feeding during the NICU stay	(1) Direct variable costs to the hospital. (2) Net revenue to the hospital. (3) Length of stay in days.	Paediatrics Outcomes (1) Weight gain. (2) Length of stay. (3) Days of parenteral nutrition. Maternal Outcomes None Other Outcomes None	(1) Compared to infants exclusively breastfed during the NICU stay, formula-fed infants had excess direct cost of approximately US$3300 dollars and higher net revenue of approximately US$10 800. Neither difference was statistically significant.	The analysis is limited by the use of clinical and economic data from a single site.
Weimer, 2001	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Prevalence of in-hospital breastfeeding and prevalence of any breastfeeding at 6 months.	(1) Excess cases and cost savings to health system for 3 paediatric health conditions. (2) Cost savings due to parental time and lost wages. (3) Cost of premature death.	Paediatrics Outcomes (1) Diarrhoeal/gastrointestinal illness in first year of life. (2) Otitis media in first 6 months of life. (3) NEC during NICU stay. (4) NEC deaths. Maternal Outcomes (1) Parental lost earnings Other Outcomes None	(1) $3.1 billion cost savings attributable to preventable NEC death. (2) $.5 billion cost savings due to direct medical and indirect non-medical costs related to 3 paediatric illnesses.	The analysis is likely an underestimate of the economic impact because the analysis relied on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration) and the estimates available may underestimate the impact of breastfeeding.
Hyun et al., 2002	South Korea	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Individual	Breastfeeding and formula feeding during the first year of life.	(1) Cost of formula. (2) Cost of formula supplies including bottles, teats, cleansers, sterilizers, brushes, cases for dried formula and washing liquid. (3) Cases and hospital costs for 4 paediatrics illnesses. (4) Additional food costs for lactating mother. (5) Cost of maternal absenteeism due to illness.	Paediatrics Outcomes (1) Respiratory disease (cold). (2) Otitis media. (3) Gastrointestinal disease. (4) Allergy Maternal Outcomes. (1) Food Other Outcomes (1) Formula. (2) Bottles. (3) Washing liquid. (4) Sterilizers. (5) Brushes. (6) Teats. (7) Case for dried formula.	(1) Compared to breastfed infants, formula-fed infants cost an extra ₩1.73 million per infant during the first year of life. (2) The mean cost of formula feeding was ₩ 1 870 125 compared to extra food costs for lactation of ₩203 004 during the first year of life. (3) Excess medical cost for formula fed infants was ₩62 920 for respiratory illnesses.	The article was not in English and thus this was not able to be determined.
Smith et al., 2002	Australia	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro costing and static modelling Perspective Institutional (Hospital)	Exclusive breastfeeding, predominant breastfeeding, and formula feeding for the first 24 weeks of life.	(1) Average cost of hospitalization in Australian public hospitals for defined DRGs for 5 paediatric illnesses.	Paediatrics Outcomes (1) Acute otitis media. (2) Gastrointestinal illness. (3) Respiratory illness. (4) Eczema. (5) NEC Maternal Outcomes None Other Outcomes None	(1) Less than 10% of infants are exclusively breastfed for the first six months of life. (2) Early weaning adds AUS$1–2 million in hospital costs for the 5 studied conditions.	The modelling matched hospital data to population risk estimates which may underestimate the impact of feeding type. The literature on the impact of breastfeeding vs formula feeding includes a lack of precision for duration and exclusivity which may mean an underestimate of the impact of early weaning on the risk of illness. The economic is limited to hospital costs and thus does not account for indirect and intangible costs.
Cattaneo et al., 2006	Italy	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Individual	Breastfeeding and formula feeding during the first year of life.	(1) Cost paid by the family for inpatient, outpatient, and emergency hospital care. (2) Cost paid by the family for private paediatrician visits. (3) Cost paid by the family for drugs from the pharmacy. (4) Estimated family cost for formula using market prices and number of days of formula feeding.	Paediatrics Outcomes (1) Episodes of care for conditions potentially related to breastfeeding or formula feeding (e.g. not related to congenital anomalies or accidents/trauma). Maternal Outcomes None Other Outcomes None	(1) Fully breastfed infants at 3 months had a lower cost of ambulatory (€34.69 vs €54.59 per infant/year) and hospital (€133.53 vs €254.03 per infant/year) healthcare compared to infants not breastfed or not fully breastfed. (2) Fully breastfed infants had statistically lower utilization of ambulatory and hospital healthcare. (3) The median cost for formula per infant per year was €247.90. (4) After adjusting for other factors that impact cost, breastfeeding for an additional month reduced the cost of healthcare by €20.79 and the cost of healthcare plus formula costs by €144.36.	Measurement error due to the inability to cost healthcare visits not paid by the family was the primary limitation stated. Misclassification error on infant feeding type was the second limitation because infant feeding data relied on self-report.
Büchner et al., 2008	Netherlands	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Prevalence of any breastfeeding through 6 months (measured monthly).	(1) Excess cases and cost savings to health system for 8 paediatric health conditions. (2) Excess cases and cost savings to health system for 3 maternal health conditions. (3) Change in DALY per 1000 newborns. (4) Change in healthcare costs per newborn.	Paediatrics Outcomes (1) Otitis media. (2) Gastrointestinal infection. (3) Asthma. (4) Respiratory infection. (5) Eczema. (6) Chron’s disease. (7) Leukemia. (8) Obesity Maternal Outcomes (1) Rheumatic arthritis. (2) Premenopausal breast cancer. (3) Ovarian cancer. Other Outcomes None	(1) $50 million euro net present value could be saved annually if all mothers breastfeed for at least 6 months. (2) Best Case: 100% breastfeeding for at least 6 months generates a gain in DALY of 28 per 1000 newborns and a healthcare cost savings of 250 euro per newborn. (3) Worst Case: 100% formula feeding generates a loss in DALY of 25 per 1000 newborns and excess healthcare cost of 220 euro per newborn. (4) A 5% shift in current breastfeeding behaviour would generate a cost savings of 20 euro per newborn and 0.002 DALY gain per newborn.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration) and the model could not capture impact of mixed feeding.
Bartick and Reinhold, 2010	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Any and exclusive breastfeeding were modelled for up to 12 months of age.	(1) Excess cases, direct costs, and indirect costs for 9 paediatrics illnesses. (2) Excess cases and VSL costs of premature death for 6 paediatric illnesses.	Paediatrics Outcomes (1) Gastroenteritis. (2) Lower respiratory track illness. (3) Otitis media. (4) NEC. (5) SIDS. (6) Obesity. (7) Atopic dermatitis. (8) Asthma. (9) Leukemia. (10) Type 1 Diabetes. Maternal Outcomes None Other Outcomes None	(1) If 90% of US families met medical recommendations to breastfeed exclusively for 6 months, we would see cost savings of $13 billion and prevent 911 excess deaths. (2) Cost savings include: $9.6 billion for premature death and $2.2 billion in medical costs.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration).
Jegier et al., 2010	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Individual	Not applicable	(1) Cost per 100 ml of human milk for mothers of VLBW infants during the NICU stay.	Paediatrics Outcomes None Maternal Outcomes (1) Maternal time to pump. Other Outcomes (1) Pump rental. (2) Pump kit.	(1) The cost per 100 ml of human milk was $0.95–$1.55 excluding maternal opportunity cost and was $2.60 to $6.18 including maternal opportunity cost. (2) Mean daily milk output for mothers of VLBW infants was 558.2 ml (SD: 320.7) and mean time spent pumping was 98.7 min (SD: 38.6). (3) The cost per 100 ml was most sensitive to the cost of the breast pump rental and the breast pump kit.	The analysis was limited to only those items universally required by all mothers for the provision of human milk to a NICU infant.
