Systematic review on the use of cost-benefit analysis to evaluate food environment interventions

ABSTRACT

Introduction

The global obesity epidemic is a complex issue influenced by various factors, including the food system and its environment. Increasingly, interventions targeting systemic changes in the food environment are being implemented to address obesity. Cost-benefit analysis (CBA) is gaining recognition as a valuable tool for evaluating these interventions, due to its ability to capture broader societal impacts beyond health outcomes. However, its application in this context remains poorly understood.

Methods

A systematic review was conducted following PRISMA guidelines, searching academic databases (MEDLINE, EMBASE, PsycINFO, CINAHL, Web of Science, EconLit, CRD). Study quality was assessed using the Making an Early Intervention Business Case checklist, designed to evaluate CBA studies.

Results

Of 6508 references screened, 28 studies met the inclusion criteria. The review identified common methodological approaches, summarized key findings, challenges and implications, and provided clear recommendations for improvement. The review also synthesized evidence to show that food environment interventions offer value for money with positive returns.

Conclusion

This first systematic review of CBAs in food environment interventions identified common methods, challenges, and areas for improvement. While CBA is a valuable tool for evaluating food environment interventions, more robust and standardized methodological approaches are needed to enhance its reliability and applicability.

Registration: PROSPERO (CRD42024503615)

1. Introduction

Obesity is recognized as a multifaceted public health issue caused by many interconnecting determinants, rendering it a complex problem to address [1–3]. While fundamentally rooted in a persistent energy imbalance, it is influenced by individual, social, biological and environmental factors [2,4,5]. The food system is one of the prominent factors driving obesity levels, of which the food supply chain and the food environment are major determinants [6–9]. The food environment can be defined as the conditions under which food is accessed such as the availability, affordability, convenience, promotion, and quality [10]. It also represents the setting in which individuals engage with the broader food system to obtain the foods they eventually prepare and eat [10].

Elements of the food environment, including the widespread availability of energy dense, nutrient poor foods, aggressive marketing, strategic product placement, and price promotions have shifted normative eating behaviors [6]. These shifts have contributed to increased consumption of foods high in fat, sugar, and salt and reduced the consumption of whole, minimally processed foods, thereby increasing the prevalence of overweight, obesity, and diet-related non-communicable diseases (NCDs) [8,11–13]. The industrialized food system has made unhealthy foods more accessible and affordable, but it is through the food environment that consumers encounter and make dietary choices. For instance, the rise in snacking, dining out, and reliance on convenience foods is heavily influenced by how food is marketed, priced, and placed in the retail and out-of-home environment [6].

These dietary shifts, driven by the design of the food environment, not only contribute to worsening public health outcomes but also impose significant financial strain on health systems and national economies. The global economic burden of obesity for example amounts to $2 trillion annually, equivalent to 2.8% of world Gross Domestic Product (GDP) [2]. Healthcare systems allocate 2–7% of their budgets directly to obesity prevention and treatment, while up to 20% of healthcare spending addresses obesity-related conditions such as type 2 diabetes and cardiovascular disease [3].

Recognizing the central role of the food environment in shaping dietary behaviors, there is a growing demand for preventive interventions that modify how consumers access and interact with healthier affordable diets [14,15]. There has been a corresponding rise in suggested policy interventions such as retail and advertising restrictions, price promotions, food labeling, and portion size reductions, all designed to address various facets of the food environment. However, these interventions are by nature highly heterogeneous in their design, scope, and implementation. This diversity spans sectors, contexts, and stakeholders making them inherently complex [16,17].

As the volume of these interventions increase, evaluating the economic value is important for policymaking and priority-setting, especially in the context of public sector finite and constrained budgets [14]. Such evaluations are essential to provide decision-makers with the information necessary to allocate resources efficiently [18], ensuring that funds are directed toward services or interventions that offer the highest value for money. The most common methods of economic evaluation applied to healthcare settings are either cost-utility analysis (CUA) or cost-effectiveness analysis (CEA). These analytic methods help ascertain whether the cost differences between alternative options are justified compared to their corresponding health outcomes. However, for public health interventions, it is important to consider the broader societal costs and benefits these interventions may provide beyond individual health effects as these are interventions designed to increase societal welfare [19]. This was emphasized in the statement from the United Nations Food Systems Summit (UNFSS), to fully recognize the true value of these interventions by comprehensively evaluating the economic, social, and environmental impacts, along with any associated externalities [20].

A cost-benefit analysis (CBA) is an economic evaluation that quantifies and compares the monetary value of all relevant costs and benefits associated with a policy, program, or intervention [18]. Unlike other approaches, CBA allows for a comprehensive assessment of societal welfare by capturing both market and non-market effects, including broader community and cross-sectoral impacts [18]. This makes CBA particularly well-suited to evaluating food environment interventions, which are inherently multisectoral and often produce complex ripple effects across health, economic, and social domains. It can also account for temporal variations in these impacts, acknowledging that while implementation costs might be immediate, benefits often materialize over longer timeframes [18]. This enables decision-makers to evaluate interventions holistically, considering all direct and indirect costs and benefits, while also capturing the distribution of costs and benefits across different stakeholders and sectors. Such comprehensive assessment is needed for interventions to maximize societal benefits.

1.1. Rationale for the study

A preliminary scoping search found no existing systematic review examining the use of CBA in evaluating food environment interventions, despite increased design and implementation of such interventions, and growing interest in the use of CBA for evaluation. This review is therefore warranted to explore how CBA is currently applied to inform future methodological development.

2. Aims and objectives

The review was designed to synthesize evidence relevant to the use of CBA for the evaluation of food environment interventions. It identified studies reporting the application of CBA to assess the economic viability of a range of interventions, with the goal of identifying recurring patterns in the methods used, consolidate common findings, and forge a deeper understanding of how CBA is being applied.

The objectives of this review were:

• To identify common practice for the application of CBA to food environment interventions.

• To critically evaluate the methodological challenges encountered when conducting CBA of food environment interventions.

• Highlight knowledge gaps and suggest areas for future research.

3. Methods

The protocol for this review was registered in PROSPERO (CRD42024503615) [21]. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [22]. The PICO framework [23] was employed to define the database search terms.

