Health, poverty and poor diet quality are inextricably linked. Poor diet quality is linked to many adverse health outcomes, both for children: obesity(Reference Okubo, Crozier and Harvey1–Reference Perry, Keane and Layte3), gastrointestinal issues and constipation(Reference Rajindrajith, Devanarayana and Benninga4), dental caries(Reference Moynihan and Petersen5,6) , hypertension(Reference Dalwood, Marshall and Burrows7), diabetes(Reference Dalwood, Marshall and Burrows7) and growth stunting(Reference Miller, Webb and Cudhea8) and for pregnant women: gestational diabetes(Reference Gao, Zheng and Jiang9), gestational hypertension(Reference Gresham, Collins and Mishra10) and excessive weight gain(Reference Liu, Zhu and Ferrara11). Looking at the impacts from a societal perspective, poor health can result in time away from school or work, increased healthcare costs and losses to the economy. Food insecurity has been associated with increased healthcare costs(Reference Berkowitz, Seligman and Meigs12) and poor health outcomes(Reference Friel and Ford13). Poverty has been linked with childhood obesity(Reference Mears, Brindley and Baxter14), with children from the most deprived decile being twice as likely to be obese as children from the least deprived decile(15).
Maintaining a good quality diet is particularly challenging for those on low incomes. Children from deprived backgrounds are more likely to have poorer diet quality than children from more affluent backgrounds(Reference Truesdale, Matheson and JaKa16,Reference Darmon and Drewnowski17) . In the UK, healthy diets are comparatively more expensive, with F&V costing significantly more per 1000 kcal energy provided (£11·79) than foods and drinks high in fat and sugar (£5·82/1000 kcal)(18). This makes it increasingly difficult for families under financial strain to maintain healthy diets.
F&V vouchers aim to improve dietary quality by safeguarding or increasing spend on F&V in low-income families. They are intended to ensure that families can access F&V that may be out of reach otherwise. Critics may argue that F&V vouchers could be used to offset current spending and could paradoxically decrease diet quality by freeing up money to be spent on unhealthy foods(Reference McKendrick19,Reference Parnham, Millett and Chang20) . Evidence to support interventions such as F&V vouchers can be challenging to gather, and we are not aware of any previous mixed methods reviews that have considered the impact of F&V voucher interventions on the diet and health of pregnant women and families with young children. This review aimed to systematically synthesise published studies (peer reviewed and grey literature) to assess the impact of F&V vouchers on the diets and health of recipients (pregnant women and families with children under the age of 5). The review also aimed to explore recipients’ experiences of F&V vouchers, where F&V voucher schemes face challenges, and what might be done to mitigate these issues.
Methods
The PICO framework for the review was as follows:
-
• Population: Low-income pregnant women and families with children under the age of 5, in Organization for Economic Cooperation and Development countries(21), used as a proxy for high-income countries
-
• Intervention: Means-tested voucher schemes that support healthy diets by at least partly targeting F&V intake.
-
• Comparator: No voucher scheme, food-based voucher schemes not targeting F&V consumption or non-food-based voucher schemes
-
• Outcomes:
-
• Primary outcomes: F&V intake and diet quality.
-
• Secondary outcomes: F&V purchasing: quantity or expenditure, nutritional value of food shopping, nutritional biomarkers, recipients’ experiences of the scheme and of food shopping, cooking and providing food for themselves or their family, healthcare providers experiences of the scheme, childhood or maternal weight status, breast-feeding rates, maternal diabetes, low or high birthweight, childhood healthcare contacts or healthcare utilisation, parental mental health, expenditure on food and food insecurity.
The protocol for this systematic review was registered on Prospero (PROSPERO 2022 CRD42022364740) on 09/11/2022(Reference Grove, Ziauddeen and Smith22).
Searches
A search was conducted on six electronic databases: EMBASE (via Ovid), MEDLINE (via Ovid), The Cochrane library, Web of Science, CINAHL (via EBSCO) and IBSS. Searches were restricted to English language articles published from the year 2000 to 30/04/2024. Grey literature searches consisted of grey literature database searches, Google searches and targeted review of specific websites (charitable organisations, think tanks and government bodies). Searches took place on 01/11/2022–03/11/2022 and were updated on 30/04/2024. The full search terms used for Medline was:
[‘healthy start’.mp. or ‘best start’.mp. or WIC.mp. or ‘Farmers Market Nutrition Program’.mp. or ‘women, infants, and children’.mp. or (‘food subsid*’ or ‘food aid’).mp. or voucher*.mp. or coupon*.mp. or (Food Assistance/ or ‘food assistance’.mp.) or ‘fruit* and vegetable* prescription*’.mp. or ‘food buck*’.mp.] AND
[family.mp. or exp Family/ or families.mp. or (‘pre-school’ or preschool).mp. or (exp Infant/ or infant*.mp.) or (Child/ or child*.mp.) or pregnan*.mp. or Pregnant Women/ or parent*.mp. or exp Parents/] AND
[(low adj2 income*).mp. or (exp Poverty/ or poverty.mp.) or exp Socioeconomic Factors/ or depriv*.mp. or disadvantage*.mp. or underprivilege*.mp.]
