Improving Food Insecurity Toolkit to Improve Screening, Follow-up, and Referral Process for Food Insecure Patients at the Urban Indian Center of Salt Lake

Food insecurity is the inability to purchase the proper quality or quantity of food for optimal health. More than one in four American Indians and Alaskan Natives experience food insecurity (Move for Hunger, 2023), which is associated with higher rates of chronic disease and depression (Hager et al., 2010; Cain et al., 2022). Providers face many barriers to adequate care for food insecurity, especially a lack of knowledge and confidence. Local Problem: Providers at the Urban Indian Center of Salt Lake are concerned about food insecurity in the Native American community. The Urban Indian Center of Salt Lake lacks a streamlined approach for screening, identifying, and referring food insecure patients. Key improvement opportunities include inconsistent screening, education, and follow-up practices, low confidence in addressing positive food insecurity screenings, and a lack of knowledge about community resources. Methods: This quality improvement project used quantitative data from a chart review to assess the total number of patients seen, the number of patients who received a food insecurity ICD-10 code, and the number of patients who received a referral to the dietician. This project also collected quantitative and qualitative data via surveys to evaluate providers' comfort and knowledge levels and identify perceived barriers and facilitators to addressing food insecurity. Interventions: This project was implemented at the Urban Indian Center of Salt Lake, a small urban clinic providing free healthcare to the Native American population. A food insecurity toolkit was developed, implemented, and evaluated. The toolkit contained information on food insecurity, a process flowchart to streamline patient care, and a list of resources. Results: During the eight-week pre-implementation period, zero of 313 patients (0.00%) received a food insecurity ICD-10 diagnosis, and three of those 313 received a referral to the dietician for food insecurity. During the eight weeks post-implementation, six of 215 patients (0.03%) received a diagnosis, while none received a referral to the dietician. Each of the four interviewed providers reported increased knowledge in addressing positive screeners and greater comfort in discussing food insecurity resources. All providers thought the toolkit was feasible for daily practice. Zero respondents would change the contents of the toolkit, and all endorsed the integration of the toolkit into daily practice. Conclusion: Following the toolkit implementation, providers exhibited more knowledge and confidence in screening for and addressing food insecurity. Despite improvements, toolkit implementation revealed underlying gaps in how providers chart efforts related to screening for and addressing food insecurity. Future strategies should include streamlining charting processes related to screening and referring patients for food insecurity.

1 Implementing a Food Insecurity Toolkit to Improve Screening, Follow-Up, and Referral Processes for Food Insecure Patients at the Urban Indian Center of Salt Lake Emily R. Royce, Danielle Puri, and Jennifer Clifton College of Nursing: The University of Utah NURS 7703: DNP Scholarly Project III March 31, 2024 2 Abstract Background: Food insecurity is the inability to purchase the proper quality or quantity of food for optimal health. More than one in four American Indians and Alaskan Natives experience food insecurity (Move for Hunger, 2023), which is associated with higher rates of chronic disease and depression (Hager et al., 2010; Cain et al., 2022). Providers face many barriers to adequate care for food insecurity, especially a lack of knowledge and confidence. Local Problem: Providers at the Urban Indian Center of Salt Lake are concerned about food insecurity in the Native American community. The Urban Indian Center of Salt Lake lacks a streamlined approach for screening, identifying, and referring food insecure patients. Key improvement opportunities include inconsistent screening, education, and follow-up practices, low confidence in addressing positive food insecurity screenings, and a lack of knowledge about community resources. Methods: This quality improvement project used quantitative data from a chart review to assess the total number of patients seen, the number of patients who received a food insecurity ICD-10 code, and the number of patients who received a referral to the dietician. This project also collected quantitative and qualitative data via surveys to evaluate providers’ comfort and knowledge levels and identify perceived barriers and facilitators to addressing food insecurity. Interventions: This project was implemented at the Urban Indian Center of Salt Lake, a small urban clinic providing free healthcare to the Native American population. A food insecurity toolkit was developed, implemented, and evaluated. The toolkit contained information on food insecurity, a process flowchart to streamline patient care, and a list of resources. Results: During the eight-week pre-implementation period, zero of 313 patients (0.00%) received a food insecurity ICD-10 diagnosis, and three of those 313 received a referral to the 3 dietician for food insecurity. During the eight weeks post-implementation, six of 215 patients (0.03%) received a diagnosis, while none received a referral to the dietician. Each of the four interviewed providers reported increased knowledge in addressing positive screeners and greater comfort in discussing food insecurity resources. All providers thought the toolkit was feasible for daily practice. Zero respondents would change the contents of the toolkit, and all endorsed the integration of the toolkit into daily practice. Conclusion: Following the toolkit implementation, providers exhibited more knowledge and confidence in screening for and addressing food insecurity. Despite improvements, toolkit implementation revealed underlying gaps in how providers chart efforts related to screening for and addressing food insecurity. Future strategies should include streamlining charting processes related to screening and referring patients for food insecurity. Keywords: food insecurity, social needs screening, American Indian/Alaskan Native, Native American 4 Implementing a Food Insecurity Toolkit to Improve Screening, Follow-Up, and Referral Processes for Food Insecure Patients at the Urban Indian Center of Salt Lake Problem Description Food insecurity (FI) is the inability to purchase the proper quantity and quality of food for an active and healthy life, and the condition can be long-term or temporary (Feeding America, 2023). According to the United States Department of Agriculture, 12.8% of US households experienced FI in 2022. In Utah, 10% of households experience FI (Johnson, n.d.). While FI can affect anyone, it disproportionately affects American Indians and Alaskan Natives (AI/AN). In general, the AI/AN population experiences FI at