• Population: Stakeholders and population affected by the food environment (e.g. policymakers, urban or rural communities, schools, children, retail stores etc.)

• Intervention: Food environment interventions.

• Comparator: Any relevant comparator

• Outcome: Net present value, Benefit-cost ratio, Return on Investment

3.1. Eligibility criteria

Studies were included if they were [1]: studies applying CBA to food environment interventions [2]; full CBA’s, with at least one or more comparator and comparing costs and benefits in monetary terms [3]; interventions focused on elements of the food environment, such as food/calorie labeling, healthy food promotion, food provision, retail strategies, pricing policies, targeting consumer behaviors across various environments including communities, schools, workplaces, restaurants, retail environments [4]; reported results in terms of net present value (NPV), benefit-cost ratio (BCR) and return on investment; and [5] published in English and from the year 2000 onwards as research conducted prior to this period may be outdated [24]. Studies were excluded if they [1]: only conducted a cost-effectiveness, cost-utility, or other forms of economic analysis [2]; contained a partial economic evaluation; 3) analyzed interventions outside the scope of the food environment, such as general economic development programmes, not directly impacting food consumption; food supply chains or agricultural interventions [4] abstracts only.

3.2. Information sources

Searches were conducted in academic databases such as MEDLINE, EMBASE, APA PsycINFO, and Health Management Information Centre (HMIC) via the Ovid platform. Other databases include the Cumulative Index to Nursing and Allied Health Literature (CINAHL), Web of Science, EconLit, Center for Reviews and Dissemination (CRD)Databases (Database of Abstracts of Reviews of Effects (DARE), the National Health Service Economic Evaluation Databases (NHS EED), and the Health Technology Assessment Database (HTA), which were all conducted between 18 July 2025 and 23 July 2025.

3.3. Search strategy

The above databases were searched using a combination of four sets of keywords created by the first author (BO) and checked by the wider team. The first set of keywords focused on the methods, and this included the search terms cost-benefit analysis and economic evaluation (where CBA was conducted alongside other economic evaluation methods). The second set focused on food and its substitutes e.g. sugar, beverages, salts, fats, fruits and vegetables etc. The third set focused on elements of the food environment interventions such as product placement, price, labeling, availability, tax, subsidies, reformulation. Lastly, the fourth keyword set focused on the population of interest e.g. schools, customers, children etc. to focus our review on studies relevant to the scope. Additionally, medical Subject Headings (MeSH) were incorporated where applicable. Boolean operators (‘AND’ ‘OR’), truncation symbols (e.g. ‘*’), and wildcard characters were used to refine the search and capture variations in terminology. A layered Boolean strategy was used, initially combining terms within concept blocks using ‘OR’ then linking blocks with ‘AND’ to avoid prematurely excluding relevant studies. In Ovid, combined free text and subject heading (ti,ab,kf), and (esp) were used to capture both indexed and unindexed articles. Search strategies in CINAHL, EconLit, and Web of Science were adapted to align with each platform’s syntax requirements. No subject filters were applied to maximize transparency and reproducibility. The search was validated by checking that each concept block retrieved relevant papers individually and when combined, didn’t exclude any relevant paper. The search strategy can be found in Appendix Table 1. References were managed in Endnote to remove duplicates and facilitate citation management.

Table 1.

Study characteristics.