Inclusion and exclusion criteria
Interventional, cohort, cross-sectional, case–control, qualitative and mixed methods studies were all included, as well as grey literature with original data from charitable bodies, governmental agencies or think tanks. Conference data, letters and other grey literature were excluded.
One of the most well-known means tested voucher schemes for women and children is the American Special Supplemental Nutrition Program for Women, Infants, and Children (WIC). This US-federal food assistance scheme provides credit to be used to purchase a wide range of food items, designed to improve health by supporting recipients (at-risk and income eligible pregnant women and children under 5 years) to eat a nutritionally balanced diet(Reference Rasmussen, Latulippe and Yaktine23). WIC also includes an educational element. The WIC programme itself does not meet the inclusion criteria for this review, due the large range of foods provided other than F&V and the educational element, However, any ‘add on’ programmes offering additional funds to be spent on F&V to WIC recipients, with no compulsory educational elements do meet inclusion criteria.
Screening process and data extraction
A sample of titles (10 %) was reviewed independently for inclusion by GG and NZ. Good agreement was achieved (> 80 %) and GG then independently screened the remainder of the titles. A sample of abstracts for the selected titles (10 %) was then reviewed independently by GG and NZ, following the same process as that employed for the titles as good agreement was again achieved. Finally, GG and NZ independently reviewed all full-text articles selected by abstract screening.
Two authors (GG and NZ) independently undertook data extraction on a 20 % sample and discussed any discrepancies. No significant discrepancies were found, and GG then completed data extraction on the remaining papers. Data were extracted using a standardised form and included funding information, study location, study design, inclusion/exclusion criteria, recruitment method, population studies, sample size, demographic information of participants (age, sex, ethnicity and socio-demographic information), intervention (including type, duration and cost), outcomes (including timepoints measured), analysis methods, loss to follow-up and main findings. Any relevant outcome data were collected and recorded as presented in the original paper.
Risk of bias assessment
Risk of bias assessment was undertaken by GG and NZ, who reviewed papers independently and then discussed quality to come to an agreed final assessment. Appraisal tools were selected depending on study type, non-randomised interventional and cohort studies were assessed using Cochrane ROBINS-I tool(24), cross-sectional studies using the National Heart, Lung and Blood Institute Quality Assessment for Observational Cohort and Cross-sectional Studies tool(25) and case control and qualitative studies using Critical Appraisal Skills Programmes checklists(26). Grading of Recommendation, Assessment, Development and Evaluation(Reference Schünemann, Brożek and Guyatt27) was used to assess certainty of evidence(Reference Guyatt, Oxman and Vist28,Reference Murad, Mustafa and Schünemann29) across all quantitative outcomes included in this review, where more than one paper contributed findings.
Data synthesis
As expected, there was considerable heterogeneity in the papers selected for inclusion in the review as mixed methods, quantitative and qualitative studies all met inclusion criteria. As such, it was not possible to undertake meta-analysis or other synthesis methods and a formal narrative approach to data synthesis was utilised, following Popay et al.’s Guidance on the Conduct of Narrative Synthesis in Systematic Reviews (2006)(Reference Popay, Roberts and Sowden30). Following the research protocol, studies were grouped by study type (quantitative, qualitative and mixed methods) and then by intervention type (for example, Healthy Start (HS) in the UK).
PRISMA guidelines were followed to ensure transparent reporting of data synthesis(Reference Page, McKenzie and Bossuyt31). Vote counting and concept mapping were used to synthesise the data and to explore relationships between data.
Outcome measurement
The primary outcome of interest in this review is F&V intake and diet quality. The outcomes and definitions used in the original paper have been used in this review. Measures of diet quality may include portions of F&V consumed, deduced from diet recall, food diaries and FFQ, as well as dietary and nutrient intakes calculated from food diaries and FFQ, or other assessments of diet quality.