a rate of one in four—slightly more than double the rate of Americans as a whole (Move For Hunger, 2023). Additionally, AI/ANs living in Utah underutilize available food resources. For example, they have the lowest utilization of the Special Supplemental Nutrition Program for Women, Infants, and Children of all urban Indian service areas in the United States (Urban Indian Health Institute, 2023). Medical providers at the Urban Indian Center of Salt Lake (UICSL) are concerned about the high prevalence of FI in and underutilization of resources by the AI/AN population. To screen patients and intervene more effectively, UICSL providers agreed to participate in a quality improvement (QI) project to refine the clinic’s care of food insecure patients. Providers noted three problem areas: (1) poor consistency with screening, education, and follow-up for foodinsecure patients, (2) a need for more confidence in knowing what to say and do if their patient screens positive for FI, and (3) a lack of knowledge regarding community resources that address FI. Overall, the high prevalence of FI in the AI/AN population, paired with provider self-reports of poor consistency of screening, follow-up, and referral processes, presents an opportunity to improve clinic systems that address FI. 5 Available Knowledge Food insecurity is prevalent and transcends age, race, and culture (Nikolaus et al., 2022). In the United States, FI is rarely associated with an underweight BMI and is distinct from starvation. Rather, it is present in otherwise healthy-appearing individuals. Because FI is not associated with outward physical signs, such as compromised height and weight, the condition is invisible unless screened for (Hager et al., 2010). However, there are some methods for detecting FI. One example is the two-question FI screening tool, the Hunger Vital Sign, which has high sensitivity and moderate specificity for detecting FI in various populations (Hager et al., 2010). Across all ages, FI is associated with poorer general health (Dong et al., 2023; Gundersen & Ziliak, 2015). In AI/AN youth, FI is associated with a higher body mass index and hypertension as well as sleep disturbances (Dong et al., 2023). In adults, FI is associated with higher rates of diabetes, hypertension, and hyperlipidemia and decreased nutrition intake (Gundersen & Ziliak, 2015). In seniors, FI is associated with more limitations in activities of daily living—as if they were 14 years older than their current age (Gundersen & Ziliak, 2015). FI also has a dose-dependent relationship with depression—the more food insecure a patient is, the more depressed they are likely to be (Kolovos et al., 2020; Cain et al., 2022). The same dose-dependent relationship exists between FI and general mental health (Cain et al., 2022). This finding is true in many age groups (Kolovos et al., 2020; Cain et al., 2022). Providers face several barriers when screening for and addressing FI (Kopparapu et al., 2020; Schickendanz et al., 2019). First, there are complexities in administering FI screenings and referrals, such as time constraints during visits, too many steps in the referral process, poor communication, a lack of knowledge about resources and referral processes, and a general lack of available resources (Kopparapu et al., 2020; Schickendanz et al., 2019). Second, few 6 providers screen for FI, though most believe screening is important (Schickendanz et al., 2019). Third, some providers do not feel confident addressing social determinants of health such as FI (Schickendanz et al., 2019). Researchers Caldwell et al. (2023) found that screening and referring patients for FI lessens the amount of FI in a given population. These researchers also found that infrastructure, staff training, and streamlined workflows are essential for successfully implementing FI screening processes. After identifying FI needs, the researchers found a strong need for an in-person handoff (ie. warm handoff) to ensure patients’ connection with resources (Caldwell et al., 2023). Lastly, the researchers found that patient preferences included a list of relevant locations (e.g., food banks, community organizations, farmer’s markets), assistance signing up for financial programs, and nutritionist referrals (Kopparapu et al., 2020). Rationale This project aims to improve the consistency and quality of FI screening, follow-up, and referral. The model used to achieve this goal will be the Institute of Health Care Improvement’s (IHI) quality improvement model (2023). This model uses aims, measures, and changes to guide quality improvement projects (IHI, 2023). The three questions we asked ourselves were: (1) What are we trying to accomplish? (2) How will we know that a change is an improvement? and (3) What change can we make that will result in improvement? (IHI, 2023). First, the aim of this project was to implement a FI toolkit that assists providers in following up on positive FI screening results. Second, the measures used to quantify the change include the number of FI International Classification of Disease (ICD-10) codes assigned to visits, the number of referrals placed to the dietician for food insecure patients, and data from pre- and post-implementation provider surveys. Last, the specific changes consist of developing 7 a toolkit and a process flowchart that providers can use to improve the FI screening and referral processes. The toolkit will be implemented through Plan-Do-Study-Act (PDSA) cycles because of their iterative nature. The project will be planned according to recent evidence and best practice guidelines and then undergo one PDSA cycle. After information is gleaned from the first PDSA cycle, the project will be turned over to UICSL’s quality improvement committee for further PDSA cycle testing. Specific Aims The purpose of this Doctor of Nursing quality improvement (QI) project is to develop and implement a FI toolkit that aims to improve provider knowledge and increase the frequency of follow-ups and referrals for all food insecure patients at the Urban Indian Center of Salt Lake (UICSL). Methods Context This quality improvement project took place at the Urban Indian Center of Salt Lake (UICSL). This small urban clinic provides free healthcare to registered AI/AN tribal members. Most patients live in or around Salt Lake City. At the time of the intervention, this clinic had three part-time family nurse practitioners, one part-time psychiatric mental health nurse practitioner, and one nurse. These workers are all employees of the University of Utah College of Nursing and are contracted to work at UICSL. The medical assistants, dieticians, and clinic managers are employees of UICSL. Only the four nurse practitioners were given the pre- and post-surveys. The clinic nurse does not typically screen for and address food insecurity with UICSL patients and so was not included in the pre- and post-surveys. In this paper, the terms 8 'provider(s)' and 'nurse practitioner(s)' will be used interchangeably since, during the intervention period, nurse practitioners