Introduction

The COVID-19 pandemic has exacerbated racial/ethnic and socioeconomic disparities in unmet social needs, including food insecurity.1,2,3 In March 2020, national estimates of food insecurity more than tripled, with Black and Hispanic households disproportionately affected.4 As federal relief programs hang in the balance,5 community-based organizations (CBOs) like food pantries have led the response to increased food need.6

Food insecurity should be understood as a leading health issue, compelling healthcare intervention to promote access to adequate and consistent sources of food.7,8 Well-established evidence has shown that decreased food access is associated with higher levels of chronic disease in adults and children, including hypertension, coronary heart disease, stroke, hepatitis, cancer, diabetes, arthritis, chronic obstructive pulmonary disease, kidney disease, and depression.9,10,11,12

The pandemic highlighted four critical challenges that complicated efforts to address food access. First, health systems serving the same communities were individually navigating the landscape of CBOs with food services (e.g., food pantries, food banks), a time-consuming and effort-redundant process. Second, CBOs were struggling to meet the influx of food requests, each limited by their available resources and potential regulatory constraints. Third, social distancing precautions, quarantine and isolation policies, and housing and transportation barriers created a need for food delivery.13 Notably, in [city], neighborhoods hit hardest by the pandemic also had less access to food and vehicles (Appendix A). Fourth, small businesses that delivered food (e.g., catering companies) were facing financial strain.

In response to these interconnected challenges, we engaged key stakeholders from health systems, CBOs, and local Black-owned small businesses across the city to develop and iteratively refine the Food Access Support Technology (FAST): a centralized city-wide digital platform to coordinate referrals for food delivery to households. In this article, we describe the iterative rapid-cycle development of the FAST platform. We then present our evaluation of both the early implementation outcomes and preliminary effectiveness of FAST, using both quantitative process metrics and qualitative semi-structured interviews with a variety of platform users.

Setting and Participants

The design process engaged key stakeholders city-wide through a food access workgroup within an established coalition of multiple health systems, CBOs, and the [city] Department of Public Health (known as Collaborative Opportunities to Advance Community Health or COACH), formed to identify and address pressing food access needs. Workgroup members described the absence of robust infrastructure to coordinate food distribution and referrals, resulting in an inefficient allocation of available resources.

These workgroup meetings provided a forum to introduce the concept of FAST and garner critical feedback from its intended end-users. We presented and elicited feedback on the following key topic areas: data privacy, current workflow challenges for fulfilling food requests, interface preferences (e.g., mobile versus desktop), and pressing or challenging use cases. Focus groups, user interviews, and weekly team assessments informed iterative, rapid-cycle changes to key platform features. During the development phase, stakeholders from COACH reacted to design mock-ups and pilot-tested an interactive demo of FAST before its official launch in March 2021.

At launch, three teams of COACH members led the implementation of FAST, with many members also involved with the initial conception and development process. First, the platform team executed technical and logistical changes for the mobile and web interface. Hired developers from Transmogrify built FAST with a React front-end and a Java and PostgreSQL back-end, hosted on Amazon Web Services. Second, the small business engagement team expanded the collaborative model and interacted with delivery groups, specifically partnering with businesses owned by Black community members in order to augment the capacity of existing CBOs. Members of COACH guided recruitment strategy, using their knowledge and experience to identify appropriate organizations and provide input on organizational readiness. Our staged roll-out model began with one health system and larger community partner organizations, before also incorporating a second health system and smaller organizations. We identified new organizations through the COACH network, as well as online meetings for local CBOs addressing food insecurity. Third, the evaluation team defined implementation and operational outcomes. Platform data provided real-time insights into critical needs, which motivated expansion efforts. For example, identifying the community need for prepared meals prompted our team to apply for and receive two additional grants to support the distribution of prepared meals.