Results
Screening was documented using a PRISMA flow diagram (Figure 1)(Reference Page, McKenzie and Bossuyt31). Searches identified 7344 records from databases and registers, including 2900 duplicate records. 4444 titles and 753 abstracts were screened for inclusion. Full-text reports (n 77) were then assessed for eligibility, with sixteen studies (n 18 reports) included(Reference Parnham, Millett and Chang20,Reference Herman, Harrison and Jenks32–Reference Jacobs, Holcomb and Margellos-Anast48) . Of these, nine studies (n 10 reports)(Reference Parnham, Millett and Chang20,Reference Ford, Mouratidou and Wademan39–Reference Thomas, Moore and Onuselogu47) were in the UK and seven studies (n 8 papers)(Reference Herman, Harrison and Jenks32–Reference Kropf, Holben and Holcomb38,Reference Jacobs, Holcomb and Margellos-Anast48) in the USA. Four studies were non-randomised trials(Reference Herman, Harrison and Jenks32–Reference Ridberg, Levi and Marpadga35), one was a cohort(Reference Wang, Seligman and Levi36), three were cross-sectional(Reference Parnham, Millett and Chang20,Reference Blumberg, Fowler and Bai37,Reference Kropf, Holben and Holcomb38) , four were before and after(Reference Ford, Mouratidou and Wademan39–Reference Mouratidou, Ford and Wademan41,Reference Thomas, Moore and Onuselogu47) , five were qualitative(Reference Lucas, Jessiman and Cameron42–Reference Ohly, Crossl and Dykes45,Reference Jacobs, Holcomb and Margellos-Anast48) and one was mixed methods(Reference Dundas, Boroujerdi and Browne46).
PRISMA 2020 flow diagram.
In total 103 grey literature records were identified, thirty-seven of which were duplicates, and thus sixty-six records were assessed for eligibility. Eight grey literature reports were included in this review(49–55).
Schemes included
Within this review, a range of programmes have been explored, including HS, Rose vouchers, Sainsbury’s top up vouchers and Best Start foods (BSF) schemes in the UK and add-on programmes linked to WIC (including the Farmers’ Market Nutrition Programme (FMNP) and EatSF) in the USA. Detailed information on each of the schemes is provided in Table 1. HS replaced the previous Welfare foods Scheme in 2006(Reference Ford, Mouratidou and Wademan39), first with a paper-based scheme and more recently a digital scheme(56). The paper-based scheme provided vouchers that could be used to purchase fruits, vegetables, cow’s milk or formula(Reference Crawley and Dodds57). The digital HS scheme(56) has increased value (£4·25 per week) and includes a wider variety of foods that can be purchased. In Scotland, HS has been replaced with BSF(58). Welfare foods Scheme and HS/BSF are government-funded schemes. BSF is similar to HS, with some differences in eligibility criteria, foods permitted and a higher voucher value. Sainsbury’s is a UK supermarket chain that provided additional vouchers to HS redeemers as part of its food donation programme, which could be redeemed against fresh or frozen F&V only for 6·5 months in 2021(59). Finally, the Rose voucher scheme(60) is run by a charity and provides vouchers that can be exchanged for fresh fruit and vegetables but only operates in some parts of the UK. In the USA, the FMNP provides recipients of the US-federal WIC food assistance program with additional vouchers that can be used to purchase F&V from farmers’ markets and roadside stalls. Similar to WIC, exact benefits vary from state to state, and FMNP currently operates in forty-nine states(61). The food costs and 70 % of administrative costs of FMNP are supported through federal funding to state agencies. EatSF was funded through the Department of Public Health, City and County of San Fracisco and other supporters and provides pregnant WIC recipients in San Francisco with additional vouchers to spend on F&V(62).
Fruit and vegetable voucher schemes included in this review
Peer reviewed papers
Table 2 summarises the characteristics of the included papers including strengths and limitations.
Summary of included peer reviewed papers
Scheme details: please see Table 1 for details of the schemes included.
*, ‡Indicates associated papers, same study population.
†Indicates associated papers, different study populations.
§ Assessed using the ROBINS -I tool, which results in a rating of low risk of bias, moderate risk of bias, serious risk of bias, critical risk of bias or no information.
|| Assessed using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, which results in a rating of good, fair or poor.
¶ Assessed using the CASP checklist for qualitative studies, which does not give specific scoring.
Quantitative papers
F&V intake, diet quality, food purchasing, portions of fruit and vegetables and nutrient intake
Eleven papers considered the impact of F&V vouchers on either food consumption (most commonly F&V intake)(Reference Herman, Harrison and Afifi33–Reference Ridberg, Levi and Marpadga35,Reference Blumberg, Fowler and Bai37–Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) , nutrient intake(Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) and/ or food purchasing(Reference Parnham, Millett and Chang20,Reference Herman, Harrison and Jenks32,Reference Griffith, von Hinke and Smith40,Reference Thomas, Moore and Onuselogu47) (Table 3). Seven studies examined the impact of F&V vouchers on food consumption and nutrient intake. Four studies found F&V vouchers were associated with an increased intake of F&V combined(Reference Herman, Harrison and Afifi33,Reference Ridberg, Marpadga and Akers34,Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) , and two with increased intake of vegetables alone(Reference Blumberg, Fowler and Bai37,Reference Kropf, Holben and Holcomb38) . One study found no differences in F&V intake between the intervention group receiving $40 voucher for F&V and the control group(Reference Ridberg, Levi and Marpadga35).
Studies considering the impact of fruit and vegetable vouchers on fruit and vegetable intake, diet quality, nutrient intake and food purchasing
*, ‡ indicates associated papers, same study population. † indicates associated papers, different study populations.