Shin, 2010	South Korea	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics and Macroeconomics Methodological Approach Static Modelling Perspective Societal	Exclusive breastfeeding.	(1) Cost of formula. (2) Cost of formula supplies including bottles, cleansers, bottle sterilizers, brushes and clamps. (3) Excess cases, hospital costs, and hospital days for 7 paediatrics illnesses. (4) Excess cases, hospital costs and hospital days for 3 maternal illnesses. (5) Cost of maternal absenteeism due to illness.	Paediatrics Outcomes (1) Otitis media. (2) Pneumonia. (3) Other respiratory infections. (4) Gastroenteritis. (5) Urinary tract infection. (6) Sepsis. (7) NEC. Maternal Outcomes (1) Breast cancer. (2) Ovarian cancer. (3) Maternal absence from work due to illness. Other Outcomes (1) Formula. (2) Bottles. (3) Cleansers. (4) Bottle sterilizers. (5) Brushes. (6) Clamps.	By increasing exclusive breastfeeding rates from 35% to 50%. (1) ₩216.4 to ₩407.5 billion wons can be saved per year overall. (2) ₩162 to ₩294 billion wons can be saved per year related to formula and formula equipment. (3) ₩7.9 to ₩13.8 billion wons can be saved per year related to infant illness. (4) ₩24.8 to ₩57.7 billion wons can be saved per year related to maternal illness. (5) ₩21.6 to ₩42.5 billion wons can be saved per year related to maternal absenteeism from work.	The article was not in English and thus this was not able to be determined.
Renfrew et al., 2012	UK	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Any and exclusive breastfeeding were modelled for up to 6 months of age.	(1) Excess cases and direct treatment costs for 5 paediatrics illnesses. (2) Excess cases and direct treatment costs for 1 maternal illness. (3) QALY gained for sudden infant death syndrome and breast cancer. (4) Change in lifetime cost of economic productivity.	Paediatrics Outcomes (1) Gastroenteritis. (2) Respiratory disease. (3) Otitis media. (4) NEC. (5) SIDS. (6) Obesity. (7) IQ. Maternal Outcomes (1) Breast cancer Other Outcomes None.	(1) Improving breastfeeding would result in avoiding £17 million in health costs for infant illness and £31 million in incremental benefit would be gained annually among first-time mothers. (2) Improving breastfeeding would annually avoid £4.7 million in family cost and £1.3 million QALYs. (3) Future research is needed on additional maternal and infant health conditions and improved methodological and data measurement is needed in breastfeeding research.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration).
Bartick et al., 2013	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Dynamic Modelling Perspective Societal	Breastfeeding rates in 2008 compared to a defined optimal breastfeeding rate of 90% breastfed for 12 months with 40% of those continuing to breastfeed up to 18 months.	(1) Excess cases for 5 maternal conditions. (2) Direct medical costs as reported in the literature for 5 maternal conditions. (3) Indirect costs as reported in the literature for 5 maternal conditions. (4) Cost of premature death, defined as death before age 70, using value of a statistical life cost methodology.	Paediatrics Outcomes None Maternal Outcomes (1) Breast cancer. (2) Premenopausal ovarian cancer. (3) Type 2 diabetes. (4) Hypertension. (5) Myocardial infarction. (6) Premature death Other Outcomes None	(1) Suboptimal breastfeeding results in statistically significant excess cases of breast cancer (4981), hypertension (53 847) and myocardial infarction (13 946). (2) The cost of suboptimal breastfeeding is $17.4 billion in cost to society resulting from premature death, $733.7 million in direct medical costs and $126.1 million in indirect costs.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding in the published literature is often poorly defined (e.g. any, exclusive, total duration). Complete cost from all perspectives, particularly maternal time, was not able to be measured
Jegier et al., 2013	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Institutional (Hospital)	Not applicable.	(1) Cost per 100 ml of human milk from the infant’s mother for VLBW infants during the NICU stay. (2) Cost per 100 ml of preterm formula for VLBW infants during the NICU stay. (3) Cost per 100 ml of donor human milk for VLBW infants during the NICU stay.	Paediatrics Outcomes None Maternal Outcomes None Other Outcomes (1) Pump rental. (2) Pump kit. (3) Milk containers.	(1) The median cost to the hospital per 100 ml of human milk was US$0.51 for those who produced ≥700 ml per day to US$7.93 for those who produced <100 ml per day. (2) Providing a pump, kit and containers for mothers who produced at least 100 ml per day cost the hospital less than the hospital would pay to procure donor human milk (US$14.84 per 100 ml) and commercial preterm formula ($3.18 per 100 ml). (3) The cost per 100 ml of human milk was most sensitive to the cost of the containers, with cost increasing 55% when container cost was doubled and 109% when container cost was tripled.	The analysis was limited to only those items universally required by all mothers for the provision of human milk to a NICU infant.
Ma et al., 2013	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Any and exclusive breastfeeding were modelled for up to 12 months of age.	(1) Excess cases, direct costs, and indirect costs for 3 paediatrics illnesses. (2) Excess cases and VSL costs of premature death for 3 paediatric conditions.	Paediatrics Outcomes (1) Gastroenteritis. (2) Lower respiratory track illness. (3) NEC. (4) SIDS. Maternal Outcomes None Other Outcomes None.	(1) If 90% of Louisiana families met medical recommendations to breastfeed exclusively for 6 months, we would see cost savings of $216 million and prevent 18 infant deaths.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration)
Patel et al., 2013	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Institutional (Hospital)	(1) Average daily and cumulative dose of human milk in on day of life 28 during the NICU stay.	(1) Cost for each chargeable item during the infant’s hospital stay.	Paediatrics Outcomes (1) Sepsis. Maternal Outcomes None Other Outcomes None	(1) Increasing average daily dose of human milk at 28 days of life, reduced the risk of sepsis by 1.9% (P = 0.008). (2) Infants who had an average daily dose of human milk at 28 days of life of 50 ml/kg/day or more had the statistically lowest hospital costs (adjusted costs US$114 870 ± US$24 782) compared to those who received <25 ml/kg/day (adjusted costs US$146 384 ± US$38 988).	This was a single centre study and thus may not be generalizable to other centres and the selection of day 28 as the interval to measure average daily dose of human milk may not be the only or best interval to capture protection from sepsis.
Colchero et al., 2015	Mexico	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Any and exclusive breastfeeding, partial breastfeeding, and formula feeding at 3 days of life, from 0 to 6 months and from 6 to 12 months.	(1) Fixed treatment costs, and variable treatment costs for 5 paediatrics illnesses. (2) Excess cases and costs of premature death for 5 paediatric conditions and sudden infant death syndrome. (3) Cost of formula.	Paediatrics Outcomes (1) Gastroenteritis. (2) Lower respiratory track illness. (3) Upper respiratory track illness. (4) NEC. (5) Otitis media. (6) SIDS. Maternal Outcomes None Other Outcomes (1) Formula.	(1) Inadequate breastfeeding costs $746.6 million to $2.42 billion, where infant formula accounts for 11–38% of total costs (2012 dollars).	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration). Incidence and mortality data in Mexico is not reported by month of age which led to authors having to create two age categories and estimate for those categories.
Johnson et al., 2015	United States	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Institutional (Hospital)	(1) Dose, cumulative dose, and average daily dose of human milk at day of life 14 during the NICU stay. (2) Exposure to any formula by day of life 14 during the NICU stay.	(1) Cost for each chargeable item during the infant’s hospital stay.	Paediatrics Outcomes (1) NEC Maternal Outcomes None Other Outcomes None	(1) The marginal cost of NEC on the hospital stay was US$43 818 after controlling for patient factors, NEC risk and dose of human milk received during days 1–14. (2) Each ml/kg/day of human milk on days 1–14 of life decreased non-NEC related NICU costs by US$534.	This was a single centre study and thus may not be generalizable to other centres. The presence of other morbidities was not controlled for which may inflate the marginal cost impact of NEC. The institution’s feeding protocols may have led to the underestimate of the impact of human milk fortifier and formula on both NEC and cost because very few infants in this setting receive either during the first 14 days of life.