Study &Year of Publication	Country	Study Setting	Aim/Objective	Study Design/Methods
Impact on Urban Health, [29]	England	State-funded schools across England	To provide evidence on the costs and benefits of expanding Free School Meal provision in England	Cost-Benefit Analysis (CBA)
Nilsson et al, [30]	Sweden	National level	To analyze whether the accompanied costs outweigh the benefits or if a vice versa connexion prevails” in relation to a Pigouvian tax on meat consumption in Sweden	Cost-Benefit Analysis
Fanneh et al, 2019	The Gambia	Nationwide school system	The main objective was to assess the monetary cost and economic benefits of providing school meals and to identify the value created in terms of increased education, improved health and nutrition and value transfer to the beneficiaries for the academic year 2014/15.”	Cost-Benefit Analysis
Qureshy et al, [32]	India	National level	The primary aim was to present ‘ex-ante estimates of the benefit cost ratio of an ambitious plan to fortify extruder rice with iron, folic acid, and vitamin B12’ in India. mentioned: ‘By highlighting the range of plausible outcomes, we hope to motivate efforts to narrow these uncertainties.’	Cost-Benefit Analysis
Broeks et al, [33]	Netherlands	National level	This study estimated the effects of a tax (15% or 30%) on meat and a subsidy (10%) on fruit and vegetables consumption in the Netherlands using a social cost-benefit analysis.	Cost-Benefit Analysis
Ananthapavan et al, [52]	Australia	Supermarkets (‘seven stores (3 intervention and 4 control stores) part of the Champions IGA supermarket group, and all were located in regional Victoria, Australia’)	To assess the economic credentials of implementing a shelf tag intervention across the four largest supermarket chains in Australia	Cost-Benefit Analysis
Nicholas et al, 2018	Lao PDR	School-based	To inform evidence-based policymaking and contribute to overall advocacy efforts for improved investments in a sustainable, nationally owned school meals programme	Cost-Benefit Analysis
An et al, [35]	United States	School-based	To project the societal cost and benefit of an expansion of a water access intervention that promotes lunchtime plain water consumption by placing water dispensers in New York school cafeterias to all schools nationwide	Cost-Benefit Analysis
Burney et al, [36]	USA- Tennessee	16 Tennessee counties	A CBA of EFNEP was conducted to determine if participants’ savings in food expenditures exceeded program implementation costs	Cost-Benefit Analysis
Dollahite et al, [53]	United States	Cooperative Extension, 35 countries	To evaluate the New York State Expanded Food and Nutrition Education Program using economic methodology	Cost-Benefit Analysis
Fitzgerald et al, [54]	Ireland	Workplace	To evaluate the costs, benefits and cost-effectiveness of complex workplace dietary interventions, involving nutrition education and system-level dietary modification, from the perspective of healthcare providers and employers	Cost-Benefit Analysis
Grosse et al, [51]	United States	National level	To calculate the economic impact of fortification using both cost – benefit and cost-effectiveness analytic techniques based on pre-fortification and post-fortification epidemiological data	Ex-post Cost-Benefit Analysis (CBA) and Cost-Effectiveness Analysis (CEA)
Kuester et al, [37]	USA, Oklahoma	Seven county units in Oklahoma (Choctaw, Oklahoma, Okmulgee, Pittsburg, Pontotoc, and Tulsa)	The purpose of this study was to estimate the economic benefit of the Oklahoma Expanded Food and Nutrition Education Program	Cost-Benefit Analysis
Rajgopal et al, [38]	USA	Virginia EFNEP program	To provide an estimated cost-benefit ratio for the Expanded Food and Nutrition Education Program (EFNEP), based on potential prevention of diet-related chronic diseases and conditions.	Cost-Benefit Analysis
Salgado Hernandez et al, 2023	Mexico	Urban Mexico	To estimate the net benefits of SSB taxes compared to a scenario of no tax in urban Mexico	Cost-Benefit Analysis
Schuster et al, [40]	USA (Oregon)	Oregon State University Extension Service EFNEP program in 2 urban counties	To apply Virginia’s cost-benefit analysis (CBA) model developed for a large Expanded Food and Nutrition Education Program (EFNEP) to Oregon’s small EFNEP. To estimate a cost-benefit ratio for Oregon’s EFNEP based on retrospective analysis of program costs and optimal nutrition behaviors (ONBs) in relation to potential health-related savings for diet-related chronic diseases/conditions	Cost-Benefit Analysis
Sharieff et al, [41]	Pakistan	Urban slum of Karachi, Pakistan	To project clinical and economic effects of home-fortification in children in an urban slum of Karachi, Pakistan.	Cost-Benefit Analysis
Verguet et al, [42]	Study analyzed 14 countries: Botswana, Brazil, Cape Verde, Chile, Côte d’Ivoire, Ecuador, Ghana, India, Kenya, Mali, Mexico, Namibia, Nigeria, and South Africa	Low and middle-income countries	To estimate the costs and benefits of school feeding programs across four sectors: health and nutrition, education, social protection, and the local agricultural economy	Cost-Benefit Analysis
Zimmermann et al, [43]	Philippines	National level	To analyze the potential benefits of Golden Rice in the Philippines	Cost-Benefit Analysis
Block Joy et al, [44]	United States	17 counties” varying “from urban (Los Angeles, Orange, San Francisco) to rural (Tulare, Stanislaus, Butte)	To evaluate the economic value of EFNEP by using behavior changes among EFNEP participants	Cost-Benefit Analysis
Gray et al, [45]	Canada	National level	To examine the impacts of a mandatory reduction of trans fat content by estimating the potential health benefits and potential adverse impacts on the agri-food sector	Cost-Benefit Analysis
Sandmann et al [46]	Germany	National-level preventive health programme	To evaluate the economic benefit of vitamin D and calcium food fortification for fracture prevention in elderly German women.	Cost-benefit analysis
Zan et al [47]	USA	Ohio, low-income adult participants in EFNEP (Communitybased adult classes delivered by Ohio EFNEP; 5,593 adults recruited, 3,473 completed pre/post surveys	To evaluate the cost-benefit of food safety education within EFNEP in Ohio	Cost-benefit analysis
Boo et al [48]	Nicaragua	a communitylevel integrated ECD intervention serving 82 000 disadvantaged children … in 66 highly vulnerable rural municipalities	This paper estimates the cost – benefit ratio for an integrated early childhood development program in Nicaragua (PAININ).	Cost benefit analysis
Alderman et al [49]	Ghana	A large-scale school feeding program in Ghana	A new method to incorporate distributional benefits from poverty reduction into standard education economic evaluations	Cost benefit analysis
Gelli et al [50]	Malawi	implemented in 60 communitybased childcare centers (CBCCs) in Zomba district in southern Malawi.	To estimate the cost efficiency, costeffectiveness, benefitcost ratio, and net benefit of using communitybased early childhood development centers as platforms for an intervention	Cost-benefit analysis
Chow et al [56]	India	Rural and urban states in India that consume mustard oil in significant quantities and that have been shown to have a high prevalence of VAD	To examine the costs and health benefits of GM fortification of mustard oil in India	Cost-benefit analysis
Stein et al [55]	India	Nationally representative household food consumption survey (NSSO 2000) and expert interviews	To update its evaluation from a scientific point of view and to investigate the economic rationale for its further development and future dissemination	Cost-benefit analysis

3.4. Screening of studies

The article selection process followed the four-stage screening process developed by Roberts et al. [25], to identify relevant articles.

3.4.1. Stage 1: categorisation based on title and abstract

All duplicate entries were removed using Endnote. Title and abstract screening were independently conducted by two reviewers (BO and LF) for all retrieved citations, based on the predefined eligibility criteria. Any discrepancies between the reviewers were resolved through discussion.

The studies were categorized based on the following:

• The study presents primary research on CBA for food environment interventions, with original or primary data on costs and benefits.

• The study includes CBA on food environment interventions utilizing both primary and secondary data sources.

• The study may provide useful information for the review but does not fully align with categories A or B or lacks sufficient information in the title and abstract.

• The study reports CBA outside the scope of the food environment, such as those focusing on agricultural practices or supply chain logistics and studies not relevant to CBA or the food system.

Studies in categories B, and C were progressed to stage 2 and full texts retrieved, while those in category D were excluded from further review.

3.4.2. Stage 2: categorisation based on full text

Full-text screening was also conducted independently by both reviewers. Inter-reviewer agreement was assessed through consensus discussions at both stages. A formal kappa score was not calculated, but all discrepancies were resolved through discussion. Reasons for exclusion at the full-text stage were recorded and summarized in the PRISMA flow diagram (Figure 1). During the full-text review, studies were further classified using Burgaz et al. [9] food environment intervention classification. Although some interventions were aligned to multiple categories, each paper was assigned to one class that best represented the primary focus. Studies that did not align with any category or failed to meet the inclusion criteria were classed as ‘not relevant’ and excluded from further review. The intervention classification was as follows:

• Food Composition: Includes mandatory or voluntary reformulation strategies to improve nutritional profiles of processed foods.