were the sole type of providers at UICSL. Many UICSL clients are considered low-income—a significant risk factor for FI (Ashbrook et al., 2021; Nikolaus et al., 2022). Consistent screening, follow-up, education, and referral can improve the lives of this higher-risk population (Caldwell et al., 2023). The FI toolkit aims to assist providers in identifying food insecurity through screening and connecting patients with valuable food resources or additional support as needed. Intervention(s) The planned intervention was to develop, implement, and evaluate an FI toolkit using one PDSA cycle. The toolkit was loosely based on the American Academy of Pediatrics and Food Research and Action Center’s FI toolkit for pediatricians (Ashbrook et al., 2021). The lead nurse practitioner and dietician helped tailor the toolkit to fit UICSL’s clinic and population needs. The toolkit contained information on FI, such as risk factors, health implications, and a list of community resources. The toolkit also included a process flowchart that outlined a list of steps to follow depending on the patient’s FI status. The toolkit’s goal was to aid providers in improving the consistency of screening, quality of follow-up on screener results, and placement of patient referrals if indicated. Study of the Intervention(s) A pre-implementation survey was given to providers to assess their primary demographic data, current FI practices, barriers and facilitators in addressing FI, and their comfort and knowledge levels with screening for and addressing FI. Eight weeks later, a post-implementation survey was given to assess the same information and the feasibility, usability, and satisfactoriness of the FI toolkit. A chart review was also conducted before implementation and 9 eight weeks following implementation. This chart review evaluated the total number of patients the clinic saw, the number of FI ICD-10 codes placed, and the number of referrals made to the dietician for food insecure patients. The FI toolkit process flowchart counseled providers to refer food insecure patients to the dietician if they had hypertension, diabetes, obesity, or some other deviations from the growth curve. Patients could also be referred based on provider discretion. Before this quality improvement project, providers and clinic staff were not trained on the risk factors and health implications of FI. There were also no processes outlined and readily available for managing food-insecure patients. Therefore, it can be reasonably assumed that an increase in FI ICD-10 codes resulted from this intervention. Measures A chart review and provider surveys were developed and administered to measure the toolkit’s effectiveness in improving provider knowledge, follow-ups, and referrals for all foodinsecure patients at UICSL. The chart review compared data from pre- and post-implementation. The time frame for data retrieval was eight weeks pre-implementation and then eight weeks postimplementation. The outcomes analyzed included the total number of patients seen, the number of FI ICD-10 codes used, and the number of referrals placed to the dietician. Providers added an ICD-10 code to visits where FI was specifically addressed to track the frequency and prevalence of FI in the AI/AN population. Clinic leaders decided that FI referrals would be placed per provider discretion and for patients with hypertension, diabetes, obesity, underweight BMIs, and deviations from the growth curve. These indications for referral are based on evidence of their association with FI (Dong et al., 2023; Gundersen & Ziliak, 2015). Additionally, the total number of patients seen during the two eight-week periods provides a base for comparison to 10 understand the prevalence of FI in this population and the proportion of patients who need additional help from the dietician. A 34-question pre-implementation survey using multiple choice, select all that apply, and free text options was developed to assess the following items: providers’ primary demographic data, current FI practices, barriers and facilitators to FI screening and referral, and knowledge and comfort levels related to screening for and addressing FI. Demographic data included each provider/nurse practitioner’s specialty and how frequently they work (e.g., PRN, part-time, fulltime). Providers’ current FI practices were assessed by asking how frequently they screen for FI and whether they use the validated two-question Hunger Vital Sign screener or assess FI in some other way. This section also included questions about whether providers discuss positive screening results or review resources with patients. Providers were asked which resources they provide and given the answer options of “yes,” “no,” and “sometimes.” If the providers answered “no” or “sometimes,” then branching logic was used to evaluate reasons such as “Not enough time during visit to discuss resources.” (for other options, see Table 2). The following section asked about barriers to screening for and addressing FI and what providers thought could help increase patient access to FI resources. The knowledge section evaluated providers’ confidence based on their agreement with statements such as, “I know what to do if my patient screens positive for food insecurity.” Providers were also assessed on the health implications of FI and when to refer patients to the dietician. Finally, providers were asked about their comfort level with interpreting a positive screening result, discussing a positive screener with a patient, and introducing FI resources. After the implementation period, respondents were asked the same questions and additional open-ended questions assessing the toolkit’s usability, feasibility, and satisfactoriness. 11 Analysis Descriptive statistics were used to analyze the study sample and the chart review data for the eight-week pre- and post-intervention periods. Due to a small sample size (n=4), no inferential statistical test was used. Content analysis was also used to analyze the open-ended survey questions regarding feasibility, usability, and provider satisfaction. All content analysis data was read, organized, summarized, and used to inform toolkit improvements. Ethical Considerations This study is a quality improvement study by nature and is not subject to institutional review board (IRB) oversight. However, because the UICSL serves a vulnerable population, this project had to request an IRB exemption from the Indian Health Service (IHS). The Phoenix Area IHS IRB reviewed and approved this quality improvement project. Participants voluntarily responded to the survey questions, and their answers are anonymous. There are no conflicts of interest. Results The data for this quality improvement project were collected from a pre- and postimplementation chart review, provider surveys, and a log of how many food resource sheets were given out during the eight-week implementation period. Results from each area will be discussed in the subsequent sections. Due to the small sample size of four survey respondents, only descriptive