Pokhrel et al., 2015	UK	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static and dynamic modelling Perspective Payer	(a) Exclusive breastfeeding at 4 months. (b) Any breast milk at discharge from the NICU. (c) Lifetime months of any breastfeeding.g	(1) Excess cases and direct treatment costs for 4 paediatrics illnesses. (2) Excess cases and direct treatment costs for 1 maternal illness. (3) QALY gained, and it associated monetary value using willingness to pay per QALY for breast cancer.	Paediatrics Outcomes (1) Gastroenteritis. (2) Respiratory disease. (3) Otitis media. (4) NEC. Maternal Outcomes (1) Breast cancer Other Outcomes None	(1) Supporting mothers who are exclusively breastfeeding at 1 week to continue to exclusively breastfeed at 4 months would save over ₤$11 million pounds annually for 3 infectious infant illnesses. (2) If the number of first-time mothers who currently breastfeed for 7–18 months in their lifetime was doubled, over ₤31 million pounds could be saved from reductions in breast cancer cases and increases in quantity and quality of maternal life.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration) and thus may underestimate the impact of breastfeeding on disease risk.
Colaizy et al., 2016	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Dynamic Modelling Perspective Societal	Current breast milk intake as a proportion of all enteral feeds during the NICU stay and followed until 36 weeks post menstrual age compared to a simulated cohort where 90% of ELBW infants received optimal breast milk intake defined as 98% or more of enteral feeds being breast milk from the infant’s mother.	(1) Hospital direct and indirect medical costs for the NICU stay. (2) Medicare physician reimbursement for neonatology. (3) Parental non-medical expenditure for NICU infants. (4) Death cost using value of a statistical life cost methodology.	Paediatrics Outcomes (1) NEC. Maternal Outcomes None Other Outcomes None	(1) NEC incidence was 1.3% among those receiving ≥98% human milk, 8.2% among those received a mixed diet, and 11.1% among those fed only preterm formula (P = 0.002). (2) Compared to optimal human milk feeding practices, current human milk feeding among ELBW infants results in 928 excess cases of NEC and 121 deaths. (3) The cost of NEC for suboptimal compared to optimal human milk feeding for ELBW infants is an additional US$27.1 million in medical costs, US$563 655 in indirect costs, and US$1.5 billion in premature death.	The model for optimal breastfeeding represents an ideal that may not be attainable given the variety of difficulties mothers of ELBW infants face when establishing and maintaining lactation. The analysis used a single centre estimate for breastfeeding rates during the NICU stay which may not reflect the population average. The cost estimates used were significantly lower than other published studies because they were marginal estimates for the cost of NEC after controlling for other factors and morbidities that impact total NICU costs. The estimates for NEC incidence did not include all possible variables and thus may be subject to some residual confounding. Last, the analysis did not consider the impact of donor human milk nor human milk fortifier both of which may impact the risk of NEC.
Mahon et al., 2016	UK	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Current rate of 35% to 100% breastfeeding during the NICU stay and continued exclusive breastfeeding to 6 months.	(1) Cases averted, cost savings and QALY gained from 2 paediatric conditions and death that occurs during the NICU stay. (2) Cases averted, cost savings, and QALY gained for 4 paediatrics conditions that occur after the NICU stay and 2 illnesses developed later in adulthood. (2) Cases averted and costs savings for premature death for sudden infant death syndrome after the NICU stay.	Paediatrics Outcomes (1) NEC during the NICU stay. (2) Sepsis during the NICU stay. (3) Death during the NICU stay. (3) SIDS after the NICU stay. (4) Acute otitis media after the NICU stay. (5) Leukaemia after the NICU stay. (6) Neurodevelopmental impairment and disability after the NICU stay. (7) Childhood obesity and its resulting impact on Type 2 diabetes later in life. (8) Childhood obesity and its resulting impact on coronary heart disease later in life. Maternal Outcomes None Other Outcomes None	(1) The NHS would save £30.1 million and would avert 190 deaths annually if all NICU infants received 100% human milk due to reductions in cases of NEC and sepsis. (2) Total lifetime savings to the NHS if all NICU infants received 100% human milk would be £904 per infant and would gain an average of 0.2 QALY per infant. (3) Per cohort of preterm infants, there would be 238 averted deaths per year which is associated with a reduction in lifetime productivity of £153.4 million.	Year of pounds adjusted for inflation not stated.
Rollins et al., 2016	Brazil, China for medical costs; 96 countries for cognitive loss costs	Disciplinary Approach Primary: Nutritional Economics Secondary: Epidemiological and Health Economics Methodological Approach Static Modelling Perspective Societal	Varied by analysis which included exclusive breastfeeding to 6 months and continued breastfeeding for 1–2 years.	(1) Cost of lost earnings in billions and as % of GNI. (2) Costs of medical treatment for up to 8 paediatric illnesses. (3) Paediatric and maternal deaths. (4) Environmental resources used to produce formula.	Paediatrics Outcomes (1) Deaths in childhood. (2) IQ loss (96 countries). (3) Country specific health outcomes: In Brazil and China (1) Gastroenteritis (diarrhoea). (2) Pneumonia (respiratory track illness). (3) Otitis media. (4) NEC. (5) Bronchiolitis. In US and UK (1) Gastroenteritis (diarrhoea). (2) Pneumonia (respiratory track illness). (3) Otitis media. (4) NEC In USA (1) Gastroenteritis (diarrhoea). (2) Pneumonia (respiratory track illness). (3) Otitis media. (4) NEC. (5) Obesity. (6) Atopic dermatitis. (7) Asthma. (8) Leukemia. Maternal Outcomes (1) Breast Cancer Deaths. Other Outcomes (1) Environmental impact of breastfeeding in terms of resources used (water, cans, metal and paper).	(1) The annual economics losses globally associated with lower intelligence was US$302 billion, US$231.4 billion from high-income countries. (2) Improved breastfeeding practices would annual avert 823 000 deaths in children under 5 and 20 000 breast cancer deaths among women. (3) A 10% increase in exclusive breastfeeding at 6 months or continued breastfeeding for 1–2 years would yield health expenditures savings of US$312 million in the US, US$7.8 million in the UK, US$30 million in Urban China, and US$1.8 million in Brazil. (4) Improving breastfeeding from present levels to 90% meeting medical recommendations would reduce health expenditures by US$2.45 billion in the US, US$22.6 million in urban China, and US$6 million in Brazil. (5) Improving breastfeeding to 45% meeting medical recommendations would reduce health expenditures by US$29.5 million in the UK.	The analysis is limited by the use of existing literature and the limitations within each of those studies.
Walters et al., 2016	Cambodia, Indonesia, Laos, Myanmar, Timor-Leste, Viet Nam, Thailand	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Societal	Simulated exposures to breast milk including 100% receive some up to 6 months, 100% are exclusively breastfeed to 6 months with continued breastfeeding to 2 years, and 90% of women have 2 years of cumulative lactation.	(1) Cases, direct treatment costs and indirect family costs for 2 paediatrics illnesses. (2) Cases of mortality for 2 paediatric illnesses and 1 maternal illness. (3) Potential earnings lost due to lost IQ points. (4) Cost of formula. (5) Cost of national breastfeeding program.	Paediatrics Outcomes (1) Gastroenteritis. (2) Acute respiratory track illness (as proxy for pneumonia). (3) IQMaternal Outcomes. (1) Breast cancer Other Outcomes (1) Formula. (2) Breastfeeding program costs.	(1) Lost potential earnings from cognitive losses in IQ total $$1.6 billion per year across all countries. (2) Inadequate breastfeeding generates annual healthcare treatment costs for diarrhoea and acute respiratory illness of $0.3 billion. (3) Inadequate breastfeeding generates 1749 maternal deaths from breast cancer per year. (4) Implementing a national breastfeeding program in Viet Nam would avert 200 child deaths and has a 139% return on investment.	Model simulations use many assumptions and relies on precision of published estimates and many of those estimates came from high income countries. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration). Data were not available in all countries, so they extrapolated from one country to the others. Cost data was not available on a national level and this they used local cost data and government reports to estimate to a national level.