Healthy Start
Two studies found that HS was associated with increased F&V intake and increased macro and micro nutrient intakes(Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) . When considering the impact of F&V vouchers on food purchasing, Griffith et al. used purchasing data from a panel of households in the UK to examine differences between households that were likely and not likely to be eligible for HS vouchers(Reference Griffith, von Hinke and Smith40). They found that £1 of HS vouchers resulted in £0·14 increase in spending on F&V(Reference Griffith, von Hinke and Smith40). Levels of fibre, vitamin A, Fe and carbohydrate in food purchases made in eligible households increased, whilst fat and sugar did not(Reference Griffith, von Hinke and Smith40). In contrast, Parnham et al. found no significant differences in F&V purchasing when comparing HS participants with eligible non-participants(Reference Parnham, Millett and Chang20). Thomas et al. explored Sainsbury’s top up vouchers distributed to HS recipients using loyalty card data(Reference Thomas, Moore and Onuselogu47). They found that recipients spent more on F&V and bought more portions of F&V in the intervention period compared with the control period. They also found that those who redeemed the vouchers purchased more F&V than those who did not(Reference Thomas, Moore and Onuselogu47).
Farmers Market Nutrition Program and EatSF
Herman et al. compared supermarket and farmers’ market vouchers for F&V and found that there was increased F&V servings at both intervention sites (primarily driven by vegetable consumption)(Reference Herman, Harrison and Afifi33) compared with the control group. The differences in F&V consumption at the intervention sites (both the farmers market’ and supermarket sites) compared with the control group (who received diaper vouchers) remained statistically significant 6 months after the end of the intervention(Reference Herman, Harrison and Afifi33). F&V consumption by individual participants over time was also significantly increased in both the farmers’ market and supermarket groups, but this only remained significant in the farmers’ market group at follow-up(Reference Herman, Harrison and Afifi33). Two further studies examining the FMNP found the programme to be associated with increased vegetable intake, but not with increased fruit intake(Reference Blumberg, Fowler and Bai37,Reference Kropf, Holben and Holcomb38) . Two studies by the same group exploring EatSF found contrasting results, with the pilot study (recruitment from February to August 2017, 700 participants) reporting increased F&V consumption frequency(Reference Ridberg, Marpadga and Akers34), but a later study (recruitment from September 2020 to June 2021, 770 participants) finding no significant differences(Reference Ridberg, Levi and Marpadga35).
In the USA, Herman et al. also looked at food purchasing and found that the items most frequently bought with F&V vouchers were oranges, apples, bananas, peaches, grapes, tomatoes, carrots, lettuce, broccoli and potatoes(Reference Herman, Harrison and Jenks32).
Food security
Three papers considered food security as an outcome(Reference Ridberg, Marpadga and Akers34,Reference Ridberg, Levi and Marpadga35,Reference Kropf, Holben and Holcomb38) (Table 4), all in the USA exploring EatSF(Reference Ridberg, Marpadga and Akers34,Reference Ridberg, Levi and Marpadga35) and FMNP(Reference Kropf, Holben and Holcomb38). Ridberg et al. explored EatSF in San Francisco, where pregnant women were automatically enrolled in EatSF whilst attending a pregnancy WIC appointment. They formed a control group of non-pregnant women who were receiving standard WIC benefits. Ridberg et al. report that significantly more women in the intervention group were food insecure at baseline (53 % v. 38 %), and amongst those who were food insecure, more women in the intervention group became food secure at 3 month follow-up than in the control group (23 % v. 14 %, P = 0·04, unadjusted estimate)(Reference Ridberg, Marpadga and Akers34). In their later work, Ridberg et al. found no significant difference in food security status between pregnant women receiving WIC and EatSF, and pregnant women in neighbouring counties receiving WIC alone(Reference Ridberg, Levi and Marpadga35). Kroft et al. sent postal surveys to female head of household registered for WIC in Athens county, Ohio, USA, where the FMNP was available to all WIC recipients(Reference Kropf, Holben and Holcomb38). They found no significant differences in food security (using unadjusted estimates) between those receiving WIC alone and those receiving WIC and FMNP(Reference Kropf, Holben and Holcomb38).
Studies considering the impact of fruit and vegetable vouchers on food security, breast-feeding, low birthweight and maternal outcomes
* indicates associated papers, same study population.
Health outcomes
Two papers explored differences in health outcomes (Table 4)(Reference Wang, Seligman and Levi36,Reference Dundas, Boroujerdi and Browne46) .