statistics will be used. A retrospective chart review was completed by UICSL research assistants to obtain preand post-implementation outcome measures. These measures were used to understand the impact of the FI toolkit on patients diagnosed with food insecurity and food insecure patients who needed a referral to the UICSL dietician. The findings from the chart review are listed in Table 1. UICSL providers saw a total of 313 patients during the pre-implementation period (9/4/23– 12 10/30/23) and 216 during the post-implementation period (11/6/23–12/29/23). During the eightweek pre-implementation period, zero patients were diagnosed with food insecurity (as measured by the number of FI ICD-10 codes assigned). In the eight weeks following implementation, 6 of 215 (0.03%) patients were diagnosed with FI. While the post-implementation data show a sixfold increase in FI-related ICD-10 codes, the overall prevalence of those diagnosed with FI remained very low (0.03%). These results are surprising given that 10% of Utah households experience food insecurity and one in four Native Americans are food insecure (Johnson, n.d.; Move For Hunger, 2023). While we acquired the total number of patients seen by the clinic, we could not obtain the number of patients screened for FI due to inconsistent charting practices and an inability to query screening results. These missing data significantly limit our ability to understand the prevalence of food insecurity among those screened for it. The chart review also assessed the number of referrals to the dietitian during the pre- and post-implementation periods. Three referrals were placed to the dietician via the EHR preimplementation, and zero were placed post-implementation. Fewer referrals post-implementation is surprising, despite the increase in the number of FI ICD-10 codes. The Discussion section will provide additional interpretation and speculation regarding the chart review results. Survey findings will be presented in the order of the survey sections. Demographic data were only assessed in the pre-implementation survey. Food insecurity practices, barriers and facilitators to screening for and addressing FI, and comfort and knowledge levels regarding FI were assessed in both surveys. Provider feedback pertaining to the usability, feasibility, and satisfactoriness of the toolkit was only assessed in the post-implementation survey. The UICSL is a relatively small urban clinic, so the sample size of providers (n=4) is likewise small. Four out of four (100%) providers responded to the pre-implementation and post- 13 implementation surveys. Results showed that 75% (n=3) were family practice nurse practitioners, and 25% (n=1) were psychiatric mental health nurse practitioners. Seventy-five percent (n=3) of nurse practitioners worked part-time, and 25% (n=1) worked PRN. Refer to Table 2 for a complete summary of the following data. In the preimplementation phase, three providers (75%) said they screen more than half of their patients every six months for FI. One provider (25%) responded that they screen less than half of their patients for FI every six months. However, in the post-implementation phase, there was an improvement in consistent screening practices; two providers (50%) said they screen more than half of their patients for FI every 6 months, and the other two (50%) said that they always screen for FI every 6 months. During the pre-implementation phase, when providers were asked about their use of the validated two-question FI screening tool (Hunger Vital Sign [Hager et al., 2015]), one provider (25%) reported using this tool, two providers (50%) didn’t know what this tool was, and one (25%) said that they were not using the tool but instead assessing FI in some other way. The three providers who fell into these last two categories assessed FI status by gleaning information during visits. Post-implementation data showed that the use of the two-question screener increased from 25% to 100%, demonstrating increased consistency in the method for screening for FI. The surveys also assessed whether the providers discussed the FI screener with their patients. In both surveys, three providers said they always discuss a positive screener, and one said they sometimes discuss a positive screener. Barriers cited for not discussing the screener included time constraints, forgetfulness, thinking someone else would discuss the screener, and not seeing screening results until the patient had left. 14 Pre- and post-implementation results showed that all providers were consistent in handing out food resources to food insecure patients (75% pre-implementation and 100% postimplementation). UICSL’s internal food resource page was the most common resource provided. The food bank was the only other additional food resource cited. Overall, post-implementation survey findings show improved consistency in screening practices, use of the two-question FI screener, and resource provision for food insecure patients. While educating clinic staff on the toolkit and concluding that it is the provider’s responsibility to discuss the screener result or delegate someone else to address it, providers still listed “thinking someone else would discuss the screener” as a barrier. It is also interesting to note that providers reported consistently handing out food resources even though they placed few FI ICD10 codes in patient's charts. There was no change in the frequency that providers endorsed addressing positive FI screenings during their visits from pre- to post-implementation. There were several reported barriers to screening for FI in the pre-implementation survey. These included a lack of training (50%), insufficient time to screen (25%), discomfort with asking about FI (25%), and discomfort in addressing food insecurity (25%) (see Table 3). While no providers cited a “lack of support staff to administer screener” as a barrier in the preimplementation survey, three (75%) providers cited this as a barrier in the post-implementation survey. This was likely caused by the changes in staffing during the intervention period. Providers continued to report a lack of training and time as barriers to screening in the postimplementation survey. Providers did not provide further detail regarding areas where they would like more training regarding FI and the toolkit. Providers were also asked about barriers to addressing FI in visits (see Table C), and their answers seem to contradict the previously reported barrier of “lack of training”. The pre- 15 implementation survey showed that 75% of providers considered a “lack of knowledge about resources for food insecure patients” to be a barrier to addressing FI. In the post-implementation survey, 0% of providers reported this lack of resources as a barrier. This survey result suggests that the toolkit effectively increased provider knowledge about FI resources. This contraindication will be discussed later in the manuscript. In the pre-implementation survey, all providers preferred the method of increasing access to FI resources by