Bartick et al., 2017	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Dynamic Modelling Perspective Societal	Current rates of any and exclusive breastfeeding and a hypothetical 90% of women breastfed to medical recommendation were modelled for up to 24 months of age for each infant. Lifetime lactation was measured up to 4 years for each mother across all of her births.	(1) Excesses case and deaths for 8 paediatric illnesses, 6 maternal illnesses, and sudden infant death syndrome. (2) Direct and indirect medical costs for 8 paediatric illnesses and 5 maternal illnesses. (3) Indirect non-medical costs for 8 paediatric illnesses and 5 maternal illnesses. (4) Costs of premature death (VSL) for 3 paediatric illnesses, and 5 maternal illnesses, and sudden infant death syndrome.	Paediatrics Outcomes (1) Acute Lymphoblastic leukaemia. (2) Acute otitis media. (3) Chron’s Disease. (4) Ulcerative Colitis. (5) Gastroenteritis. (6) Lower respiratory track illness. (7) Obesity among non-Hispanic white infants. (8) NEC. (9) SIDS. Maternal Outcomes (1) Breast cancer. (2) Premenopausal ovarian cancer. (3) Type 2 diabetes. (4) Hypertension. (5) Myocardial infarction. (6) Premature death Other Outcomes None	(1) Suboptimal breastfeeding results in statistically significant excess cases and deaths for all the paediatric and maternal illnesses measured except premenopausal ovarian cancer. (2) The cost of suboptimal breastfeeding was $14.2 billion in cost to society resulting from premature death, $3 billion in direct and indirect medical costs and $1.3 billion in indirect non-medical costs. (3) 80% of the total medical cost and 50% of the cost for premature death due to suboptimal breastfeeding was attributable to maternal costs.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding in the published literature is often poorly defined (e.g. any, exclusive, total duration). Complete cost from all perspectives, particularly maternal time and the cost of breast milk replacement, was not able to be measured. The model assumed that the observed relationship between breastfeeding and risk reduction was causal though it is possible that some of the relationship may be confounded. The analysis also assumed a steady state of breastfeeding rates even though breastfeeding rates have increased year over year.
Stuebe et al., 2017	United States	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Dynamic Modelling Perspective Societal	Changes in 10% increments to exclusive breastfeeding from 0 to 6 months and any breastfeeding from 0 to 12 months.	(1) Excesses case and deaths for 8 paediatric illnesses, 6 maternal illnesses, and sudden infant death syndrome. (2) Direct and indirect medical costs for 8 paediatric illnesses and 6 maternal illnesses. (3) Indirect non-medical costs for 8 paediatric illnesses and 6 maternal illnesses. (4) Costs of premature death (VSL) for 3 paediatric illnesses, and 6 maternal illnesses, and sudden infant death syndrome.	Paediatrics Outcomes (1) Acute Lymphoblastic leukaemia. (2) Acute otitis media. (3) Chron’s Disease. (4) Ulcerative Colitis. (5) Gastroenteritis. (6) Lower respiratory track illness. (7) Obesity among non-Hispanic white infants. (8) NEC. (9) SIDS Maternal Outcomes (1) Breast cancer. (2) Premenopausal ovarian cancer. (3) Type 2 diabetes. (4) Hypertension. (5) Myocardial infarction. (6) Premature death Other Outcomes None	(1) A 5% increase in breastfeeding rates was associated with a statistically lower incidence of cases and costs for childhood infectious illness like acute otitis media. (2) The US as a whole and each US state saw some significant differences in excess cases and costs due to suboptimal breastfeeding with the degree of difference related to size difference between the current rate of breastfeeding and the proposed optimal rate.	Model simulations use many assumptions and relies on precision of published estimates. Breastfeeding in the published literature is often poorly defined (e.g. any, exclusive, total duration). Complete cost from all perspectives, particularly maternal time and the cost of breast milk replacement, was not able to be measured. In this analysis, fewer simulations were purposefully run to be more conservative to reflect the uncertainty of state level data for the model inputs, especially for premature infant breastfeeding.
Unar-Munguia et al., 2017	Mexico	Disciplinary Approach Primary: Health Economics Secondary: Epidemiological and Health Economics Methodological Approach Static Modelling Perspective Societal	Current exclusive breastfeeding rates at 6 months and continued breastfeeding between 12 and 36 months was compared to 95% of parous women breastfeeding exclusively for 6 months and continued breastfeeding between 12 and 36 months.	(1) Total direct medical treatment costs for cases treated at the public healthcare system. (2) Foregone wages for caregivers who support breast cancer patients. (3) Cost of productivity, defined as foregone wages and short and long term subsidies for women eligible for social security benefits, for the breast cancer patients for morbidity and mortality.	Paediatrics Outcomes None Maternal Outcomes (1) Breast Cancer Other Outcomes None.	(1) Suboptimal exclusive breastfeeding at 6 months results in statistically significant excess per 100 000 women of 537 cases of breast cancer, 126 premature deaths, 2629 DALYs, and US$13.97 million. (2) The total economic cost of suboptimal breastfeeding was US$245 million, 80% of which were attributable to medical costs. (3) Sensitivity analyses demonstrated that increasing the discount rate to 5% would reduce the total economic burden estimates by 50% while using per Capita GDP instead of foregone wage would increase the total economic burden estimates by double.	Conservative estimates were used at each decision point which may have resulted in an underestimate of the economic burden of breast cancer due to suboptimal breastfeeding. Further, not all costs could be measured which might also contribute to an underestimate. Third, the model used literature estimates which may not perfectly capture breast cancer incidence and mortality as well as the impact of breastfeeding on breast cancer. Last, the model had some overlap because there was not information available that would delineate the number of women who breastfeed exclusively for the first 6 months and continue breastfeeding after 1 year from those that did not exclusively breastfeed for the first 6 months but did continue breastfeeding after a year.
Siregar et al., 2018	Indonesia	Disciplinary Approach Primary: Health Economics Secondary: Epidemiological Methodological Approach Static Modelling Perspective Societal	Current breastfeeding rates was compared to breastfeeding to medical recommendation.	(1) Excess cases and costs incurred to providers for 2 paediatric health conditions. (2) Excess cases and costs incurred to patients for 2 paediatric health conditions. (3) Excess cases and costs incurred to healthcare system for 2 paediatric health conditions. (4) Change in healthcare costs per newborn.	Paediatrics Outcomes (1) GII (diarrhoea). (2) Pneumonia/respiratory disease. Maternal Outcomes None Other Outcomes None	(1) The cost to treat diarrhoea and respiratory illness was US$11.37 per case. (2) The total annual cost for not breastfeeding according to medical recommendations was US$118 million which includes US$88 million in medical costs to the healthcare system costs and US$30 million for patient costs, such as transportation and missed work, in Indonesia.	(1) The estimates for costs were limited to only five Indonesian regions which may not reflect the totality of costs experience in Indonesia. (2) They assumed that the impact of not breastfeeding was the same across all Indonesian areas. (3) They did not include patient productivity costs if the patient worked inside the home, such as a stay at home parent.
