Despite well-established health benefits of fruit and vegetable consumption, increasing evidence recognises a distinction between fruits and vegetables in determinants, impacts on health and intake levels(Reference Appleton, Hemingway and Saulais1,Reference Glasson, Chapman and James2) . Increasing work also suggests health benefits from a high consumption specifically of vegetables, yet intakes remain low, and barriers to increasing consumption are prevalent making intervention difficult(Reference Appleton, Hemingway and Saulais1). A systematic review of interventions to promote consumption of vegetables as a distinct food group found that a variety of successful strategies have been tried(Reference Appleton, Hemingway and Saulais1). However, long-term effects and cost-effectiveness were rarely considered, and certain groups, e.g. individuals of low socio-economic status, were noticeably absent(Reference Appleton, Hemingway and Saulais1).
In a recent pilot of a farm-to-institution intervention for low-income adults served by the Special Supplemental Nutrition Programme for Women, Infants and Children (WIC), the investigators found post-intervention between-group differences in vegetable intake favouring participants who received the intervention relative to those who did not(Reference Di Noia, Monica and Sikorskii3). Receipt of the intervention was also associated with the redemption of Farmers’ Market Nutrition Programme (FMNP) vouchers provided by WIC for produce purchases at farmers’ markets. In this Short Communication, we estimate intervention costs and relate them to primary outcomes of vegetable intake and FMNP voucher redemption. Findings will inform decision makers in other WIC and maternal and child health agencies contemplating implementing the programme or a similar intervention and may serve as a reference for future research in this area(Reference Sukhanova, Ritzwoller and Alexander4).
To determine whether an intervention is a good value for money, it is necessary to estimate and compare the ratio of intervention costs to effects(Reference Cobiac, Vos and Veerman5). To date, there have been few economic analyses of interventions to promote fruit and vegetable consumption relative to the many analyses of programmes targeting other lifestyle behaviours, e.g. physical activity and obesity(Reference Cobiac, Vos and Veerman5). Quality-adjusted life-years have been used extensively in cost-effectiveness analyses for quality of life outcomes(Reference Carson, Hidalgo and Ard6). This is natural because improved quality of life is one of the possible benefits of dietary interventions(Reference Carson, Hidalgo and Ard6). An advantage of quality-adjusted life-years is that they provide a common metric for estimating and comparing the outcomes of what are often heterogeneous interventions(Reference Neumann and Cohen7). However, quality-adjusted life-years may not be consistent with the way decision makers value health interventions. For example, WIC administrators may favour programmes resulting in the largest between-group differences in dietary behaviours and FMNP voucher redemption, which is low in the WIC population (nationwide in 2019, 55 % of FMNP voucher recipients redeemed the vouchers)(8). Thus, the performance of a programme expressed as cost/natural unit, for example, the cost per additional serving of vegetables consumed per person/d, may be more informative to such stakeholders regardless of their efficiency from a quality-adjusted life-year-optimising standpoint(Reference Dalziel and Segal9). For this reason, we did not calculate cost-effectiveness ratios using quality-adjusted life-years or other such metrics as health care cost savings, but instead as cost per intervention effect.
Methods
Setting, design and sample
The research protocol is described elsewhere(Reference Di Noia, Monica and Sikorskii3). In brief, the intervention was piloted in three sites of an urban WIC agency located in New Jersey, USA (one randomised to the intervention study group and two to the control group) with 297 adults (160 enrolled at the intervention site and 137 at control sites)(Reference Di Noia, Monica and Sikorskii3). Intervention participants received routine WIC services, for example, vouchers for nutritious foods and nutrition and breast-feeding education and the intervention(10). Control participants received routine services only. Data were collected via telephone and in person at baseline and at mid- and post-intervention (3 and 6 months post-baseline, respectively). This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving study participants were approved by the William Paterson University Institutional Review Board for Human Subject Research (2018-339). Verbal informed consent was obtained from participants completing telephone assessments and was formally recorded; written informed consent was obtained from those completing in-person assessments.
Intervention
In preliminary work, we examined the feasibility and acceptability of the planned intervention(Reference Di Noia, Monica and Sikorskii11). Findings of this work informed revisions for finalising the programme. The intervention combined behaviourally focused instruction and handouts with the introduction of a WIC-based farmers’ market, field trips to an area farmers’ market, telephone coaching and support and recipe demonstrations and tastings, as summarised in Table 1.