Healthy start
Dundas et al. used secondary data analysis of existing data sets (Growing up in Scotland (GUS) and Infant Feeding Survey (IFS) to explore breast-feeding, low birth weight, child weight and maternal mental health, with conflicting findings(Reference Dundas, Boroujerdi and Browne46). When comparing those who were eligible and receiving HS (R) and those who were eligible but not receiving HS (E), group R were significantly less likely to breastfeed than those in group E in IFS data, but no significant differences were found in GUS data(Reference Dundas, Boroujerdi and Browne46). When comparing group R with those nearly eligible (NE) in GUS, maternal mental health was significantly better in group NE(Reference Dundas, Boroujerdi and Browne46). When comparing groups R and NE in IFS, group NE had fewer low birthweight infants (group NE = 0·052 % v. group R = 0·071 %, P = 0·025)(Reference Dundas, Boroujerdi and Browne46).
EatSF
Wang et al. utilised birth records to assess for associations between the intervention (EatSF) and health outcomes (low birth weight, maternal gestational diabetes, maternal weight gain)(Reference Wang, Seligman and Levi36). They found no significant differences between the intervention and control groups, although it is likely that only a small proportion of the intervention group (∼11 %) received the intervention, due to low programme enrolment in the intervention county(Reference Wang, Seligman and Levi36).
Qualitative papers
Five qualitative studies explored HS in the UK(Reference Lucas, Jessiman and Cameron42–Reference Dundas, Boroujerdi and Browne46), and two explored the FMNP in the USA(Reference Blumberg, Fowler and Bai37,Reference Jacobs, Holcomb and Margellos-Anast48) (Table 5).
Studies considering the impact of fruit and vegetable vouchers: qualitative findings
Healthy Start
Three studies held interviews with parents(Reference Lucas, Jessiman and Cameron42,Reference Ohly, Crossl and Dykes45,Reference Dundas, Boroujerdi and Browne46) , and two undertook qualitative work with both parents and professionals(Reference McFadden, Green and Williams43,Reference Moonan, Maudsley and Hanratty44) . Common themes throughout the studies included the way in which vouchers are used: as a financial benefit to subsidise food already being purchased(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43,Reference Ohly, Crossl and Dykes45) , to increase the quantity or variety of F&V purchased(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43,Reference Ohly, Crossl and Dykes45) , or as a safety net, to be used to ensure that the family had something to eat(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43) . All five papers found that participants felt the monetary value of the HS vouchers was insufficient(Reference Lucas, Jessiman and Cameron42–Reference Ohly, Crossl and Dykes45). Issues with the application process and eligibility criteria were highlighted(Reference Lucas, Jessiman and Cameron42–Reference Moonan, Maudsley and Hanratty44,Reference Dundas, Boroujerdi and Browne46) , as well as with awareness of the scheme(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43,Reference Dundas, Boroujerdi and Browne46) .
Farmers Market Nutrition Program (FMNP)
Jacobs et al. explored participant’s and staff’s experiences of the FMNP using semi structured interviews(Reference Jacobs, Holcomb and Margellos-Anast48). Blumberg et al.’s, mainly quantitative, study included open survey questions on barriers and enablers of voucher redemption, which were qualitatively analysed(Reference Blumberg, Fowler and Bai37).
The heterogeneous nature of the interventions explored, the study designs used and the outcomes explored make robust synthesis of the data challenging. Much of the reported data is observational or non-randomised, which limits conclusions that can be drawn and confidence in any quantitative results. This, in part, reflects the challenges of evaluating public health interventions(Reference Bonell, Hargreaves and Cousens63–Reference Weatherly, Drummond and Claxton65).
Table 6 presents the GRADE assessment and summary of quantitative findings. Overall, certainty in the evidence was low. There was more consistence of results from qualitative work, with reasonable triangulation of concepts between studies, and, in general, better study quality.
Summary of findings including GRADE assessment
Table adapted from Schünemann H, Brożek J, Guyatt G, et al. GRADE Handbook: Handbook for grading the quality of evidence and the strength of recommendations using the GRADE approach.: Cochrane Training, 2013. https://gdt.gradepro.org/app/handbook/handbook.html, and Murad MH, Mustafa RA, Schünemann HJ, et al. Rating the certainty in evidence in the absence of a single estimate of effect. Evid Based Med 2017;22(3):85–87.
‘High: We are very confident that the true effect lies close to that of the estimate of the effect.
Moderate: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect.
Very Low: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect’ (section 5)(Reference Schünemann, Brożek and Guyatt27).
* GRADE Quality of evidence grades, taken from GRADE Handbook(Reference Schünemann, Brożek and Guyatt27): Schünemann H, Brożek J, Guyatt G, et al. GRADE Handbook: Handbook for grading the quality of evidence and the strength of recommendations using the GRADE approach.: Cochrane Training; 2013. https://gdt.gradepro.org/app/handbook/handbook.html.