Program Description

FAST facilitates communication and resource sharing between health systems, CBOs, and small businesses to coordinate food delivery to people’s homes (Appendix B). Platform end-users are organizations, who complete a detailed screener so that requests can be matched to them based upon their available services, eligibility criteria, and notification preferences. Food recipients do not engage directly with the platform:

  • Step 1: Posting a Request. After obtaining consent, a provider logs into the password-protected platform to post a request for food. The request includes the recipient’s address, contact information, type of food (unprepared or prepared meals), household size, and amount of food requested, as well as dietary restrictions (e.g., allergies, low sodium) and eligibility criteria for food partners (e.g., age). Both health systems and CBOs can initiate requests for their patients or clients.

  • Step 2: Matching to Food. Once a request is posted, the platform uses pre-determined eligibility criteria to match the request with relevant participating partners and notifies them via text and/or email. Any partner notified can then claim the request.

  • Step 3: Delivering to Homes. Food partners that cannot deliver their own food may re-post the request for delivery only, notifying eligible delivery partners. Once claimed, the patient’s phone number and address will be revealed for delivery completion. In an intentional effort to reinvest in the community, we partnered with Black-owned small businesses to handle the delivery component.

FAST acts as a one-stop shop, providing an efficient posting and claiming process within the platform that previously required half a dozen phone calls between many different organizations. FAST allows requests to be completed in days rather than weeks or months.

Program Evaluation

Evaluating in Real-Time

The FAST platform has built-in data capture of process and implementation outcomes, including status of request, requested food type (unprepared food, prepared meals, or delivery only from a CBO), amount of food requested (in weeks), dietary restrictions, urgent status, household size, and recipient zip code. These data can provide real-time insights into trends in food need and critical gaps in organizational capacity across the city.

Guided by the Consolidated Framework for Implementation Research,14 we conducted semi-structured interviews with five referring providers, six food partners, and one delivery partner to elicit their experiences with FAST, how they used the platform within their current workflow, and perspectives on the current food landscape. In a directed content analysis, our team developed and iteratively refined a codebook to capture emerging themes from coded interview transcripts.

A Look at Fast Requests

Between March 2021 and July 2022, the FAST platform received 364 requests from 2 health systems, representing 207 unique households in 51 distinct postal codes (Appendix A). Of these requests, 258 (70.9%) were completed, via 12 food partners and 2 delivery partners. The remaining were in the process of either completion (8.0%), canceled (15.7%, most often due to platform entry errors), or unfulfilled because recipients were unreachable (5.5%). About 41% of requests were for unprepared (31.0%) or prepared meals (9.6%)—primarily initiated from a health system—with the remaining for delivery only, primarily initiated from a CBO. Most requests were for one (26.1%) or 2 weeks’ (44.0%) worth of food, though requests were completed for more than 10 weeks’ worth of food. Almost half (45.1%) of requests were marked as urgent. Of the 207 unique recipients, 132 recipients (63.8%) lived in a single-person household, and 45 recipients (21.7%) had more than one request (Table 1).

Table 1 FAST requests and recipients: March 2021–July 2022
Full size table

FAST facilitated request completion in a median of 5 (IQR 0–7) days. Specifically, food and delivery partners typically claimed requests in less than a day (IQR 0–5 and 0–1 respectively), with delivery itself completed in a median of 3 (IQR 0–6) days. Requests for prepared meals (7 days, IQR 3–10) took longer to complete than requests for unprepared food (5 days, IQR 0–22.5). Partners completed requests marked as “urgent” (1.5 days, IQR 0–5) faster than non-urgent requests (6 days, IQR 5–8). In addition, our analysis demonstrated increased month-to-month variability in completion time for the first 6 months of FAST, likely the result of early troubleshooting during initial organizational onboarding and platform fine-tuning.

Perspectives of FAST Users

Semi-structured interviews with health systems, food partners, and delivery partners revealed critical insights about both the current food access needs city-wide and user experiences with FAST (Table 2).

Table 2 User perspectives on the implementation of FAST
Full size table

According to our partners, food insecurity in Philadelphia reflects a broader set of cumulative challenges at the (1) individual-level (e.g., transportation, stigma); (2) community-level (e.g., reduced food options, limited nutritional quality); and (3) society-level, with intersecting systems of oppression (e.g., poverty, housing). Many organizations have been striving to build community programs to address food insecurity, but with minimal coordination and collaboration between them. With FAST, platform users aimed to fill an important infrastructural gap, as well as expand their service population by improving outreach, increasing access, and streamlining processes.