Table 1 Components of the farm-to-WIC intervention
WIC, Special Supplemental Nutrition Programme for Women, Infants and Children; FMNP, Farmers’ Market Nutrition Programme; FV, fruit and vegetable.
Measures
Vegetable intake was assessed objectively using a reflection spectroscopy device, the Veggie Meter® (Longevity Link Inc., Salt Lake City, UT), to assess dermal carotenoids as a biomarker of intake(Reference Scherr, Laugero and Graham12–Reference Radtke, Pitts and Jahns14). Scans of the fingertip were taken in triplicate and averaged (score range = 0–800; higher scores indicate higher carotenoid levels). Vegetable intake was also measured by self-report with the following item: ‘How many cups of vegetables (including 100 % vegetable juice) do you eat or drink each day?’(Reference Yaroch, Tooze and Thompson15). To facilitate the estimation of food portions, participants were told that a cup was about the size of their fist(Reference Ogunbode, Owolabi and Ogunbode16). FMNP voucher redemption was objectively assessed using data provided by WIC. WIC reported whether participants redeemed any vouchers (yes/no) during the voucher redemption period (1 June, to 30 November 2019).
Intervention costs
Direct intervention costs were categorised as fixed, variable or step variable costs. Whereas fixed costs include expenses to develop and implement an intervention regardless of the number of people who receive it, variable costs include expenditures tied directly to the number of recipients(Reference Johns, Baltussen and Hutubessy17). A portion of the variable costs, referred to as ‘step variable costs,’ arise from intervention activities that are performed in discrete units (e.g. with groups of participants)(Reference Graybeal, Franklin and Cooper18). The method of calculating step variable costs was to estimate and assign the cost of an activity to a group of participants and to account for the number of times the activity was offered. Research costs to evaluate the intervention were excluded. FMNP vouchers were provided to participants through the WIC programme; consequently, there were no intervention costs for the vouchers(19). Overhead costs associated with use of the WIC office, equipment and support services were estimated at a rate of 20 % of direct costs. Total costs (direct and overhead) were divided by the number of intervention participants (160) to obtain a unit cost per participant(Reference Given, Bradley and You20). Total and per participant intervention costs were estimated over a 6-month period in 2019 US dollars from the perspective of the agency implementing the intervention.
Analysis
Between-group differences in vegetable intake and FMNP voucher redemption were examined with linear mixed-effects and logistic regression models adjusted for baseline measures and prognostic factors (potential influences on intake)(Reference Di Noia, Monica and Sikorskii3). When prognostic factors had a different distribution by study group at baseline, they were included as covariates in the analyses(Reference Di Noia, Monica and Sikorskii3). To quantify the magnitude of the differences, adjusted Cohen’s d was calculated as the difference between least square means divided by the square root of the residual variance. In a synthesis of reviews and meta-analyses of interventions to promote fruit and vegetable intake in adults, between-group differences in intake at or above 0·30 were considered meaningful(Reference Sweet and Fortier21). For FMNP voucher redemption, effect size was measured using the OR, where 1·5 is a small effect, 2·5 is a medium effect and 4 is a large effect(Reference Rosenthal22). Incremental cost-effectiveness ratios (ICER) were calculated as the total 6-month per-participant intervention cost divided by the post-intervention between-group difference in the outcome. For self-reported vegetable intake, the ICER was expressed as the cost per cup/d by dividing the total per participant cost in the ICER calculation by 180 (the number of intervention days). Analyses were conducted with SAS, version 9.4(23). Statistical significance was set at P < 0·05.
Results
Participant characteristics
Baseline characteristics of participants in the intervention group and the full sample are shown in Table 2. Most participants were Hispanic (73 %), foreign-born (70 %), reported a high school education or less (55 %) and were overweight or obese (75 %).
Table 2 Selected baseline participant characteristics
CDC, Centers for Disease Control and Prevention.
* Includes nine participants reporting more than one race.
† As weight and height were measured at time of study entry, among pregnant participants, pre-pregnancy BMI could not be determined. Pregnant women (n 54) were therefore excluded from analyses of BMI.
