Designing Food Voucher Programs to Reduce Disparities in Healthy Diets

Participants for the trial will be recruited via advertisements at community sites including primary care clinics, housing assistance sites, and soup kitchens in the city of San Francisco. Recruitment materials will direct interested individuals to attend one of the study information sessions at a 'store front' site in San Francisco for an initial screening evaluation by the study staff, and a study orientation session to detail the study purpose, procedures, and participant rights. The orientation sessions will consist of a 60-minute session at a centrally-located room. After the orientation session, participants will be invited to sign a consent form, complete the baseline assessment, and be randomized to one of the four intervention conditions in a 1:1:1:1 allocation. We will use permuted block randomization so that sample sizes in each intervention group are nearly equal. Allocation sequence will be concealed prior to assignment; staff for sequence generation, allocation concealment, and implementation will be separate.

Those agreeing to participate will sign a written consent form after the study orientation and before randomized to one of four voucher conditions, as part of our 2-by-2 study design: (1) vouchers targeted to fruits and vegetables and valid for sequential weeks of the month; (2) vouchers targeted to fruits and vegetables and valid over the entire month; (3) vouchers applicable to any food and valid for sequential weeks of the month; and (4) vouchers applicable to any food and valid over the entire month. Half of participants will receive a blue voucher valid for any food (not tobacco, alcohol, or prepared foods), while the other half will receive a green voucher valid only for qualified fruits and vegetables (fresh or frozen fruits, vegetables, or herbs without added sugars or fats, to correspond to the eligibility guidelines of existing voucher programs). All vouchers will have the same individual value ($5), but one half of participants will receive four vouchers that are each valid for the whole month (a lump sum payment of $20), while the other half will receive four vouchers that are each valid for one sequential week of the month, as labeled on the voucher (a distributed payment of $5 per week). Therefore, all intervention groups will be balanced on time and attention from study staff, as well as effort to obtain the vouchers. By having all vouchers in small increments of $5, we minimize the need for a large voucher to be spent all at once, reducing the risk of food spoilage affecting intervention effectiveness. Having all vouchers be worth the same value also simplifies processing of the voucher at the level of the food vendor. We have specifically designed the monetary amounts of the trial to be sufficient for achieving nutritional goals (the Healthy People 2020 goal of consuming 1.4 cups of fruits and vegetables per day), but not be unduly large so as to be coercive to low-income populations.

Vouchers will be mailed to each participant toward the end of the month prior that the vouchers were valid for, i.e. vouchers for March 2017 will be mailed at the end of February 2017. Vouchers will be prepared following the guidelines in the Trial Protocol.

To redeem vouchers, participants present the voucher to a cashier, and sign and date the voucher at the time of redemption. The cashier processes the voucher just as they processes vouchers for the Women, Infants and Children's program or manufacturer's coupons, and the vendor then submits the redeemed voucher for payment (the financial infrastructure is handled by the San Francisco General Hospital Foundation). The names of the vendors who will accept the vouchers are printed on each voucher, along with their address and business hours. All voucher rules will be reviewed with participants during the orientation session. We will continue routine meetings with the neighborhood food vendors' managers to ensure continued active participation and correct usage of the vouchers.

Following informed consent, participants will complete a baseline survey and a 24-hour dietary recall before being randomized to one of the four study conditions. The recall includes three distinct passes to obtain a participant's food intake during the previous 24 hours-a quick list, a detailed description, and a review-including foods obtained at home and away from home, and including bartered, foraged, food pantry and other non-purchased items. Participants are provided standard 2-dimensional booklets for reference to portion sizes. Our research assistants are blinded to the participant's intervention arm and skilled in guiding dietary recalls among diverse participants in a non-judgmental, comprehensive manner. The baseline recall will be conducted in person at the store front site; subsequent recalls will be conducted over telephone. Recalls will be performed during week 1 and week 4 of the month-0 (baseline), month-6, and month-12 assessments. Two recalls each will be done during week 1 (one weekday and one weekend) and two during week 4 (one weekday and one weekend) of the assessment months. This minimizes the number of recalls required to capture intra-person variation in diet between weekdays and weekends, while simultaneously capturing the monthly consumption cycle among participants (the key hypothesized mediator for the lump sum vs. distributed voucher effect).

In addition to dietary recalls, research assistants will orally administer surveys to prevent literacy-based biases, consistent with recent Institute of Medicine recommendations. These 20-minute surveys will be conducted at months 0, 6, and 12 and will cover attitudes and preferences regarding the vouchers and fruits and vegetables generally, shopping patterns, food expenditures, household environment, and food insecurity.

Each participant will receive vouchers for a total of 6 months. The participants will then be tracked by the recruitment/retention coordinator through brief telephone contact monthly to ensure retention until they complete the month 12 survey and dietary recall (6 months after the end of the intervention), after which they are discharged from the study.

In addition to participants who receive the vouchers, we will also conduct a brief survey with managers of the ten stores who have signed memos of understanding with the City of San Francisco and San Francisco General Foundation to accept the vouchers. The managers will be given informed consent forms to consider participating in a post- intervention survey. The paper-based vendor survey will be hand-delivered to vendors' managers to assess: (i) whether managers have any difficulties with the voucher system and whether those difficulties are at the point of sale, at the point of reimbursement, and/or when answering questions to customers about the vouchers; (ii) what if any activities were used by vendors to promote fruit and vegetable purchases during the voucher trial period; (iii) what if any activities they had engaged in to sell more fruits and vegetables; and (iv) a free text entry to elicit what changes vendors would propose to the voucher system in future