Interviewees described FAST as simple to use. Users detailed creative practices to incorporate FAST into their routine workflows, including their usual delivery routes, adding FAST intake into standard screening processes, and leveraging FAST for urgent requests. We anticipate further expansion of workforce capacity and refinement of the technical platform to meet evolving needs. Our users offered several recommendations for platform improvement, which we have since incorporated. Users most often asked for increased platform integration with CBO intake and more nuance for request statuses (Appendix A).

Discussion

Our findings support the promise of centralized platforms to meet food access needs by facilitating streamlined partnerships and real-time coordination of resources between health systems, community food partners, and small-businesses or CBOs that offer food delivery services. During our pilot, FAST served over 200 unique households city-wide, completing most delivery requests in less than a week and urgent requests in just days. Since then, FAST has gone on to serve over 500 households and has developed partnerships to fulfill identified gaps such as emergency food and prepped meals.

FAST expanded upon prior successful interventions, which delivered food to targeted populations, such as older adults or patients with diabetes,15,16,17,18,19,20,21,22 to address food access barriers among the broader city population resulting from the pandemic.23,24 The novelty of FAST lies in its multi-stakeholder approach to resource coordination, filling an important reported gap in the existing landscape of food insecurity and food provision. FAST differs from existing care coordination platforms (e.g., UniteUs, NowPow, FindHelp) in that our model is based on an open marketplace of food delivery requests that any eligible partner can proactively claim, as opposed to a community directory that providers send referrals out to. Additionally, FAST intentionally invests back into the community by partnering with Black-owned small businesses to address unmet service needs of existing non-profit partners.

The implementation of the FAST platform revealed several challenges and generalizable insights for those looking to build similar innovations. First, the platform’s basic units are not individual organizations, but rather organizational clusters, and that should be accounted for in implementation timelines. For example, when a new health system joined the platform, simultaneously recruiting CBOs that served that geographic population became essential to fulfill those posted requests. Second, with built-in data capture, FAST can identify resource gaps as they appear, such as the need for more delivery partners or for prepared meals. Adapting to emerging needs, however, has required agility and resources. We have proactively engaged in strategic alliances and applied for intra- and extramural funding to support emerging needs.

FAST offers real-time insights about city-wide food access needs from platform data and user perspectives that may benefit communities and policymakers. For example, initial platform data revealed the need for prepared meals and expansion of partnerships and pilot funds enabled meeting such identified gaps. In addition, FAST quickly revealed the constraints that CBOs faced with the influx of referrals from multiple health systems in the same geographic area that stemmed from federal incentives to screen patients for social needs. Such insights can inform policy that considers resourcing CBOS to assist health systems with identified social needs.

This study should be interpreted in the context of its rapid-cycle development and implementation, which includes several limitations. The sample size only reflects the platform pilot period, and these data have already informed iterative quality improvement efforts since then. In addition, the platform and its partners are based in a single urban city in the northeastern United States, which may limit generalizability to other settings and differing populations, organizational infrastructure, and needs. Finally, it must be emphasized that the effectiveness of a platform like FAST relies on the quality, efficacy, and cooperation of its partners, in order to build capacity to meet critical gaps identified by the platform. While the platform can seek to optimize resources among CBOs, FAST by itself cannot address the root of inequities in race, geography, and funding underlying food access. The intentional identification of Black-owned small businesses as partners for FAST has been an example of one effort to move upstream and re-invest in our community. Still, our goal is to continue to think creatively and commit to community-owned and community-operated means to improve services in this city.

The success of FAST stems from vital input from our community partners. Future directions include expanding partnerships and evaluating longer-term outcomes, including quality metrics and qualitative experiences at both the organizational and recipient levels. As new policies call on health systems to screen for social needs at point-of-care, FAST offers a rapid-cycle, community-engaged model for efficient resource coordination to support health systems and CBOs in their effort to address patient needs and promote health.