Intervention costs
As shown in Table 3, fixed development costs included personnel and participant incentives to pretest and finalise the intervention and translate materials to Spanish (total costs: $6372). Training and supervision costs encompassed personnel (a trainer and trainees (four interventionists and their supervisor), and the supervisor’s time overseeing the interventionists (total costs: $2539). Fixed costs for the WIC-based market included supplies (e.g. tablecloths) and signage for point-of-sale messaging on display at the market. An area farmer provided produce for the market (deliveries were on Mondays, Wednesdays and Fridays), produce bags and bins to display foods, and the collaborating agency provided shelf space and refrigerators to store unsold items for the next market day and canvas bags for participants to carry their produce at no cost to the intervention. Total market costs were $450. Costs for trips to the area farmers’ market included bus service for 12 Saturday trips @ $400/trip and supplies, e.g. trash bags (total costs: $4820). For recipe demonstrations and tastings, costs were incurred for food ingredients and supplies (e.g. cooking equipment; total costs: $937). Fixed costs for coaching and support calls encompassed supplies, e.g. file folders and forms (total costs: $30). Total fixed costs were $15 148.
Table 3 Intervention costs
Hourly rates for part-time staff include fringe at a rate of ·0765.
Variable costs included interventionist time scheduling field trips and coaching and support calls. Relative to calls, additional time was required to schedule trips owing to the additional information discussed, e.g. details of the trip and what to set aside. Variable costs also included interventionist time providing 1:1 instruction at the WIC-based and area farmers’ market, including coordination time between contacts. As coaching calls involved goal setting and progress monitoring, more time was needed for coordinating tasks, e.g. reviewing what was discussed on the previous call. Other variable costs included ingredients for recipe packs and photocopying of handouts. Total variable costs were $12 960. Step variable costs included interventionist time conducting group-based instruction and recipe demonstrations and tastings plus set-up time (total: $832). Overhead costs were $2152. The overhead rate was not applied to the translation services, marketplace activities or use of the bus. Total intervention costs were $31 092 ($194 unit cost per participant).
Intervention outcomes
Mid-intervention, Veggie Meter scores were higher in the control relative to the intervention group (mean (se) = 269·98 (10·06) and 227·11 (9·91), respectively)(Reference Di Noia, Monica and Sikorskii3). Post-intervention, the scores were higher in the intervention relative to the control group (236·01 (9·77) and 212·10 (9·79), respectively)(Reference Di Noia, Monica and Sikorskii3). Self-reported vegetable intake did not differ by group at mid-intervention; post-intervention, it was higher in the intervention relative to the control group (1·83 (0·11) and 1·55 (0·12) cups/d, respectively)(Reference Di Noia, Monica and Sikorskii3). Adjusted rates of FMNP voucher redemption were 87 % in the intervention group and 28 % in the control group. Cohen’s d was 0·35 for objectively measured intake and 0·34 for self-reported intake(Reference Di Noia, Monica and Sikorskii3). For FMNP voucher redemption, the adjusted OR was 17·39 (95 % CI (8·64, 35·02)).
Based on adjusted post-intervention differences between intervention and control groups, ICER were $8·10 for a 1-unit increase in Veggie Meter scores ($194/24 units of study group difference), $3·85 for a 1-cup/d increase in self-reported vegetable intake (($194/180 d)/0·28 cups/d of study group difference) and $3·29 for a 1 % increase in the FMNP voucher redemption rate ($194/(87 %–28 %)). Lower ICER are generally indicative of greater cost-effectiveness(Reference Campbell, Carr and DeVellis24).
Discussion
The study findings highlight the promise of a farm-to-WIC intervention for promoting vegetable purchases and consumption among urban, WIC-enrolled adults. Cohen’s d for between-group differences in objective and self-reported vegetable intake (0·34–0·35) was meaningful(Reference Sweet and Fortier21); for voucher redemption, the between-group difference of 59 % and OR of 17·39 were large(Reference Rosenthal22).