• Food Labeling: Covers various informational strategies including nutrition facts panels, front-of-pack labeling, and menu labeling initiatives.

• Food Promotion: Regulates marketing of unhealthy foods through mandatory and voluntary restrictions across various media channels.

• Food Provision: Strategies implemented in various food provision settings such as preschools, schools, childcare facilities, universities, public canteens, healthcare facilities, and workplaces.

• Food Retail: Addresses product placement, availability, and marketing within retail environments including supermarkets, vending machines, food service establishments, and online platforms.

• Food Prices: Encompasses fiscal measures including taxes on unhealthy foods and sugar-sweetened beverages, as well as subsidies and vouchers for nutritious options.

Figure 1.

PRISMA flow diagram.

3.4.3. Stage 3- quality assessment

The included articles were independently assessed by both reviewers (BO and LF) using the Making an Early Intervention Business Case: checklist and recommendations for cost-benefit analysis [26]. This checklist was chosen over others such as the Drummonds and CHEERS checklist [27,28], as it is specifically tailored to CBA methodology and based on the CBA framework developed by the New Economy and Greater Manchester Combined Authority. It includes specific questions relevant to CBA studies that help to identify areas for improvement [26].

The quality assessment checklist is detailed in Appendix 2. Each included study was assessed based on 26 items listed under 10-criteria. Studies were graded as low, medium, or high quality based on their fulfillment of these domains. Studies were categorized according to the number of items fulfilled. Papers were considered low quality (≤50%) if 13 or fewer items were fulfilled, medium quality (51%-77%), if 14–20 items were fulfilled and high-quality papers ( > 77%), if 21–26 items were fulfilled.

3.4.4. Stage 4: data extraction

Data extraction was conducted independently by both BO and LF to minimize bias and enhance precision and was done using Microsoft Excel. A subset of included studies were extracted in duplicate by the two reviewers to check consistency. The remaining studies were extracted by one reviewer (BO) and verified by a second reviewer (LF). Disagreements were resolved by discussion until a consensus was reached. The data extracted included comprehensive study characteristics and relevant methodological details. Methodological descriptors such as the type of CBA (i.e. trial or model-based CBA), the type of food environment intervention, description of the intervention and comparator, and the source of data (e.g. primary data from surveys, secondary data from published sources) were extracted. Additional extraction fields included the costing method (e.g. top-down or bottom-up), resource items used, cost categorization (direct, indirect, or in-kind), and outcomes measured. The review recorded whether outcomes were monetized, the agencies or sectors that incurred costs and benefits, and the distribution of these costs or benefits across stakeholders (i.e. who gains or losses). Furthermore, information was extracted on the time horizon, discount rates, stakeholder Involvement, costs and benefits drivers, CBA results (NPV/BCR). Details on the type of sensitivity analysis conducted and key findings from the sensitivity analysis were also documented.

3.5. Data synthesis

Due to methodological heterogeneity across studies, a narrative synthesis approach was adopted. This synthesis covered general study characteristics, intervention attributes, CBA methods, and key economic findings.

4. Results

4.1. Screening and selection of studies

The search process yielded a total of 6508 records. Specifically, 2633 records were retrieved from Ovid Medline, Embase, APA PsycINFO, HMIC, 319 from CRD/NHS EED/DARE/HTA, 2806 from CINAHL/ECONLIT, and 739 from Web of Science. Additionally, 11 papers were manually identified from google and reference lists. After removing 113 duplicates, 6395 records remained for title and abstract screening. Of these, 6337 studies were classified as category D, as they were out of scope. The remaining studies included two in category A, 26 in category B, and 30in category C, resulting in a total of 58 studies selected for full-text screening. Following this screening, 28 studies were deemed eligible for inclusion in the review. For data extraction and quality assessment, 21 of the 28 included studies were reviewed independently by both authors, while the remaining 7 studies, identified from an updated search, were extracted and assessed solely by the lead author (BO). See PRISMA diagram (Figure 1)

4.2. Study characteristics

All the 28 studies conducted a CBA as either the primary evaluation or in combination with other economic evaluation methods. All detailed study characteristics are presented in Table 1. 22 studies applied only a CBA[29–50], three of these were ex-ante CBAs [30,32,43], and three were ex-post or retrospective CBAs [38,40,51]. Additionally, six studies were CBAs conducted alongside other economic evaluation methods such as CUAs and CEAs [51–56], of these, one was an ex-post CBA [51] and two were ex-ante [55,56]. Most evaluations were conducted in the United States (n = 7) [35–38,40,44,47,51,53], with single studies from Australia [52], England [29], Sweden [30], the Netherlands [33], Ireland [54], Canada [45], Germany [46], and low- and middle-income countries including India [32,55,56], The Gambia [31], The Philippines [43], Mexico [39], Lao PDR [34], Nicaragua [48], Ghana [49], Malawi [50], and Pakistan [41] and a multi-country evaluation covering 14 low- and middle-income countries (Botswana, Brazil, Cape Verde, Chile, Côte d’Ivoire, Ecuador, Ghana, India, Kenya, Mali, Mexico, Namibia, Nigeria, and South Africa) [42].

Most of the interventions (n = 9) evaluated were categorized as food provision [29,31,34,35,42,48–50,54], eight were classified as food composition [32,41,43,45,46,51,55,56], one was classified as food labeling [52], three were categorized under food prices [30,33,39] and seven were classified as food promotion interventions [36–38,40,44,47,53].

Several evaluations were conducted at the national level (n = 11) [30–33,35,42,43,45,46,51,55]. Educational settings were prominent, with many interventions being evaluated in school-based settings [29,31,34,35,42,49]. Interventions implemented in community-based settings were diverse including urban areas, regional districts and mixed urban-rural settings [36–38,44,47,48,50,56]. The remaining studies evaluated interventions conducted in supermarkets [52] and workplace settings [53,54].