providing a page of resources in exam rooms. In the post-implementation survey, most providers (75%) switched their preference to providing a resource handout at the end of the visit. This finding suggests that following toolkit implementation, provider preferences for addressing food insecurity switched from a passive to an active methodology. This could be due to increased confidence in addressing FI. Two knowledge-based questions were placed in the pre- and post-implementation surveys to assess the effectiveness of the toolkit training session (see Table 4). In both surveys, none of the providers got either question completely correct, though they answered most of the select all that apply questions correctly. This implies that the quality of the training session for the FI toolkit could be improved. For example, future trainers could use quiz questions during training to increase provider knowledge retention. The surveys also assessed whether providers knew what to do if their patient screened positive for FI. Pre-implementation survey data showed that all providers (100%) somewhat agreed that they knew what to do if a patient screened positive for food insecurity, while in the post-implementation survey, 100% fully agreed that the knew what to do. After toolkit implementation, providers endorsed greater comfort discussing a positive screening with patients (100% post-implementation versus 50% pre-implementation). Post-implementation data showed 16 that 100% of providers were comfortable discussing food insecurity resources with patients, compared to just 25% pre-implementation. The survey findings illustrate that although the toolkit training could be improved, providers endorsed increased knowledge of what to do with a positive screening result and increased comfort in discussing positive FI screenings and FI resources with patients. At the end of the post-implementation survey, providers were asked multiple-choice and open-ended questions regarding the feasibility, usability, and satisfactoriness of the intervention. Please refer to Table 5 to see complete data for the multiple-choice questions. The first multiplechoice question asked whether providers thought the two-question FI screener accurately identified food insecure patients; 100% of the providers responded yes. The second question asked about the feasibility of implementing the toolkit into standard practice; 100% (n=4) of providers reported this as very feasible. All providers recommended incorporating the toolkit into standard clinic practice. Open-ended questions allowed participants to provide descriptive feedback about the toolkit (see Table 6). When asked what they liked about the toolkit, providers cited various aspects, such as the convenience of resources, the thoroughness and accuracy of the resource sheet, the fast documentation, and the overall ease of use. Regarding the detail of the resources, one provider liked “how thorough the resource sheet was with up-to-date information and how [the resource sheet] included details to help the patient decide what resource would be best for them.” When asked what they would change about the toolkit, all providers (100%) indicated no desire to change its contents. Upon being asked whether there was anything else participants deemed valuable to know for this QI project, three participants (75%) stated they had nothing further to add, while one participant (25%) stated, “I really like that the food resources handouts 17 are available for providers to give out to patients and families if they screen positive. It is also nice to have a dietitian consult with patients who screen positive.” The open-ended questions indicated that participants are satisfied with the toolkit’s convenience, thoroughness, and detailed resources. All providers would like to use the toolkit in daily practice, and most believe this intervention is feasible. We also tracked the number of FI county resource pages (obtained from the Utahns Against Hunger website) and UICSL-specific resource pages handed out (see Table 7). The medical office gave out five resource sheets for Salt Lake County, one resource sheet for Utah, Summit, and Wasatch Counties (resources for these three counties were combined into one page), and four UICSL-specific resource sheets. The front desk provided only one UICSLspecific sheet. The number of resource sheets provided was not tracked before project implementation, so it was impossible to compare the frequency of resource provision pre- and post-implementation. As previously mentioned, the inability to obtain the number of patients screened for FI is a significant limitation in this study’s data. Obtaining information about the total number of patients screened for food insecurity, the number who screened negative, and the number who screened positive would lend perspective to the six FI ICD-10 codes assigned to visits postimplementation. Another notable absence is data regarding referrals to the dietitian. It is plausible that providers directed patients to the dietitian, potentially noting this in the written section of the plan, without entering a formal referral into the EHR. Per the dietician, providers often asked her in-person (via warm hand-off) to provide additional support at the end of the visit. She noted that this may have resulted in fewer formal referrals placed in the chart for FI because the need for 18 additional support was already communicated to her in person. These absences mean more patients may have been referred to the dietician than the data shows. Lastly, it is important to note that the updated UICSL-specific FI resource sheet was not implemented until a month after the intervention began. The initial plan was to put this resource sheet into effect at the start of implementation; however, calling resources and searching websites for detailed information took a significant amount of time. In the meantime, we used a different resource (the Utahns Against Hunger food resource sheets), which contained a pareddown version of the information on the UICSL-specific sheet (see Appendix C for both resources). We determined that taking extra time to ensure that the UICSL-specific resource sheet contained detailed information about resources (such as the availability of fresh produce versus non-perishable food, hot meals, accurate hours of operation, and requirements for utilizing resources) would be invaluable in helping patients decide which resource they prefer. The late implementation of the UICSL-specific resource page could have resulted in fewer of this specific type of resource page handed out, but it would not have reduced the total number of FI resource options provided to the patient. Discussion Summary This QI project evaluated how implementing a food insecurity toolkit would affect provider screening, follow-up, and referral processes for food insecure patients at UICSL. The IHI Model for Improvement was used to set the project’s aims, measures, and changes. The principal aim was to implement a food insecurity toolkit into practice. The measures—including the number of FI ICD-10 codes, the number of dietitian referrals, and the results of the pre- and post-implementation surveys—increased knowledge about barriers to and facilitators of improving screening, follow-up, and referral practices. 