Grey literature
Eight grey literature reports have been included in this review (Table 7). It was not possible to formally quality assess these documents due to a lack of detail in the methodological information available. Five reports explored HS(49,51,52,55,66) , one explored BSF(50) and two evaluated Rose vouchers(53,54) . Amongst the reports focussing on HS, common themes identified included lack of clarity around various aspects of the scheme(52,55,66) , issues around access to retailers signed up to HS(49,51,52) , the use of HS to increase quantity or variety of F&V purchased(51,55,66) and the need to make changes to the eligibility criteria(49,51) . After the introduction of BSF the Scottish government commissioned an evaluation of the scheme. Recipients reported using BSF to purchase a greater quantity or variety of F&V, to reduce financial pressures or as a safety net(50). There were concerns about lack of understanding of some aspects of the scheme, and some felt that eligibility criteria should be broadened(50). In general, the use of a pre-paid card rather than paper vouchers was felt to be a positive change(50).
Grey literature reports considering the impact of F&V voucher schemes
* The food matters document focusses specifically on the participatory workshops that contributed to one of the peer reviewed papers included in this review, by McFadden et al. (Reference McFadden, Green and Williams43).
Finally, two evaluations of the Rose voucher scheme were undertaken(53,54) . Recipients reported consuming more F&V, some used the vouchers to reduce financial pressures and others to purchase larger quantities or varieties of F&V(53,54) . Some recipients felt that the scheme supported healthy habits and that the scheme was likely to change their habits in the longer term, with some reporting improved health outcomes (reduced constipation, feeling healthier, weight loss, improved skin, improved energy levels, improved mental wellbeing), which they saw as being a result of the scheme(53,54) .
Discussion
This systematic review explores the impact of F&V voucher schemes on a range of outcomes. The most commonly included group of outcomes were F&V purchasing and F&V consumption. Overall, F&V voucher schemes did appear to increase fruit and/or vegetable consumption, but confidence in this finding was low. Qualitative data was more consistent. F&V vouchers were used in three main ways; as a financial benefit to subsidise food already being purchased, to increase the quantity or variety of F&v. purchased, or as a safety net, to be used to ensure that the family had something to eat.
There was a lack of consistency of results across the studies included in this review, with different outcomes being considered and some studies finding a positive impact of F&V vouchers whilst others found no significant differences. There are several possible reasons for this. Firstly, evaluating interventions such as F&V voucher programmes is challenging(Reference Jackson64). Often, researchers have to utilise existing datasets or use proxies to determine either exposure or outcome variables, which introduces bias into the study. In many of the studies included in this review, estimates were unadjusted for some or all major confounders that could be expected to impact the results(Reference Herman, Harrison and Jenks32,Reference Ridberg, Marpadga and Akers34,Reference Ridberg, Levi and Marpadga35,Reference Blumberg, Fowler and Bai37–Reference Mouratidou, Ford and Wademan41,Reference Thomas, Moore and Onuselogu47) , mostly due to the data being unavailable. Studies included in this review used a wide range of methods and data sources, and studied several different populations. Some studies, such as that by Ridberg et al., may have been impacted by the COVID pandemic and the introduction of other assistance schemes that could have diluted the impact of the intervention(Reference Ridberg, Levi and Marpadga35). Others reported large differences in the rates of overestimation and underestimation of food intake between intervention and control groups, which, again, may have impacted their results(Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) . When taken in totality, these factors make it challenging to draw firm conclusions from the available data.
Some studies found that F&V vouchers increased fruit and/or vegetable purchasing(Reference Griffith, von Hinke and Smith40,Reference Thomas, Moore and Onuselogu47) or consumption(Reference Herman, Harrison and Afifi33,Reference Ridberg, Marpadga and Akers34,Reference Blumberg, Fowler and Bai37–Reference Ford, Mouratidou and Wademan39,Reference Mouratidou, Ford and Wademan41) , whilst others found that they made no significant differences(Reference Parnham, Millett and Chang20,Reference Ridberg, Levi and Marpadga35) . In the case of the UK-based HS and BSF schemes, vouchers can be used to purchase both cow’s and infant formula milk, as well as F&V, which may have diluted the impact of the vouchers on F&V purchasing and consumption outcomes, particularly in the case of families with infants who are formula fed or not yet fully weaned. This is due to the comparatively high cost of infant formula, which means that it is unlikely that families would be able to purchase both sufficient formula for their child’s needs and F&V with the vouchers provided(67). Overall, when considering the impact of F&V voucher schemes on F&V purchasing and consumption (and associated nutrient intake), the weight of evidence would suggest that, F&V vouchers schemes are likely to increase F&V purchasing and F&V consumption, but to what degree is unclear and confidence in this finding is low. Some studies examining F&V vouchers in older children have found an association between F&V vouchers and increased fruit and/or vegetable purchasing or consumption(Reference Saxe-Custack, LaChance and Hanna-Attisha68–Reference Buscail, Margat and Petit72).