Oliveira et al., 2019	United States	Disciplinary Approach Primary: Health Economics Secondary: Epidemiological Methodological Approach Static Modelling Perspective Institutional (Governmental agency and payer).	Breastfeeding rates in 2016 compared to a hypothetical optimal breastfeeding of 90% to medical recommendation for the first year of life (exclusive for 6 months with continued breastfeeding through 12 months).	(1) Excesses case and deaths for 8 paediatric illnesses, 5 maternal illnesses, and sudden infant death syndrome. (2) Direct and indirect medical costs for 8 paediatric illnesses and 5 maternal illnesses. (3) Indirect non-medical costs for 8 paediatric illnesses and 5 maternal illnesses. (4) Costs of premature death (VSL) for 3 paediatric illnesses, and 5 maternal illnesses, and sudden infant death syndrome. (5) Cost of food packages. (6) Cost of WIC administrative services.	Paediatrics Outcomes (1) Acute Lymphoblastic leukaemia. (2) Acute otitis media. (3) Chron’s Disease. (4) Ulcerative Colitis. (5) Gastroenteritis. (6) Lower respiratory track illness. (7) Obesity among non-Hispanic white infants. (8) NEC. (9) SIDS. (10) Infant food package utilization. Maternal Outcomes (1) Breast cancer. (2) Premenopausal ovarian cancer. (3) Type 2 diabetes. (4) Hypertension. (5) Myocardial infarction. (6) Premature death. (7) Maternal food package utilization. Other Outcomes (1) Food Packages. (2) WIC administrative costs.	(1) If 90% of WIC met medical recommendations to breastfeed exclusively for 6 months with continued breastfeeding for 12 months, there would be an 8% increase in WIC participants per month and a US$252.4 million dollar increase in WIC program costs. (2) Under optimal conditions, federal Medicaid expenditure would decrease by at least US$111.6 million and WIC households (or their healthcare payer) would see savings of US$9 billion.	Conservative estimates were used at each modelling choice which may have resulted in an underestimate of the impact improving breastfeeding rates. Further, the USBC calculator which was used to estimate costs has model parameters that are based on literature estimates. These may be subject to under-estimation and confounding. The economic costs were not exhaustive. The model did not account for any potential downstream cost impact such as price elasticity for decreased formula demand and increased demand for lactation support, equipment, and supplies. The model also did not account for additional investments that might be needed to achieve optimal breastfeeding rates.
Santacruz-Salas et al., 2019	Spain	Disciplinary Approach Primary: Health Economics Secondary: None Methodological Approach Micro-costing Perspective Payer	Breastfeeding, Mixed Feeding, and Formula feeding for the first 6 months of life.	(1) Payments made for hospitalization for the infant. (2) Payments made for primary care visits for the infant. (3) Payments made for speciality care visits for the infant. (4) Payment made for pharmacy use for the infant. (5) Payments made for emergency room visits for the infant. (6) Payments made for medical tests for the infant.	Paediatrics Outcomes (1) Claims to the public health system for all medical reasons. Maternal Outcomes None Other Outcomes None	(1) Infants exclusively breastfed for 6 months had lower cost for hospital admission (P = 0.08), primary care visits (P < 0.01), speciality visits (P = 0.14), pharmacy (P = 0.02), medical tests (P = 0.63), emergency visits (P < 0.01) and total healthcare payments (P < 0.01) compared to those who were not exclusively breastfed for 6 months (mixed fed or formula fed). (2) Controlling for maternal and infant sociodemographic variables, the mean cost for healthcare expenses for infants exclusively breastfed for 6 months was €454.40-€503.50 lower compared to those who were not exclusively breastfed for 6 months (mixed fed or formula fed).	The analysis was limited to the payer perspective for infants alone. Caregiving costs and lost familial income for illnesses were not measured. They were not able to measure maternal behaviours that might lead to early breastfeeding cessation such as mastitis. They used medical records which does not capture all sources of costs, particularly maternal time costs for feeding and any privately secured care that was not indicated in the public care system.
Walters et al., 2019	Over 130 mostly low-middle income countries plus United States	Disciplinary Approach Primary: Nutritional Economics Secondary: Epidemiological and Health Economics Methodological Approach Static Modelling Perspective Societal	Current rates per country compared to medical recommendations from WHO and UNICEF.	(1) Cases of morbidity and mortality for 3 paediatric illnesses and 3 maternal illnesses. (2) Healthcare system treatment costs for 2 paediatrics illnesses and 1 maternal illness. (3) Potential earnings lost due to lost IQ points for children. (4) Potential earnings lost due to premature mortality for children and mothers. (5) Cost of formula.	Paediatrics Outcomes (1) Gastroenteritis. (2) Acute respiratory track illness (as proxy for pneumonia). (3) Obesity. (4) IQ. Maternal Outcomes (1) Breast cancer. (2) Ovarian cancer. (3) Type II diabetes. Other Outcomes (1) Formula.	(1) Not breastfeeding to medical recommendation can be attributed globally to over 175 million excess cases of paediatric illness and over 996 million excesses cases of maternal disease annually. (2) Not breastfeeding to medical recommendation can be attributed globally to 595 379 excess paediatric deaths and 98 943 excess maternal deaths annually. (3) The cost of avoidable healthcare treatment globally due to not breastfeeding is US$1146.81 million annually. (4) The cost of cognitive loses globally due to not breastfeeding is US$285.39 billion annually. (5) The cost for premature death globally due to not breastfeeding is US$53.7 billion for paediatric deaths and US$1.26 billion annually for maternal deaths.	Model simulations use many assumptions and relies on precision of published estimates and many of those estimates came from high income countries. Breastfeeding is often poorly defined in the literature (e.g. any, exclusive, total duration). Cost data similarly relied on precision of previously published estimates and government reports.
Quesada et al., 2020	Spain	Disciplinary Approach Primary: Epidemiological Secondary: Health Economics Methodological Approach Static Modelling Perspective Payer	Exclusive breastfeeding at hospital discharge and 6 months post-partum.	(1) Costs savings for 4 paediatric illnesses resulting from increased breastfeeding rates.	Paediatrics Outcomes (1) Gastroenteritis (diarrhoea). (2) Respiratory infection (bronchitis and asthma). (3) Otitis media. (4) NEC. Maternal Outcomes None Other Outcomes None.	(1) Increasing exclusive breastfeeding rates at hospital discharge and 6 months from their 2014 levels (85% and 15%, respectively) to 2020 WHA recommendations (95% and 50%, respectively) would save Spanish healthcare system €197 million/year or €464/child born from just 4 paediatric illnesses.	Analysis was limited to cases only in the first two years of life for the 4 illnesses studied. All other potential impact was not measured.
Lechosa-Muñiz et al., 2020	Spain	Disciplinary Approach Primary: Health Economics Secondary: None. Methodological Approach Micro-costing Perspective Payer	Exclusive breastfeeding, exclusive formula feeding, and mixed feeding at hospital discharge and 2, 4, 6, 9 and 12 months post-partum.	(1) Cost of hospitalization for infectious diseases at hospital discharge, 2, 4, 6, 9 and 12 months. (2) Cost of primary care office visits for infectious diseases at hospital discharge, 2, 4, 6, 9 and 12 months. (3) Cost of drug treatment for infectious diseases at hospital discharge, 2, 4, 6, 9 and 12 months. (4) Cost of emergency room visits for infectious diseases at hospital discharge, 2, 4, 6, 9 and 12 months.	Paediatrics Outcomes (1) Infectious diseases defined by All Patients Refined–Diagnosis Related Groups including non-bacterial Gastroenteritis (diarrhoea), respiratory infection (bronchitis and asthma, upper respiratory tract, RSV pneumonia, pneumonia, other respiratory signs, symptoms and minor diagnoses), kidney and urinary tract infections, Infectious and parasitic diseases including HIV, fever, viral illness, other infectious and parasitic diseases, and other skin, subcutaneous tissue and breast disorders. Maternal Outcomes None Other Outcomes None.	(1) Children who were exclusively formula fed had higher costs during the first year of life compared to exclusively breastfed children for infectious diseases for hospitalizations (791.6€ exclusive formula vs 86.9€ exclusive breastmilk), paediatrician visits (295.7€ vs 97.9€), emergency room visits(260.1€ vs 196.2€) and total costs in the first year of life (1339.5€ vs 443.5€). (2) Children who were fed a mixed diet of breast milk and formula had costs in between exclusive formula fed and exclusive breastfed infants for all measured costs.	The analysis is limited by the use of cost data from a single area in Spain and relies on existing health records. The study is observational and thus cannot confer causality.