19 The final question in the IHI framework focuses on changes that result in improvement. In this project, the chart review data revealed that the FI toolkit had minimal to no effect on how many patients were diagnosed with FI nor how many food insecure patients were referred to the dietician. After the toolkit was introduced, only six patients of 215 (0.03%) received an FI ICD10 code. Even though this number increased from zero patients during the pre-implementation period, the prevalence of food insecurity diagnoses remained very low. These data are not congruent with the information on FI prevalence among patients provided by UICSL providers nor the latest estimates of local food insecurity among AI/AN populations (Johnson, n.d.; Move For Hunger, 2023). Additionally, the decrease in dietitian referrals from three referrals preimplementation to zero post-implementation is perplexing because the number of documented FI ICD-10 codes increased after toolkit implementation. These incongruencies suggest an underlying barrier to the implementation of this QI project. Further speculation about potential barriers to implementation will be discussed in the Interpretation section. The survey results from before and after the implementation of the toolkit corroborate providers’ initial concerns regarding discomfort in addressing FI and a lack of familiarity with available FI resources. Pre-implementation survey results showed inconsistency in the frequency of screening for FI and inconsistency in how providers screened patients for FI. By contrast, post-implementation results showed improved consistency in screening patients for FI every six months and a 300% increase in the use of the validated two-question screening tool (The Hunger Vital Sign, Hager et al., 2015). Such improvements can help streamline how patients are screened for FI and decrease the likelihood of a missed FI diagnosis. In the surveys, providers also enumerated various barriers to addressing FI—mainly a lack of time and training as well as inconsistent clinic processes for registering and rooming patients. Despite these barriers, 20 providers endorsed greater confidence in discussing FI resources after toolkit implementation, and all expressed the desire to integrate the toolkit into standard clinic practice. Despite this study’s limitations, certain strengths can also be identified. One strength of the study is that it contributes knowledge regarding clinic-level efforts to improve the detection and care of food-insecure Native Americans. Another strength of this study is that it contains specific positive provider feedback regarding the toolkit’s feasibility and usefulness. There was an overwhelming endorsement for integrating the toolkit. Interpretation Chart Review While the number of FI ICD-10 codes increased from zero to six after toolkit implementation, the reported percentage of food insecure patients at UICSL (0.03%) remained very low. This figure is inconsistent with existing literature, which shows a high prevalence of FI among Utahns (one in ten) and a higher prevalence among Native American populations (more than one in four) (Johnson, n.d.; Move For Hunger, 2023). However, based on discussions with the dietician, it appears that FI is more prevalent among UICSL patients than the data suggests. The dietician reports that from January to February 2024, she enrolled 45 UICSL clients in an FI-related program, which means that the UICSL case manager or providers identified 45 foodinsecure individuals in need of further food resources. Per the dietician, “We could have doubled that [number] if I had more money and didn’t limit [enrollment] to certain hours.” She also stated that the patients love the program, which was initiated in January 2024 and provides $25 or $35 twice a month, depending on family size, to purchase dried beans, fresh and canned goods, and frozen fruits and vegetables. When assessing the significance of this data in understanding the prevalence of food insecurity at the UICSL health clinic, it is important to note that UICSL 21 offers additional services beyond medical care, such as therapy and youth programs. Therefore, the 45 individuals referred to the dietician for enrollment could have been patients of the medical clinic or participants from other programs at UICSL. As of January 2024, the dietician started requiring providers to place a formal referral in the EHR before enrolling patients in the food insecurity program. While the toolkit guided providers to input a referral to the dietician, the toolkit did not specify where in the chart to place the referral. This oversight may have resulted in documentation of referrals in the text of a note rather than in a formal location—making it difficult to track how many referrals were actually placed. UICSL elected to have their research assistants complete a chart review for the pre- and post-implementation data and did not allow us access to patient charts. Understandably, it was infeasible for the research assistants to review the 528 patient charts individually in this project’s time frame because of their other duties. Another likely reason for the discrepancy between reported FI levels and the project results is inconsistent charting practices. For example, UICSL switched to a new EHR system at the end of May 2023 and is still in the process of moving over. Every six months, patients fill out a paper food insecurity screener as part of the UICSL’s wellness packet, and then the medical assistants enter the results into the EHR. However, there is no consistent location for the documentation of screening results, which makes it difficult to track how many patients were screened for FI. As for referrals to the dietician, if the providers did not place a formal referral in the EHR, there is no place in the charting system that can easily be queried to obtain information about who needed a referral to the dietician. Fixing these documentation issues is key to tracking the number of patients screened for food insecurity, identified as food insecure, and given more assistance with resources. 