4.3. Methodological characteristics of the CBAs

The methodological approaches across the 28 studies showed notable variation in design, perspective, and analytical methods, illustrated in Table 2. Most studies did not clearly state whether the CBA was model, or trial based. However, we were able to deduce this from the analysis. Twenty studies used model-based evaluations [29–35,38,39,41–47,51,52,55,56] while only eight were trial-based CBAs [36,37,40,48–50,53,54]. A societal perspective was predominantly adopted [29–35,37,39–42,44–53] though its definition and scope varied across the studies (e.g. ‘societal,’ ‘household/societal,’ ‘multi-sectoral/societal’). In some cases, the perspective had to be inferred from the inclusion of both direct and indirect costs and benefits [31,32,37,40,47–49,51]. Few studies used a healthcare perspective, with only two studies explicitly stating this as a perspective [45,54] and with four others, this was implied through their focus on healthcare costs and benefits [43,44,55,56].

Table 2.

Key model parameters applied in CBA of food environment interventions.

Study	Study Perspective	Type of CBA	Source of Data	Type of Food Systems Intervention	Time Horizon	Discount rate	Outcomes reported	Output metric	Sensitivity Analysis
Impact on Urban Health, [29]	Societal perspective	Model-based	Secondary	Food Provision	20-year period	3.5% standard rate (1.5% for health-related benefits)	Increased Lifetime earnings and contributions, cost-savings to schools, increased NHS savings, savings to families, Increased impact of gross value added on the economy	BCR	Not conducted
Nilsson et al, [30]	Societal	Model-based	Secondary	Food Prices	Not explicitly stated” or “35 years for discounting calculations	2.8%	Environmental Outcomes: Greenhouse gas emissions reduction. Health Outcomes: Reduction in colorectal cancer risk Consumer Welfare: Changes in consumption levels Producer Welfare: Changes in production levels and revenues	Net social benefits	Probabilistic sensitivity analysis and scenario analysis.
Fanneh et al, 2019	Unclear/not stated	Model-based	Primary and Secondary	Food Provision	15–30 years	Various scenarios (22%, 19.21%, 5%)	Educational outcomes (enrollment, attendance, dropout rates), health outcomes, income transfer	NPV/BCR	Scenario analysis
Qureshy et al, [32]	Unclear/not stated	Model-based	Secondary	Food Composition	7 years	3%	Productivity gains and cognitive benefits	BCR	Scenario analysis
Broeks et al, [33]	Societal perspective	Model-based	Secondary	Food Prices	30 years	3% (with sensitivity analyses using 1.5% and 4%)	Quality-adjusted life years, productivity levels, environmental impacts, consumer surplus	NPV	Scenario analysis
Ananthapavan et al, [52]	Both societal (CBA) and healthcare sector (CUA) perspectives	Model-based	Mixed	Food Labeling	Lifetime	3%	Health-adjusted life years gained, healthcare cost savings, consumer surplus, changes in body mass index, energy intake.	NPV/BCR	Multiple types including univariate and multivariate sensitivity analyses and PSA
Nicholas et al, 2018	Societal (considers benefits to individuals, households, and economy)	Model-based	Mixed	Food Provision	7 years, Lifetime	10%	Value transfer Improved education and increased productivity Healthier life Gender equality	NPV/BCR	Not reported
An et al, [35]	Societal perspective	Model-based	Secondary	Food Provision	Lifetime	3%	Cases of childhood overweight prevented lifetime direct (medical) and indirect costs saved	NPV/BCR	One-way sensitivity analysis and probabilistic sensitivity analysis using Monte Carlo simulation
Burney et al, [36]	University/Program provider perspective	Trial-based	Primary	Food promotion	12 months, 3–5 years	3%, 5%, and 7%	Changes in food expenditures, Food and nutrition intakes, food resource management practices	NPV	Scenario analyses
Dollahite et al, [53]	Societal perspective	Model-based	Mixed	Food promotion	5 years	5%	Quality-adjusted life years saved, health benefits, avoided healthcare costs, productivity losses avoided	BCR	Scenario analyses
Fitzgerald et al, [54]	Health system perspective and Employer’s perspective	Trial-based	Primary	Food Provision	1 year	No discounting	Improvement in health-related quality of life, reduction in absenteeism	Net benefit	Probabilistic sensitivity analysis
Grosse et al, [51]	Unclear/not stated	Model-based	Secondary	Food Composition	Lifetime	3%	Prevention of neural tube defects (NTDs): Spina bifida cases prevented Anencephaly cases prevented	NPV	Scenario analysis (best and worst case)
Kuester et al, [37]	Unclear/not stated	Trial-based	Mixed	Food promotion	1 year, 5 years and a lifetime	5% for base case, 10% for sensitivity analysis	Direct tangible benefits from avoiding/delaying health care costs, Indirect tangible benefits from avoiding/delaying productivity losses	NPV/BCR	Scenario analyses
Rajgopal et al, [38]	program sponsors, including federal leaders and legislators who determine funding and direction of the program	Model-based	Mixed	Food promotion	Based on life expectancy (78 years for women)	5%	Direct tangible benefits from potential prevention/delay of diet-related diseases	BCR	Scenario analyses
Salgado Hernandez et al, 2023	From the perspective of the government, producers, and consumers	Model-based	Secondary	Food Prices	Lifetime, 10 and 35 years)	4%	Economic benefits (the value of statistical life, healthcare savings, and tax revenue), reduced disease incidence and mortality	NPV	Probabilistic sensitivity analysis using Monte Carlo simulation
Schuster et al, [40]	Unclear/not stated	Trial-based	Mixed	Food promotion	1 year, 5 years and	5%	Health-related savings from prevention/delay of diet-related chronic diseases/conditions, Lost wages prevented (indirect benefits)	BCR	Scenario analyses
Sharieff et al, [41]	Societal perspective	Model-based	Mixed	Food Composition	55 years	3%	Gain in lifetime earnings, reduction in diarrhea, improvement in hemoglobin concentrations, reduced child mortality, higher IQ scores	Net benefit	Probabilistic sensitivity analysis and scenario analyses.
Verguet et al, [42]	Multi-sectoral/societal	Model-based	Secondary	Food Provision	Programme costs: 1 year, Education benefits: 45 years Health benefits: 5 years	3%	Health and nutrition gains (anemia and STH infection reduction), education gains, social protection benefits, local agricultural economy gains	BCR	Scenario analyses
Zimmermann et al, [43]	Health sector- (examining population-wide health impacts)	Model-based	Mixed- Primary (interviews with experts) and Secondary (national surveys, literature)	Food Composition	10 years	3%	Disability-adjusted life years gained, reduction in vitamin A deficiency-related health conditions	Internal Rate of return	Scenario analyses (pessimistic vs. optimistic) and parameter variation
Block Joy et al, [44]	Unclear/not stated	Model-based	Mixed-Primary and Secondary (Primary data from California EFNEP participants, secondary data for cost estimates)	Food Promotion	5 years	5%	Healthcare cost savings from delayed onset of chronic diseases, productivity gains (indirect benefits)	BCR	Scenario analyses
Gray et al, [45]	Health and healthcare-cost perspective	Model-based	Secondary (Uses FDA data adapted to Canadian context)	Food Composition	19 years	5%	CHD, health benefits in monetary terms	BCR	Scenario analyses with best estimate, low estimate, and high estimate
Sandmann et al, [46]	Societal perspective	Model-based	Secondary data (epidemiological stats and cost estimates)	Food Composition	1 year (annual cost and benefits presented), with projections to 2025 & 2050	None used (1-year horizon)	Fractures averted (36,705 annually), Secondary Benefits Measured: Healthcare cost savings from fewer fractures	NPV	Deterministic sensitivity analysis (univariate SA)
Zan et al, [47]	Unclear/not stated	Model-based	Mixed	Food Promotion	Not stated	5% for estimating effectiveness	Cost savings from foodborne illness (Per) and the annual number of life-years protected	Net benefit and BCR	Scenario analysis
Boo et al, [48]	Unclear/not stated	Trial based	Secondary data	Food Provision	Costs: 5 years of programme delivery (age 15); Benefits: 50 years of working life (age 1565)	5%	Decreased cognitive abilities. Malnutrition’s effect on school progression (e.g. the probability of ever enrolling, the age of initial enrollment, the grade completion rate per year in schooling, and the dropout rate) because of lower energy levels and shorter attention spans among malnourished children attending school.	BCR	Univariate sensitivity analysis
Alderman et al, [49]	Unclear/not stated	Trial-based	Mixed	Food provision	2 years for the cost and lifetime for the benefits	3%, 8%	Long-term productivity gains from improved human capital (learning-adjusted years of schooling).” “Distributional benefits from cash transfer element of meals	BCR	Deterministic sensitivity analysis
Gelli et al, [50]	Societal perspective	Trial based	Mixed	Food provision	12 months	3%, 5%, 10%	Benefits from avoided premature mortality, Benefits from increased lifetime productivity, Benefits from increased household agricultural production	Net benefit and BCR	Probabilistic sensitivity analysis … Monte Carlo 1 000 draws” plus “deterministic oneway sensitivity analyses
Chow et al [56]	Not stated	Model-based	Mixed	Food Composition	20-year	3%	DALYs averted and deaths averted	IRR	Not Conducted
Stein et al [55]	Not stated	Model-based	Secondary data	Food Composition	30 years	3%	DALYs averted, lives saved	NPV, BCR, IRR	Scenario Analysis