Ethical approval was not required by our institutions as our research did not involve human subjects.

Results

We found 90 studies that potentially met the inclusion criteria (Figure 1). We reviewed and excluded 54 studies primarily because they were either macroeconomic evaluations or were studies which only measured the costs of intervention programs Of the 36 studies included, 25 were additional to those identified in Renfrew et al., as indicated in the summary table (Table 2 and supplementary Table).

Also, as can be seen in Table 2, the majority of the studies were conducted in the past 15 years, with 24 completed since 2010. Further, Renfrew et al. focused on studies from four industrialized countries which excluded the earlier economic studies that identified health costs of suboptimal breastfeeding in developing countries such as India (Gupta and Khanna, 1999) and Indonesia (Rohde, 1981). Our analysis covered a wider range of study types and reflects 20 individual countries, including 16 in the USA, 3 in the UK and in Spain, and 2 per country in Australia, Mexico, and South Korea. Five studies evaluated more than 1 country, and 2 provided global estimates. The provision of multi-country estimates has increased since 2016 including through grey literature and journal publications by researchers in international agencies and non-government organizations (Rollins et al., 2016; Walters et al., 2016; 2019; Kakietek et al., 2017).

Disciplinary and methodological approaches used in studies have expanded over the period 1995–2023, with recent studies drawing from economics, public health and health, medical, and nutrition science disciplines, using concepts and methods from epidemiology, health economics and nutritional economics. Prior to 2010, most studies used an epidemiological frame as the primary disciplinary approach (17 of 36). After 2010, additional disciplinary approaches were more commonly utilized as the primary approach including health economic and nutritional economics. The specific distribution of primary disciplinary was 17 epidemiologic, 17 health economic and 2 nutritional economics in the human capital tradition (Hanushek and Woessmann, 2008). Health economics (typically using DALY, QALY or VLS to indicate health loss or gain, disease burden or mortality impacts of not breastfeeding/breastfeeding) was used as a secondary disciplinary lens in 11 of the 36 papers.

Methodological approaches utilized both microeconomic analysis and statistical modelling, categorized here for simplicity as, ‘micro-costing’, ‘static modeling’ and ‘dynamic modeling’. Pre-2010, studies used micro-costing methods although a couple of studies from the USA and Australia used static modelling. Studies from 2010 forward used dynamic modelling and incorporated chronic disease at the country and global level. The specific distribution of costing methods used in the studies evaluated was: 12 micro-costing, 18 static modelling, 4 dynamic modelling and 2 multi-methods. The two papers that used multiple methods both used the static approach as one of their methods.

The scale and the scope of the studies have also increased since 2010 with more recent studies including maternal health costs of not breastfeeding (Bartick et al., 2013; 2017). For example, the 1997 Australian study by Drane focused on hospital treatment costs for premature infants, while the 2017 study led by Bartick for the USA comprehensively included maternal health and morbidity costs. The specific distribution of the scope of studies is: 15 studies measured child outcomes only, 7 studies measured maternal outcomes only, 13 studies measured both child and maternal outcomes and 1 study measured the institutional cost of acquiring maternal milk. The distribution of perspectives used included societal (20), institutional/payer (13) and individual (3).

Early cost of not breastfeeding studies focused on the cost impact of short-term acute illnesses, such as otitis media and gastrointestinal illness and conditions that occurred in the first year of infant life such as NEC (Riordan, 1997; Ball and Wright, 1999; Weimer, 2001; Hyun et al., 2002). However, with increasing availability of longitudinal data and electronic health records, and simulation modelling, more recent cost studies have included chronic illnesses such as childhood obesity (Büchner et al., 2008; Bartick and Reinhold, 2010; Renfrew et al., 2012; Mahon et al., 2016; Rollins et al., 2016; Bartick et al., 2017; Walters et al., 2019) and leukaemia (Büchner et al., 2008; Bartick and Reinhold, 2010; Mahon et al., 2016; Rollins et al., 2016; Bartick et al., 2017; Stuebe et al., 2017; Oliveira et al., 2019), as well as maternal breast cancer (Büchner et al., 2008; Shin, 2010; Renfrew et al., 2012; Bartick et al., 2013; 2017; Pokhrel et al., 2015; Rollins et al., 2016; Stuebe et al., 2017; Unar-Munguia et al., 2017; Oliveira et al., 2019; Walters et al., 2019), heart disease (Bartick et al., 2013; 2017; Stuebe et al., 2017; Oliveira et al., 2019) and diabetes (Bartick et al., 2013; 2017; Stuebe et al., 2017; Oliveira et al., 2019; Walters et al., 2019).

Technical economic techniques varied across studies. Specifically, 15 studies identified a discount rate, 13 studies did not require discounting and 8 studies did not explicitly discuss discounting. Twelve studies did not identify a cost year used for their analysis. For sensitivity analyses, 20 studies did not include a sensitivity analysis, 14 did include a sensitivity analysis on at least one portion of their economic model and 2 were unable to be determined.

Costs varied depending on methodology and perspective. Medical costs were dwarfed by cognitive losses and the economic costs of premature death. The cost of premature death could also vary, depending on if it were measured using the value of a statistical life, or the cost of potential lost earnings, which is much lower. Using different methods, estimates for the annual cost of not breastfeeding for the USA came to the range of $US100 billion by different estimates. Some estimates used a combination of medical, death and cognition losses, and the cost of purchasing formula or the resources to produce it (Rollins et al., 2016; Walters et al., 2019). In these, the cognition losses comprised three quarters of the total cost and mortality cost (measured by lost earnings) compromised most of remaining costs. Medical costs were only a tiny fraction of these total costs. A few studies measured the indirect cost of lost earnings for women undergoing treatment for illnesses like breast cancer. Only two studies considered the economic cost of unpaid time by women and families in caring for sick children (Bartick et al., 2017; Stuebe et al., 2017).

Discussion

This study is the first to review the breadth of the literature on the costs of not breastfeeding since the landmark report by Renfrew et al. in 2012. We have mapped the diverse studies at the intersection of breastfeeding and economics between 1996 and 2023 and widened the scope of that report to include grey literature. We addressed a gap in the current literature on the economic costs of suboptimal feeding by characterizing and summarizing the key features and findings of the literature including classification of each identified study by approaches and data sources used, study team disciplinary makeup and costing methods and measurement. We also identified and reviewed some earlier empirical studies and major reports on the costs of not breastfeeding published between 1996 and 2012 that were not included in or post-dated the Renfrew report. We found that since 2012, studies have become more diverse in geography and scope and more multidisciplinary in approach and methodology. They also have increasingly measured societal benefits like increased cognition and preventable deaths of women and children. This likely reflects the increased recognition of breastfeeding as the foundation for optimal population health, a public health and prevention priority, and a widely available, renewable commodity. It also reflects the increased availability of more complex modelling tools and software. Our analysis highlights the need for more engagement from the economics disciplines as many papers used epidemiologic methods as the primary approach and had some technical methodological weaknesses. Our analysis also highlighted a consistent disregard for the economic value of unpaid work, particularly maternal and family time. Thus, this scoping review demonstrates that while the literature has expanded and improved our understanding of the economics of breastfeeding, it needs more interdisciplinary collaboration and improved methods, particularly in the measurement of breastfeeding and maternal health impacts. This improvement will assist policymakers and others to prioritize breastfeeding as a fully funded primary strategy for population health because of its economic and health advantages.