22 The discrepancy could have also been impacted by the fact that two of the four providers surveyed are no longer working at UICSL as of January 2024. With this turnover, there could have been less provider buy-in, which may have impacted referral and charting practices. There is also a slight possibility that the clinic saw more food insecure patients in January 2024 than in October, November, and December of 2023. Outside research shows that even though few providers screen for FI, most believe screening for it is important (Schickendanz et al., 2019). The post-implementation results of our study demonstrate that the toolkit likely helped providers elevate the importance of FI screening and close the gap between belief and action. With the toolkit, more providers reported an increased frequency in their FI screening practices. Pre- and Post-implementation Surveys Our survey showed findings similar to those of other research articles about FI. For example, other researchers found that barriers to addressing FI include time constraints during visits, poor communication, a lack of resources, and a lack of knowledge of and confidence in discussing resources (Kopparapu et al., 2020; Schickendanz et al., 2019). In the surveys, providers commonly cited barriers such as a lack of training and time to screen patients and discuss resources, a lack of support staff to administer the screeners, and a lack of knowledge about FI resources (see Tables 3 and 4). Our data show that three providers who initially listed a “lack of knowledge about resources for food insecure patients” as a barrier did not list it again in the post-implementation survey, suggesting that the toolkit effectively increased provider knowledge. It is interesting, however, that while providers endorsed greater confidence and increased knowledge regarding FI, there was no change in the number of providers who cited a “lack of training” as a barrier. It is a possibility that the option “lack of training” was too broad 23 and that the providers may have thought that clinic staff (such as medical assistants) needed more training instead of themselves. This indicates a need to understand better the areas in which providers would like more training regarding food insecurity. Providers’ increased knowledge of food resources likely contributed to their improved comfort when discussing food insecurity; this matches research findings that clinicians want more resources and information on how best to address social needs (Tong et al., 2018). UICSL providers spoke about “how easy [the toolkit] was [for grabbing] resources specific to the patient’s location,” “how thorough the resource sheet was with up-to-date information and . . . details to help the patient decide what resource would be best for them,” and how “[the toolkit] was easy to follow” (see Table 6). Even though 100% of interviewed providers recommended the toolkit be fully implemented to address food insecurity, one provider (25%) said it would be very infeasible to implement it into standard practice. All providers reported no desire to change the toolkit’s contents. Although the clinic nurse was not included in the surveys, she noted the impact of the toolkit on clinic practice and sent an email with the following statement, The workflow has made an impact on the clinic addressing food insecurity in its population and is sustainable for the clinic. While every client is screened, we now have a process for determining which clients would also benefit from a dietician visit. The provided resource materials are effective and manageable with the clinic workflow. We have noted an increased uptake in providing these resources on positive screening. This project has married well to a new produce prescription program the center offers and was a great resource to 2 BSN students’ clinical presentation. Based on all feedback, the UICSL staff is satisfied with the toolkit’s contents and believes that the toolkit is useful but perhaps not feasible in standard practice. 24 In summary, while the data indicate a minimal impact on chart review measures, the toolkit improved providers’ consistency in caring for food-insecure patients and boosted their confidence in addressing FI resources. The qualitative data also demonstrate overall provider support for the future use of the toolkit. Limitations There are several limitations to this quality improvement project. First, due to scheduling difficulties, we were unable to meet with the psychiatric mental health nurse practitioner for inperson training on the toolkit. In order to counteract this limitation, we sent multiple emails reminding providers of the location of the food insecurity toolkit, resources, and important items to remember related to the intervention. Fortunately, nearly all patients who visit with the psychiatric mental health provider are first evaluated by family practice providers and so would have already been screened for FI and given resources, if needed. Second, UICSL leadership requested that their research assistants perform the pre- and post-implementation chart review. Because many screening results and dietitian referrals were mentioned only in the notes (and not via formal electronic referral), finding this data would require looking through over 500 charts, which was not feasible for the research assistants in the project’s short implementation timeline. It is difficult to give perspective to the six patients who received a diagnosis of FI when we cannot ascertain how many were screened for food insecurity initially. There also may have been more patients referred to the dietician than shown—leading to falsely low levels of dietician referrals. Third, as of January 2024, two of the four providers surveyed (50%) no longer work at UICSL, which recently hired a new medical director and an additional interim nurse practitioner. Staffing turnover could have influenced provider buy-in of the intervention. To address this 25 limitation, we discussed this project in detail with the new medical director during her training an onboarding to the health clinic. Fourth, our sample size is small (n=4), which makes the results less generalizable when paired with the other limitations of this study. However, this data could be applied to other small urban clinics that see patients from predominantly low socioeconomic backgrounds. These results could also provide meaningful insights into the barriers and facilitators to improving provider screening, follow up, and referral processes for food insecure patients. Conclusions This QI project had minimal improvement in the number of patients who received an ICD-10 code for FI and no improvement in the number of patients referred to the dietician. However, this project did have a large impact on provider practices surrounding FI. We found that after the toolkit implementation, providers screened more regularly for FI, used the twoquestion screening tool more frequently, and felt more confident in discussing positive screening results and FI resources with their patients. All providers endorsed the toolkit’s usability and feasibility and expressed their satisfaction. UICSL will continue the effort to improve FI-related processes through additional PDSA cycles. We have also turned the project over to UICSL’s quality improvement committee to address the above discrepancies and improve the screening and referral process for food insecure patients. The project results may be useful for other small clinics looking for ways to reduce FI or address other social determinants of health. Future efforts should focus on streamlining charting processes so that information regarding screening and referral of food insecure patients can be easily searched. Future efforts could also designate a clinical team member to train new employees on the FI toolkit, process 26 flowchart, and resources. Lastly, future efforts could also designate a staff member (such as a social worker) to help patients enroll in the Supplemental Nutrition Assistance Program. This staff member could also assist the dietician in signing patients up for the fresh produce program which has shown promise in increasing access to healthy food resources. Despite some challenges and the limited data present in this study, the data show great potential for providers to increase effective FI screening and care through a toolkit resource. 