” >^{NPV: Net Present Value, BCR: Benefit-cost ratio, CBA: Cost-Benefit Analysis, CUA: Cost Utility Analysis}.

Most studies used ‘status quo’ or ‘pre-post’ as comparators, with some employing more specific control conditions [37–39,41,43–51,55,56]. The data sources varied: 12 studies relied solely on secondary data [29,30,32,33,35,39,42,45,46,48,51,55], ten used a combination of primary and secondary data [31,38,40,41,43,44,47,49,50,56]. This differed by study design as the two studies that used solely primary data were trial based CBA, while the model-based interventions were reliant on secondary sources, drawing from the literature, national databases or simulation assumptions, to monetize and extrapolate the outcomes. The time horizon also varied ranging from 1 year to a lifetime (100 years), with most studies using between 1 to 35 years [29–33,35,36,42–46,50,53–56] or a long-term horizon of longer than 50 years [35,39,41,51,52]. Some studies also used both short and long-time horizons to estimate the costs and benefits [32,34,36–40,42,48,49]. Furthermore, the time horizon was found to vary by intervention type, with no discernible pattern emerging. Discount rates of 3% (13 studies) [32,33,35,36,41–43,49–52,55,56] and 5% (eight studies) [37,38,40,44,45,47,48,53] were mostly applied. One study [57] applied a combination of rates of 22%, 19.2% and 5%, while two studies conducted no discounting due to its one-year time horizon [46,54].

Studies employed either a bottom-up or top-down costing approach, though this was rarely explicitly described, with the exception of one study by Fitzgerald et al [54]. The studies that used a bottom-up costing approach were more likely to include both direct and indirect costs compared to those using a top-down approach. Resource use was consistently detailed across all studies, providing transparency in cost and benefit calculations. However, most of the studies failed to highlight the distribution of the costs and benefits across the different agents i.e. specifying the gainers or losers due to the intervention. Only a subset of studies categorized resources into direct, indirect, and in-kind costs [30,31,35,36,38,44,46,51]. Of these, only Rajgopal et al. and Schuster et al. reported collecting only direct costs, while indirect or in-kind costs were acknowledged but not included in the analysis [38,40]. The outcomes mostly reported were health-related outcomes in terms of direct health outcomes (QALYs, DALYs, HALYs) [33,43,52–56] disease prevention (e.g. neural tube defects, chronic diseases) [29–31,33–35,37–41,44,45,51–53] and nutritional improvements (e.g. anemia reduction, vitamin A deficiency) [42,43,48]. Additional outcomes included economic outcomes such as cost-savings, lifetime earnings, productivity gains, and wage losses prevented [29,32,33,37–41,44,49,50,53,54], environmental benefits such as reduced GHG emissions [30,33] and educational benefits [29,31,34,42,47–49].