Increase in geographic diversity in studies

Geographic diversity in studies is important because of the impact of geography on both costs and health outcomes. We found that most studies (22 of 36) were conducted in the USA, Australia or the UK. The heavily skewed geographic spread of studies perhaps reflects concerns that among the OECD [Organization for Economic Cooperation and Development (OECD), 2021a], the USA has the highest health expenditures [Organization for Economic Cooperation and Development (OECD), 2021b], worst maternal and child mortality outcomes, and has among the lowest breastfeeding rates (UNICEF, 2018). This growing awareness of the health cost implications is reflected in the recommendations of the US Surgeon General’s 2011 Call to Action on Breastfeeding (United States Department of Health and Human Services, 2011). Australia has a mixed public and private health system, with high initiation but short duration of breastfeeding; nevertheless, there is a high level of awareness as the economic benefits of optimal breastfeeding have underpinned its national dietary guidelines for infant feeding since 2003 (National Health and Medical Research Council, 2003). The UK have very low breastfeeding rates but a publicly funded health system; hence, studies for the UK focus on the potential cost savings for the National Health System of improving infant feeding practices. Future research should encompass wider ranges of countries to ensure robust and more generalizable cost estimates of the impact of not breastfeeding. These are necessary to reflect the plurality of health system and social contexts that influence health costs and their distribution between the formal and informal health sectors as well as breastfeeding initiation and duration.

Continued challenges to breastfeeding exposure methodologies

Most cost of not breastfeeding studies had a clear framework for identifying the health impact of breastfeeding/not-breastfeeding, using robustly developed epidemiological methods. However, an important weakness in that area of literature was an unclear definition of the ‘exposed’ population. To some extent this is unavoidable because early feeding practices (e.g. definition of breastfed) are poorly specified in many epidemiological studies, resulting in underestimating impacts (Smith and Harvey, 2011). This is something that Renfrew et al. (2012) also noted. The difficulty in defining breastfeeding in part arises because randomized trials are unacceptable when the health consequences of not breastfeeding are so profound. Thus, we must rely on well-controlled observational studies. A decade later this remains a concern and underscores what Renfrew et al. concluded in 2012 that ‘Research into the extent of the burden of disease associated with low breastfeeding rates is hampered by data collection methods; this can be addressed by investment in good quality research’. In fact, much of the research agenda Renfrew et al. identified remains unfulfilled in the decade since 2012 due to lack of high-quality epidemiological evidence related to measuring early feeding dose–response exposures and outcomes. This is despite the ambitions of major review studies in 2016 and 2019 (Rollins et al., 2016; Walters et al., 2019), which drew on several meta-analyses in 2015 that confirmed the exposure–outcome relationships, but highlighted the ongoing lack of investment in high-quality research in this area. Our study supports this conclusion as most studies after 2012 either identify breastfeeding measurement as a limitation to their study findings or assume that existing epidemiological estimates of the impact of breastfeeding/not-breastfeeding are causal. Addressing this challenge requires urgent and increased investment in quality epidemiological research and cross-disciplinary research collaborations between epidemiologists and health economists.

Expanded availability of cost estimates

Our analysis reveals that the coverage and scope of cost estimates has increased considerably from 2012 when Renfrew et al. noted that multiple adverse health outcomes could not be included in their economic quantitative modelling due to limitations of current evidence on prevalence and costs such as for Sudden Infant Death Syndrome (SIDS), cognition losses and early years obesity. Renfrew et al. also identified a range of other child and maternal health outcomes that were excluded, despite plausible links to early feeding (asthma, diabetes, leukaemia, coeliac disease, cardiovascular disease and sepsis in the child, and ovarian cancer and type 2 diabetes in the mother) because data were inadequate to inform an economic analysis. Some gaps have been filled, such as on maternal conditions (Bartick et al., 2013; 2017). The most recent studies measure the economic impact of not breastfeeding at global levels in both acute illnesses and chronic diseases, as well as maternal and child outcomes. This is a marked improvement in the literature though much work remains to quantify the impact of not breastfeeding on child and maternal health, particularly an emerging area in mental health.

In terms of specific costs estimated, most of the reviewed studies measured the paediatric or maternal medical treatment costs of ‘breastfed’ compared to ‘non-breastfed’ infants with a smaller number measuring cognitive losses and lower productivity of the future labour force, and costs of premature death associated with not breastfeeding. In fact, 34 out of 36 studies measured the direct medical costs of diseases associated with not breastfeeding. Early studies focused narrowly on health sector and treatment costs such as in neonatal intensive care settings while more recent empirical studies measured the broader economic consequences of not breastfeeding related to cognitive development and the cost of premature maternal mortality.

It is notable in the reviewed studies that estimates for cognition losses dwarf treatment costs. For example, for the USA, the latter amounted to upwards of $US84 billion a year in 2012 dollars, while for Australia it is around $US6 billion a year. The UK is estimated to incur cognition losses of around $US15–16 billion a year from its low rates of breastfeeding (Rollins et al., 2016; Walters et al., 2019). This approach is aligned with well-established literature on the future income losses of early life malnutrition in low- and middle-income countries (LMICs) (Lutter and Lutter, 2012). For outcomes of not breastfeeding, this is particularly relevant to high income country (HIC) settings, where breastfeeding prevalence and duration is low (Victora et al., 2016). Although not as large as paediatric cognition losses, the maternal productivity and mortality costs are also especially pertinent for HIC for the same reason. The economic impact of these factors is important because they underscore the foundational nature of breastfeeding as a building block for population health and as an important tool for maximizing maternal workforce productivity and longevity. These are all necessary to optimal long-term global economic growth and vitality.

Technical and assumption weaknesses

There were several areas of economic technical weaknesses or common assumptions that put limitations on the literature. Specifically, and with few exceptions, studies took perspectives which ignored unpaid maternal time costs associated with different types of early feeding, or the indirect costs to households (e.g. lost wages) caring for sick children associated with not breastfeeding (Jegier et al., 2010; Smith and Forrester, 2013; Smith, 2019). This choice results in a consistent undervaluing of the economic impact of breastfeeding/not breastfeeding. It also may reflect biases in existing economic methodological norms that minimize or ignore the value of caregiving and unpaid work, an area that is disproportionately provided by or expected of women. There was also a common lack of clarity and consistency about the stakeholder perspective of the study; some studies relied on charges by healthcare providers to estimate disease treatment costs, while others used payments made by insurers. In fact, most health economic studies (11 of 17) took a narrow institutional view that measured direct costs from an insurance or government payer or cost-saving perspective, while a smaller number took a broad societal perspective on the indirect or lifetime economic costs associated with not breastfeeding. Sometimes, these values were added together to arrive at final cost impact despite their lack of common perspective or source. This is important because of the cost externalities and long timeframes involved and the potential perception of inflated cost measurement.

There was also considerable variability in the estimated cost values due partly to inconsistent application of economic methods. For example, in China estimates of cognition losses from low breastfeeding rates cost the economy between $US26 billion (Rollins et al.) and $US60 billion (Walters et al., 2019) a year. From these same studies, estimates were more consistent for Australia at just over $US6 billion dollars (0.6% of gross national income) and the UK at around $US15–16 billion (around 0.7% of national income) (Rollins et al., 2016; Walters et al., 2019). The methodological summaries in the expanded Table 2 in the appendix highlight that only a small number of the micro-costing, and static or dynamic modelling studies clearly identified the year or basis for discounting of future values. Taken together, these weaknesses and diverse assumptions make it difficult to compare across studies. Such variation in methods, complexity and lack of coverage of unpaid work may also cause confusion among readers unfamiliar with epidemiology and economic methods, and points to the need for clear communication with non-specialist audiences.