Considering the novelty of the intervention, direct cost comparisons with other studies are difficult. However, an indirect comparison approach (with costs of other interventions adjusted to US dollars with reference year 2019 for comparability) may be useful(Reference Le, Gold and Abbott25). The per participant intervention cost of $194 was lower than that of an educational and skill-building intervention to promote vegetable intake among socio-economically disadvantaged women ($275 per household in 2012 Australian dollars/$294 in 2019 in US dollars where AU $1 = US $0·96, September 2012)(Reference Ball, McNaughton and Le26). In the educational and skill-building intervention, staffing costs were the largest cost, and staff were paid a higher hourly rate ($41 in 2019 US dollars) than in this study, contributing to the higher cost(Reference Ball, McNaughton and Le26). The per participant cost was also lower than that of a price reduction intervention to promote vegetable purchases ($316 per household in 2012 Australian dollars/$338 in 2019 US dollars)(Reference Le, Gold and Abbott25). The higher cost of the pricing intervention was likely due to the added expense to provide the pricing discounts. The per participant cost was also lower than the $240/household cost in 2001 ($346 in 2019 US dollars) for a 6-month voucher programme ($10 voucher/week) to promote fruit and vegetable intake among WIC-enrolled women(Reference Herman, Harrison and Afifi27). Although our programme was also designed to promote vegetable purchases, we leveraged an existing resource (FMNP vouchers) to do so; as such, there were no intervention costs to provide the vouchers.
There are no directly comparable ICER for Veggie Meter and voucher redemption data. For self-reported vegetable intake, the $3·85 ICER was comparable with the cost per cups/d of vegetables consumed in the educational and skill-building intervention ($6·24 in 2012 Australian dollars/$6·67 in US dollars in 2019) and the US voucher programme ($3·44 in 2019)(Reference Ball, McNaughton and Le26,Reference Herman, Harrison and Afifi27) .
The WIC-based market was the least costly component to implement. Thus, in situations where resources are limited, WIC administrators may be able to introduce an onsite market at minimal cost. Cost savings may result from having WIC nutritionists deliver the intervention(Reference Rosal, Lemon and Nguyen28).
The details of cost calculations presented in Table 3 allow programme planners to estimate costs to implement the programme with a different number of intervention participants than in this study. Fixed costs (other than costs to develop intervention materials and messages) would apply to implement the programme as is. If, however, existing content is adapted to meet the needs of local clientele (in terms of local culture and language) or new materials are added, costs to develop, translate and pretest the materials would be incurred. Possible additional fixed costs include physical space and refrigerators to store unsold produce (unless a farmer agrees to deliver produce and pick up unsold items on the same day) and canvas carrying bags for participants. Variable costs also would apply. In non-WIC settings, programme planners considering providing participants with vouchers to purchase produce at farmers’ markets would need to plan for this additional variable cost.
Limitations and strengths
Although findings were comparable to studies of counseling and individual contact, comparisons should be interpreted with caution given differences in study designs and measures. The lack of information on comparable interventions limits the ability to make relative cost-effectiveness comparisons. The small number of study sites and single site in the intervention group limit the extent to which findings can be causally attributed to the intervention despite the randomised design(Reference Di Noia, Monica and Sikorskii3). Despite these limitations, findings add to the limited economic data on interventions to promote vegetable intake in low-income groups(Reference Appleton, Hemingway and Saulais1,Reference Michie, Jochelson and Markham29) . The objective measures of vegetable intake and FMNP voucher redemption are study strengths. The relatively new device for measuring dermal carotenoids is a novel feature of this study(Reference Di Noia, Monica and Sikorskii3).
Conclusion
Costs and cost-effectiveness ratios for a farm-to-WIC intervention compared favourably with those for other interventions targeting vegetable intake in low-income groups(Reference Ball, McNaughton and Le26,Reference Herman, Harrison and Afifi27) . Although these findings suggest that the programme may be cost effective in promoting vegetable purchases and consumption, the cost-effectiveness of the intervention could not be directly assessed as there is no threshold to compare and link ICER expressed as cost/natural unit to longer term health outcomes. Conclusions regarding whether this is in fact the case must therefore await further research(Reference Neumann and Cohen7).
Acknowledgements
Financial support: Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R21CA230476. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The study is registered at www.clinicaltrials.gov (NCT04038385). Conflict of interest: There are no conflicts of interest. Authorship: J.D., D.M., H.J. and A.S. designed the study and obtained funding for the research. J.D. and D.M. carried out the study. H.J. and A.S. analysed the data and interpreted the findings. J.D. wrote an initial draft of the manuscript. All authors provided critical feedback, assisted in revising the manuscript and have read and approved the final version. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the William Paterson University Institutional Review Board for Human Subjects Research. Verbal informed consent was obtained from all participants completing telephone assessments and was formally recorded; informed written consent was obtained from all participants completing in-person assessments.