One concern raised about the HS and BSF programmes, is that, by allowing recipients to use their vouchers to purchase infant formula, the scheme incentivises bottle feeding of infants(49). The only study to explore this outcome found inconclusive results, with a negative association between HS and breast-feeding in one dataset, and no significant difference in another(Reference Dundas, Boroujerdi and Browne46). Interestingly, Parnham et al. found that HS recipients spent significantly less on infant formula than eligible non-participants of the scheme(Reference Parnham, Millett and Chang20). Whether the purchase of infant formula should be allowed under HS is a topic that requires careful consideration. The negative health consequences of removing formula (such as the risk of families being forced to ‘water down’ formula under intense financial pressures(Reference Weale and Butler73)) from HS may be more damaging than the potential positive benefits of more F&V consumption.
The evidence for impact of F&V vouchers on health outcomes was limited and conflicting. Wang et al. found no evidence of association between EatSF and maternal/foetal health outcomes(Reference Wang, Seligman and Levi36), but it is important to note that Wang et al. used proxies to determine intervention status, which meant that only approximately 11 % of the intervention group were likely to be receiving the intervention(Reference Wang, Seligman and Levi36). Dundas et al. explored associations between HS and low birth weight and maternal mental health, and found different results in the different datasets analysed(Reference Dundas, Boroujerdi and Browne46).
Impact on food security was also unclear, with two studies reporting no significant differences between intervention and control groups(Reference Ridberg, Levi and Marpadga35,Reference Kropf, Holben and Holcomb38) , and one study finding an increase in food security amongst the intervention group(Reference Ridberg, Marpadga and Akers34). In terms of improving food security, an alternative to F&V vouchers could be a cash benefit. There is some debate about the benefits of vouchers compared with cash benefits and the impacts of these on their intended outcome, with much of the evidence coming from developing countries(Reference Gentilini74). Whether cash or voucher benefits have more impact is likely to be context and intervention specific(Reference Gentilini74), and it is therefore difficult to predict whether cash benefits may offer any additional positive impact over vouchers.
In contrast, the qualitative data included in this review are more cohesive, with striking similarities found across several different studies, and triangulation of these views between recipients and HCP in some cases. Most qualitative studies included in this review explored HS. Three main ways in which HS vouchers are used are highlighted in both the peer reviewed and the grey literature: subsidising food that would have been bought already(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43,Reference Ohly, Crossl and Dykes45) , buying greater quantity or variety of F&v.(Reference Lucas, Jessiman and Cameron42,Reference McFadden, Green and Williams43,Reference Ohly, Crossl and Dykes45) and acting as a safety net to prevent families from going hungry at times of crisis(Reference Lucas, Jessiman and Cameron42,Reference Ohly, Crossl and Dykes45) . It was clear that there were issues with the paper-based HS scheme, with difficulties around applications and eligibility frequently mentioned(Reference Lucas, Jessiman and Cameron42–Reference Moonan, Maudsley and Hanratty44), as well as an acknowledgement that the voucher value has been insufficient to keep pace with rising food costs(Reference Lucas, Jessiman and Cameron42–Reference Ohly, Crossl and Dykes45). HS has recently transitioned into a digital scheme with a prepaid card, which can be used at any retailer which accepts MasterCard(56). There has also been a small uplift in the voucher value, to £4·25 per week for pregnant women and children aged 1–4 years, and £8·50 per week for infants aged 0–1 year(56). Whilst this is likely to have resolved some of the issues highlighted in the literature, the transition has been far from smooth for many(Reference Dundas, Boroujerdi and Browne46,Reference Defeyter, Hetherington and McKean75) , and the increase in HS value has not kept pace with rising food costs(76). Additionally, the eligibility criteria have not significantly changed, so many issues around the exclusion of vulnerable groups are likely to remain, and gaps in eligibility (for example for children aged 4–5, before they may become eligible for free school meals upon starting school(77)). A review published in 2016 exploring the use of vouchers in the HS and WIC programmes, found that vouchers were used to improve dietary quality, and to reduce food expenditure(Reference Ohly, Crossland and Dykes78), both themes also found in this review.
Some alternatives to F&V vouchers have been explored in the literature, for example, F&V or produce boxes(Reference Fischer, Bodrick and Mackey79). Fischer et al. explored the impact of a fortnightly F&V box, delivered to families with preschool children, alongside nutrition education(Reference Fischer, Bodrick and Mackey79). Whilst satisfaction with the programme was high, impact on F&V intake and food insecurity was uncertain with most changes failing to meet statistical significance(Reference Fischer, Bodrick and Mackey79). A recent review exploring produce prescription interventions found that F&V boxes were acceptable to recipients, and some evidence for increased F&V consumption, but concede that evidence in this area could be improved(Reference Muleta, Fischer and Chang80). Whilst F&V boxes may be appealing in some respects, removing concerns about how vouchers are used and perhaps encouraging families to try new foods, they are logistically challenging to organise, and may be wasted if families receive produce that they do not like or do not know how to use. One issue highlighted in this review is, for some, a lack of understanding of nutrition and food preparation knowledge hinders attempts to improve diets for some families. One French study offered nutrition education workshops alongside F&V vouchers. They found that changes in F&V consumption were not associated with attendance at a workshop(Reference Buscail, Margat and Petit72). In their recent scoping review, Greatorex Brooks et al. concluded that educational elements to F&V prescription programmes needed further exploration, in order to better understand their contribution (or not) to the programme’s success(Reference Greatorex Brooks and McInerney81).