The synthesis also revealed that while sensitivity analysis was conducted in most of the studies there was considerable variation in the approach and comprehensiveness. Scenario analysis was mostly done [31–33,36–38,40,42–49,51,53,55], one study conducted only a probabilistic sensitivity analysis (PSA) [39], while others used a combination of both [30,35,41,50,52] and four studies did not conduct a sensitivity analysis at all [29,34,54,56]. Among the studies that conducted a PSA, Monte-Carlo simulation was consistently used as the analytical approach. Four studies utilized 2,000 random observations [35,39,41,52], while one study employed 3,000 observations [30] and another 1,000 observations [50]. Results were predominantly reported using 95% confidence intervals and the percentage of simulations yielding positive NPVs.

4.4. Evidence on the economic value of food environment interventions

Results were predominantly reported as BCRs [29,32,38,40,42,44–46,48,49,53] or NPVs [33,36,39,51] with some studies reporting both metrics [31,34,35,37,47,50,52,55]. Two studies used the Internal Rate of Return (IRR) [43,56], while others used terms like ‘net benefit’ or ‘net social benefits’ [30,54]. Most of the studies reported a positive return on investment, with only one study reporting a negative net social benefit [30].

For the studies classified under food provision, many focused on interventions either in a school or a workplace setting. Six studies looked at the costs and benefits of expanding provision of school meals, or different approaches to offering food and drink in a school setting, and all reported positive economic value [29,31,34,35,42,49]. One study used a CBA to understand the economic value of an education programme in a workplace setting [54] and another study on a micronutrient supplementation programme in early childhood [48,50].

For the studies categorized under food prices, the majority evaluated fiscal interventions aimed at modifying consumer behavior through taxation or subsidies. Three studies assessed the economic impact of taxes on unhealthy foods or subsidies for healthier options. Two studies assessed taxes on meat consumption and sugar-sweetened beverages, with mixed outcomes, one reporting negative net benefits, the other substantial gains [30,39]. A third study combined a meat tax with subsidies on fruits and vegetables, consistently demonstrating positive economic value [33]. Collectively, these studies indicate that while isolated taxation on high-emission or unhealthy products may not always yield favorable economic outcomes, pairing them with subsidies or targeting widely consumed unhealthy products can enhance cost-effectiveness and generate significant societal benefits.

Studies under food composition interventions focused on micronutrient fortification strategies across various delivery platforms. Several evaluated the fortification of staple foods such as rice, cereal grains and mustard, with iron, folic acid, or vitamin A, all reporting high economic returns [32,43,51,55,56]. Others assessed home fortification for young children and the use of vitamin D and calcium to prevent fractures, likewise, demonstrating favorable outcomes [41]. One study examined policies to reduce trans fatty acids through labeling and product bans, finding consistently high economic value across all scenarios [45]. Overall, these interventions show that nutrient fortification, whether voluntary or mandatory, offers potential for cost-effective health gains.

Studies classified under food promotion focused primarily on evaluations of the Expanded Food and Nutrition Education Program (EFNEP) in the United States [37,38,40,44,47,53]. Multiple analyses consistently demonstrated positive economic returns, reinforcing EFNEP’s cost-effectiveness across different contexts.

Finally, for food labeling interventions, a single study evaluated supermarket shelf tags designed to encourage healthier packaged food choices [58]. The analysis reported high economic returns, highlighting the potential of simple, information-based strategies to drive positive dietary behavior at scale.

4.5. Quality assessment

Studies were graded as low, medium, or high quality based on their fulfillment of the 28 listed items within the CBA-specific checklist (Appendix 3, Table A2). None of the studies were graded low, 21 were graded as medium quality [29–32,34–38,40–44,46–49,54–56], and seven others had a high-quality rating [33,39,45,50–53]. However, several domains were not fulfilled, these include inadequate mapping of the status quo with little or no details on the costs and outcomes and the agents incurring them [32,35,47]; limited identification of alternative interventions, with many studies evaluating existing interventions [31,34,36–38,40,42,44,53]; and lack of uncertainty analysis [29,34,54,56]. For example, a common limitation was the failing to account for what would have occurred in the absence of the intervention, and accounting for all relevant impacts [29,30,32,40,44,47,55,56]. Additionally, monetization of benefits for some studies lacked clarity on whether associated service costs were included and whether projected savings would lead to actual budget reductions [31,32,36–38,40,42,43,48,55,56]. Quality assurance processes were often missing or weak, with limited use of standard CBA guidelines or independent expert validation [30,31,35–37,46–48,55,56]. Lastly, performance monitoring and evaluation plans were absent in most of the studies, with little detail on how impacts would be tracked over time [29,30,32–36,39–43,45–50,52,55,56].

5. Discussion

5.1. Main findings

This systematic review focused on the application of CBA to evaluate food environment interventions, synthesizing evidence from 28 studies across various intervention types. The review focused only on interventions affecting the food environment and therefore prioritized interventions linked to individual food choices and nutritional outcomes. Overall, the review found that most interventions had a positive return on investment highlighting the economic value of these interventions to society, however it also identified wide heterogeneity with the methods used within the CBAs. The review revealed important insights that affect the quality and comparability of studies. The studies lacked adoption of standardized frameworks, and methodological inconsistency with respect to key aspects of the CBA, such as the perspective, discount rates, time horizons, included costs and benefits, and valuation techniques. These are briefly summarized below.