Authors disciplinary background and the need for more collaboration

Lastly, it is worth noting that much of the economic work done in breastfeeding is led by non-economists (e.g. healthcare providers, epidemiologists, public health researchers). This may explain the robustness of the ability to estimate the epidemiologic counts of disease and health impacts. It also may explain the less developed econometric modelling that has been done to date. Less robust econometric modelling particularly limits the ability to extrapolate to the financial, insurance and tax policy research on obesity and chronic disease and related productivity impacts that is needed for the impacts of low breastfeeding rates to be fully recognized in current economic and health financing policy discussions. It also lends urgency to the need for cross-disciplinary collaboration. Collaboration among current breastfeeding researchers and economists would improve our knowledge and ability to account for the impact of low breastfeeding rates on health systems Addressing such economic methodological issues in the epidemiological literature may also improve the comparability of the economic work to other preventative health measures, such as reducing lead exposure in children, or assisting smoking cessation, and may encourage greater acceptance among policymakers of the findings and importance of breastfeeding as a health policy priority.

Strengths and limitations

This study is the first comprehensive review of economic studies on the value of breastfeeding or the cost of not breastfeeding, their methodologies and it crosses disciplinary boundaries. It excluded the emerging area of macroeconomic evaluation (Smith et al., 2023), cost-effectiveness studies or economic evaluations of interventions to reduce premature weaning from optimal breastfeeding, which is an important future study.

As indicated by previous reviews, the study is informed by being multidisciplinary; it is conducted by a multidisciplinary and multisectoral, cross-country collaboration of academic health economists, working with experienced lactation and health administration practitioners and clinicians. Its scope also included some articles in languages other than English if there was an English abstract. However, the heterogeneous literature meant that there was a needed to use hand-searches as much as systematic database searchers, so a weakness is that we may have missed some studies, especially those with no English abstract or translation.

The time period for the review excludes some of earlier pioneering studies in related areas conducted during the 1970s with a focus on developing countries. For example, Oshaug, Almroth, and Berg and their colleagues (Berg, 1973; Almroth et al., 1979; Oshaug and Botten, 1994) led important early contributions on the economic costs and benefits of formula vs breastfeeding, as did others focusing on specific economic or financial aspects such as McKigney (1971), Popkin (1978), Butz (1978) and Gupta and Rohde (1993).

Our review excludes a small but growing number of economic studies of the costs of policy or program interventions to increase breastfeeding (Thomson et al., 2012; Hoddinott et al., 2014; 2015; Whelan et al., 2014; Crossland et al., 2015; Moran et al., 2015; Morgan et al., 2015; Relton et al., 2018; Carroll et al., 2020; Washio et al., 2020). Some of this emerging research investigates the effects of offering women specific financial incentives for breastfeeding, though the economic or financial incentives created by structural determinants of breastfeeding behaviour are rarely considered (Smith, 2015). This is a future area for research which is important for translation of evidence on the costs of not breastfeeding into implementation of cost-effective programs to address the problem.

Conclusions

In the past decade, the low or declining rates of breastfeeding among infants and young children have emerged as an issue of economic as well as public health importance, at global and country level. This study brings new knowledge and insights from mapping and characterizing an expanding body of existing economic literature on the cost-consequences of not breastfeeding. It has built on the landmark Renfrew report by identifying new studies in a wider range of country settings, expanding the types of studies that were eligible to include grey literature, and including maternal health outcomes and a wider number of child health and development outcomes. Our review has shown a growing emphasis on cognition and mortality-related costs and their magnitude in such studies, assisted by more sophisticated methodologies, wider coverage of countries and health or development outcomes, and more extensive cross disciplinary collaborations of researchers in this area.

While earlier studies focused on preventable paediatric treatment costs, more recent literature shows that the largest economic costs arise from human capital costs—premature deaths of mothers such as from reproductive cancers, and economic losses from poorer cognition among children not breastfed in infancy. Such costs are less visible than treatment costs, and thus may be given less attention by policymakers. While the literature provides clear evidence that increased breastfeeding can mitigate rising trends in health treatment costs, these other larger cost impacts are less understood and require better communication to the public and by policymakers, as well as an enhanced epidemiological base for estimates of exposures to insufficient breastfeeding. This underscores the importance of collaboration among researchers including to explain complex economic simulation methodologies and to translate research findings to policy initiatives. Our review also highlighted the lack of attention to costing the unpaid work burdens of those caring for unwell infants or adults, usually women.

Further, we identified that methodological improvement is needed in the foundational economic techniques in these studies including the clarity and reporting of the study perspective, as well as standardization of the value of money across countries and consistent use of conventional discounting and inflation adjustment techniques. This also requires more cross-disciplinary collaboration between researchers and practitioners in epidemiology and public health, with those in health economics.

Taken together, our broad review suggests that there remain important opportunities for useful economic analyses across multiple health outcomes and country settings, and economic perspectives. Such research should prioritize:

• Interdisciplinary approaches and comprehensive conceptual frameworks, including societal as well as health system perspectives, and accounting for unpaid work.

• The framing of breastfeeding within a preventive health paradigm so that use of breastmilk substitutes is the deviation from norm and the appropriate exposure variable.

• Full consideration of human capital, with attention to mortality costs in appropriate balance with medical and financial costs.

• Impacts on maternal health are also a key priority for economic analysis.

• Enhanced economic approaches which account appropriately for price inflation and time discounting of benefits of investments.

• Consistent costing frameworks and methodologies, both across countries and time horizons, and clarifying intangible vs financial costs, as well as greater consistency in the scope and type of costs considered.

These future research directions would expand and strengthen the evidence base for greater global and country-level investments in policies and programs to better enable women to breastfeed.

Supplementary data

Supplementary data is available at Heapol Journal online.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Funding

None declared.

Acknowledgements

None declared.

Authorship note

At the time we started this study targeting this journal, a LMIC author was not a requirement, and our colleagues from LMIC countries did not feel it was appropriate to join the study at this late stage. Further, it is impractical to have an LMIC author specialized in this topic. Global research expertise on costs of not breastfeeding remains confined to a small number of authors from high-income countries. As our review illustrates, there are very few authors from LMICs who have conducted studies in this area and have only done one off studies; there is a general lack of economists interested in the ‘female’ issue of breastfeeding in all countries. Lastly, we are three women who have led research in this area, are regarded as experts in this area and are well placed to conduct the research which purposefully included global coverage. The study was not funded and thus finding another co-author to join an unfunded project without renumeration is unlikely given the pressures of academic institutions across the globe.

Author contributions

We distributed the work in this article across all of us. Specific contributions are as follows: Conception or design of the work and critical revision of the article: B.J., J.S., M.B. Article collection and abstraction: J.S. and M.B., B.J. served as tie breaker. Analysis, interpretation and drafting the article: B.J., J.S. and M.B. All authors have read and agreed to the published version of the manuscript.

Reflexivity statement

We have reflected on how our situation influences our choice of topic and how it was conducted. We are three female researchers from high-income countries none of whom was funded to do this research. Our interest and commitment to this research topic arises because as health researchers and practitioners who work to support women to breastfeed, we are concerned at the lack of visibility of breastfeeding in health economics research and the relatedly low level of global research and policy priority given to investments in enabling breastfeeding according to health authority recommendations. Our experience as clinicians, breastfeeding peer counsellors and previously breastfeeding mothers motivates us to ensure that the many barriers to breastfeeding, including its invisibility and lack of public policy investments, are addressed from an evidence base showing its economic importance.

Ethical approval.

Ethical approval for this type of study is not required by our institution.

Conflict of interest:

None declared.