All of the studies included in this review examined targeted interventions designed to support those on low incomes, with means tested eligibility criteria. Interestingly, the level of financial hardship needed to qualify differs across the schemes. HS, Rose vouchers and Sainsbury’s top up vouchers have stringent eligibility criteria, whilst BSF has slightly more generous criteria. WIC eligibility (and therefore FMNP and EatSF eligibility) varies by state up to a maximum of 185 % of the federal poverty guidelines(82), resulting in 48 % of children under 5, pregnant and postpartum women being eligible for WIC in 2021(83). Universal provision was raised by some participants as a potential improvement to F&V vouchers(49) and was recommended by the UK Faculty of Public Health in January 2024. There are some benefits to this approach, reduced administrative load assessing eligibility, perhaps a reduction in stigma associated with the vouchers and an emphasis on the importance of healthy diets. However, increased costs of the schemes may be off putting to policy makers. This debate raises the question of whether population or targeted approaches are more successful in terms of improving population health. Clearly, the answer to the question is likely to differ depending on the population, the intervention and the desired outcome(Reference Dodge84). There is some evidence that population level health interventions have the potential for positive impact on outcomes(Reference Zulman, Vijan and Omenn85–Reference Kumanyika, Obarzanek and Stettler88). Whether this would be the case in this context is unclear currently, it may be that a combination of population level and more targeted approaches offers the most effective approach(Reference Ahern, Jones and Bakshis86). These may include focusing on specific geographic areas of higher deprivation to increase uptake or testing extended eligibility criteria in such areas. Further targeted interventions could include cooking sessions during school holidays or mobile vans that provide fresh produce to areas where fewer affordable options are available.
Strengths
This review took a systematic approach and used broad inclusion criteria resulting in the exploration of a range of outcomes. Another strength is the inclusion of quantitative, qualitative and mixed methods studies, which has allowed exploration of the impact of F&V vouchers in a more holistic way. Finally, the inclusion of grey literature has ensured that important findings were not excluded by virtue of not being published in an academic journal, limiting publication bias.
Limitations
Interventions that met inclusion criteria were only found in two geographic regions, the USA and the UK, which limits generalisability of these findings to other parts of the world. Most studies exploring HS looked at the paper-based scheme, and so do not necessarily reflect the current, digital scheme. The wide variety of study designs, methods and outcome measures make it difficult to draw direct comparisons between some of the findings, particularly the quantitative outcomes, and necessitated a narrative approach to synthesis. No quantitative evidence was found for some outcomes included in the review inclusion criteria as follows; nutritional biomarkers, childhood or maternal weight status or childhood healthcare contacts or healthcare utilisation, although many of these topics were explored in the qualitative data. Finally, the review is limited by the quality of evidence available in the literature. Whilst not a limitation of the methods of this review, the majority of the studies included were of designs that are lower in the hierarchy of evidence, and many were not able to control for confounding or had to use proxies to determine intervention or outcome status. This is not unexpected given the type of intervention and need to be pragmatic and make use of available data. However, it does limit the confidence in the findings of the studies, and, in turn, this review.
Conclusions
In conclusion, it is possible that F&V vouchers increase F&V intake, although certainty of evidence is low. It is likely that F&V vouchers have some positive benefits, and they seem to be perceived in a positive light by recipients and staff. It is possible that F&V vouchers may have more significant impacts on certain groups—for example for families with breast rather than formula fed children in the HS/BSF schemes, due to the high cost of infant formula. The food purchasing behaviours of the recipient are also likely to have an impact on the impacts of the scheme, with those using vouchers to subsidise existing choices likely to have different experiences to those choosing to buy more, or more varied F&V. There is a potential for positive mental health impact through reduced financial stress regardless of the approach used when redeeming the vouchers.
This review highlights some of the difficulties that researchers face in evaluating the impact of public health measures to improve population health. More, high-quality research is required to better understand the impacts of F&V vouchers on outcomes. This includes research which considers uptake of the schemes, captures outcomes consistently with longer follow-up, enables researchers to control for confounding and understanding the experiences of people using digital schemes.
It is clear that there are significant operational challenges associated with voucher schemes. Several factors are important to consider when designing F&V voucher schemes; eligibility criteria, accessibility of scheme, voucher value and stigma associated with the scheme, amongst others. It is important that the voucher value of F&V schemes keep pace with food costs and are taken up by those eligible for it. Evaluation of the scheme could help identify potential changes required to ensure that the target population of pregnant women and families with young children benefit from the voucher scheme.
Open access