It is recommended that CBAs adopt a societal perspective to capture all relevant impacts on society [18]. This was mostly used among the studies included, however some studies used a narrower stakeholder- perspective, which may be more appropriate, particularly when the analysis is tailored to the interests of a specific stakeholder, such as healthcare systems or government agencies. The review found that most studies failed to report the distributional effects of costs and benefits across stakeholders, making it difficult to identify gainers and losers resulting from the interventions. This omission undermines a fundamental principle of CBA and its link to welfare economics [18,59]. When the NPV is positive, societal gains occur because total social benefits exceed total social costs [18]. This implies that winners’ gains surpass losers’ losses, creating potential for Pareto improvements where beneficiaries could theoretically compensate those adversely affected. In a CBA, if the intervention’s ROI is positive, it provides an opportunity to reallocate resources to improve the wellbeing of society without making anyone worse off [18,59]. Without distributional analysis, these equity considerations remain obscured, limiting the ability to assess interventions’ true social value and potential compensatory mechanisms. The review found the CBA results expressed as either NPVs, BCRs or on two occasions using IRR. Some studies presented both NPVs and BCRs, providing a more comprehensive assessment of economic value. It is important to note that NPV is recommended as the preferred output metric for estimating return on investment as it represents optimal resource allocation to society and clearly identifies which intervention delivers the largest benefit [18,59]. BCRs, while commonly reported across the studies in this review, has several limitations as it provides only a relative measure of profitability rather than an absolute measure, making it difficult to assess the true economic impact [18]. Additionally, a BCR does not account for the magnitude and scale of total social benefits, meaning interventions with a high BCR may still have a low NPV if the overall benefits are small, potentially leading to incorrect rankings of policy options [18,59]. Of the studies reporting only BCRs, these offer an incomplete picture of economic value, particularly when comparing interventions with different scales. This limitation is especially problematic when policymakers need to prioritize between multiple competing interventions with varying implementation costs and potential impacts.

The review identified a variety of approaches to measuring the costs and effects related to the interventions. Some studies used evidence from existing policies, drawing on data from ongoing or past interventions, while others used comparative modeling techniques, projecting outcomes based on policies with shared mechanisms or target populations. Statistical analyses were also used in a subset of studies to estimate impacts more generally. In the cases where empirical evidence was limited, stakeholder consultation was used to inform the expected costs and benefits [18]. This diversity in approach highlights the need for transparent reporting of assumptions and data sources. It also raises the concern of over-reliance on effect sizes from dissimilar contexts which can reduce external validity, potentially misleading policymakers. To reduce bias and improve reliability, we recommend that conservative assumptions are used, and all assumptions are validated through stakeholder engagement.

Food environment interventions are inherently complex, and therefore economic evaluations often rely on assumptions about inputs and outcomes that cannot be predicted with certainty. While most of the studies used scenario analysis to explore uncertainty, fewer used a PSA. Applying PSA in CBA offers distinct advantages over other approaches, as it captures uncertainty across all model inputs simultaneously and accounts for interactions between parameters, providing a more comprehensive assessment of uncertainty [60,61].

5.2. Strengths and limitations

This review has notable strengths; it represents the first systematic evaluation of CBA applications in food environment interventions providing an understanding of how CBA is being used and identifying methodological gaps. It used a checklist developed by the New Economy and Greater Manchester Group specifically designed for assessing the quality of CBAs, enabling a rigorous assessment of methodological quality across the studies. The synthesis of evidence across the diverse food environment interventions enabled the identification of common challenges and practices that have previously not been systematically documented.

However, several limitations should be acknowledged. The inclusion of studies with diverse methodological approaches and reporting styles introduced heterogeneity, which made direct comparisons challenging. Furthermore, while efforts were made to include all relevant studies, gray literature was not included, this may have excluded other relevant studies. Also, the exclusion of non-English language publications may have resulted in the omission of key research conducted in other regions.

5.3. Recommendations for future research

The findings of this review suggest several areas for improving the use of CBA in evaluating food environment interventions. There is an urgent need for strict adherence to established methodological frameworks specifically designed for CBAs. Such frameworks will provide clear guidance on analyzing costs and benefits while accommodating the unique complexities of these interventions. Additionally, transparency in reporting needs significant improvement, particularly when describing benefit monetization, the distribution of costs and benefits among agents, and the methods used for uncertainty analysis. Studies should clearly document their methodological choices and assumptions, enabling better understanding and replication of analyses. Stakeholder engagement also requires a more systematic and documented approach. Future CBAs should incorporate stakeholder perspectives throughout the evaluation process, from design to interpretation of results. An improved and consistent method to addressing uncertainty and assumptions in CBA is needed, as seen in CEAs and CUAs. While most studies conducted some form of sensitivity analysis, the approach and presentation of results varied widely and often lacked comprehensiveness. Future studies should adopt more rigorous and consistent approaches to testing assumptions and presenting uncertainty results. Lastly, establishing ongoing monitoring and evaluation systems for CBA processes to track impact and progress made over time would enable continuous methodological refinement and improved accuracy. This would help develop better practices and enhance the reliability of CBAs in food environment evaluation.

6. Conclusion

This systematic review of CBAs in food environment interventions revealed both the current state of practice and important methodological challenges in the field. While CBAs are increasingly being used to evaluate these interventions, significant variations exist in the methodological approaches, highlighting the need for greater standardization and methodological improvement. The review suggests that while CBA remains a valuable tool for evaluating food environment interventions, its application could be strengthened. Addressing these challenges would enhance the reliability and usefulness of CBAs in informing food environment policy and investment decisions.

Funding Statement

This paper was funded by the Economic and Social Research Council [ES/X009343/1]. The views expressed in the manuscript are those of the authors and the funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Article highlights

• First systematic review examining the application of cost-benefit analysis (CBA) for evaluating food environment interventions

• Synthesis of 28 studies revealed significant methodological variations in model parameters used.

• Common challenges included estimating intervention effects, monetising outcomes, limited stakeholder engagement, addressing uncertainties and incremental analysis.

• The findings highlight the need for greater standardisation in CBA methodologies and guidelines for conducting economic evaluation of food environment interventions to improve comparability, reliability, and policy relevance.

• Recommendations include adopting standardised frameworks, improving stakeholder engagement, conducting sensitivity analysis and strengthening performance monitoring and evaluation in CBAs.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Author contributions

B. Osifowora contributed to the conceptualization, methodology, investigation, data curation, formal analysis, and writing – original draft. L. Fu contributed to the conceptualization, methodology, investigation, manuscript review, data curation, and writing – review and editing, final approval for publication. R. Oppong and E. Frew provided conceptualization, investigation, supervision, funding acquisition, and writing – review and editing, final approval for publication.

Data availability statement

No new data were generated or analyzed in this study; therefore, data sharing is not applicable.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14737167.2025.2559112