27 Acknowledgments The common saying “it takes a village to raise a child” could be changed in this case to “it takes a village to raise a nurse practitioner”—at least from my perspective. I am grateful for the many individuals who supported me on the journey to obtain my doctoral degree. I would first like to thank my family, especially my husband, Cameron, for being my support throughout my schooling. I’d also like to thank my mom and dad for their example of pursuing additional education, even when it is difficult. Next, I would like to thank the Urban Indian Center of Salt Lake for allowing me to complete my DNP project at their site. I would like to recognize my project chair, Dr. Jennifer Clifton, and my content expert, Danielle Puri, for their invaluable insight and guidance throughout this project. I’d also like to thank Leslie Crandall, the UICSL nurse who continually watched for patients who needed food resources and actively promoted the project among the clinical team. I want to especially thank UICSL’s dietician, Kristie Hinton, whose expertise and vision for improving food insecurity will help propel this project forward into a new and improved version. I am likewise grateful to Claudia Charles and Sarai Negrete Macias for helping retrieve the chart review data. Lastly, I would like to thank all the nurse practitioners and medical assistants who helped with this project’s creation and implementation. They are dedicated to providing the best quality care to their patients. 28 References Ashbrook, A., Essel, K., Montez, K., & Bennett-Tejes, D. (2021, January). Screen and intervene: A toolkit for pediatricians to address food insecurity. Food Research and Action Center. https://frac.org/aaptoolkit https://frac.org/aaptoolkit Cain, K. S., Meyer, S. C., Cummer, E., Patel, K. K., Casacchia, N. J., Montez, K., Palakshappa, D., & Brown, C. L. (2022). Association of food insecurity with mental health outcomes in parents and children. Academic Pediatrics, 22(7), 1105–1114. https://doi.org/10.1016/j.acap.2022.04.010 Caldwell, J. I., Palimaru, A., Cohen, D. A., Shah, D., & Kuo, T. (2023). Food insecurity screening in safety-net clinics in Los Angeles County: Lessons for post-pandemic planning. Journal of the American Board of Family Medicine, 36(2), 240–250. https://doi.org/10.3122/jabfm.2022.220175R2 Dong, L., D’Amico, E. J., Dickerson, D. L., Brown, R. A., Palimaru, A. I., Johnson, C. L., & Troxel, W. M. (2023). Food insecurity, sleep, and cardiometabolic risks in urban American Indian/Alaska Native youth. Sleep Health, 9(1), 4–10. https://doi.org/10.1016/j.sleh.2022.10.003 Feeding America. (2024). Hunger in America. https://www.feedingamerica.org/hunger-inamerica/food-insecurity Gundersen, C., & Ziliak, J. P. (2015). Food insecurity and health outcomes. Health Affairs, 34(11), 1830–1839. https://doi.org/10.1377/hlthaff.2015.0645 Hager, E. R., Quigg, A. M., Black, M. M., Coleman, S. M., Heeren, T., Rose-Jacobs, R., Cook, J. T., Ettinger de Cuba, S. A., Casey, P. H., Chilton, M., Cutts, D. B., Meyers, A. F., & Frank, D. A. (2010). Development and validity of a 2-item screen to identify families at 29 risk for food insecurity. Pediatrics, 126(1), e26–e32. https://doi.org/10.1542/peds.20093146 Institute for Healthcare Improvement. (n.d.). How to improve: Model for improvement. https://www.ihi.org/resources/how-to-improve Kolovos, S., Zavala, G. A., Leijen, A. S., Melgar-Quiñonez, H., & van Tulder, M. (2020). Household food insecurity is associated with depressive symptoms: Results from a Mexican population-based survey. Food Security, 12(2), 407–416. https://doi.org/10.1007/s12571-020-01014-1 Kopparapu, A., Sketas, G., & Swindle, T. (2020). Food insecurity in primary care: Patient perception and preferences. Family Medicine, 52(3), 202–205. https://doi.org/10.22454/FamMed.2020.964431 Move For Hunger. (n.d.). How hunger affects Native American communities. https://moveforhunger.org/native-americans-food-insecure Nikolaus, C. J., Benally, T., Maudrie, T., Henderson, A., Nelson, K., Lane, T., Segrest, V., Ferguson, G. L., Buchwald, D., Blue Bird Jernigan, V., & Sinclair, K. (2022). Food insecurity among American Indian and Alaska Native people: A scoping review to inform future research and policy needs. Advances in Nutrition, 13(5), 1566–1583. https://doi.org/10.1093/advances/nmac008 Schickedanz, A., Hamity, C., Rogers, A., Sharp, A. L., & Jackson, A. (2019). Clinician experiences and attitudes regarding screening for social determinants of health in a large integrated health system. Medical Care, 57(6), S197–S201. https://doi.org/10.1097/MLR.0000000000001051 30 Tong, S. T., Liaw, W. R., Kashiri, P. L., Pecsok, J., Rozman, J., Bazemore, A. W., & Krist, A. H. (2018). Clinician experiences with screening for social needs in primary care. The Journal of the American Board of Family Medicine, 31(3), 351–363. https://doi.org/10.3122/jabfm.2018.03.170419 Urban Indian Health Institute. (n.d.). Maternal and child health: WIC status. https://www.uihi.org/urban-indian-health/data-dashboard/ Johnson, C. (n.d.). Utah hunger statistics. Utah State University. https://extension.usu.edu/hsi/fsc-utah-hunger-statistics 31 Tables and Figures Table 1 Demographic information of nurse practitioner respondents Pre (N=4) Characteristic Specialty Family Pediatrics Acute Care Psych/Mental Health Gerontology Hours Worked per Week Full time Part time PRN Post (N=4) N % N % 3 0 0 1 0 75% 0% 0% 25% 0% 3 0 0 1 0 75% 0% 0% 25% 0% 0 3 1 0% 75% 25% 0 3 1 0% 75% 25% 32 Table 2 Current food insecurity practices Pre (N=4) Question Percent of patients you screen every 6 months? None Less than 50% Greater than 50% Always Do you use the Hunger Vital Sign screener? Yes No I don’t know what this is Post (N=4) N % N % 0 1 3 0 0% 25% 75% 0% 0 0 2 2 0% 0% 50% 50% 1 1 2 25% 25% 50% 4 0 0 100% 0% 0% If “No” or “I don’t know what this is,” do you assess food insecurity in another way? (N=3) I glean information during the visit 2 66.7% 0 Other 1 33.3% 0 Please specify Other I discuss the positive screener with the patient. Yes No Sometimes 0% 0% “Glean information and ask when appropriate” 3 0 1 75% 0% 25% 3 0 1 If “no” or “sometimes,” what are reasons you do not discuss the positive screener? (Select all that apply) I didn't see the screener until the patient had left 1 100% 1 Too many other topics to address 1 100% 0 Someone else is addressing the patient's food insecurity 1 100% 0 I forget to discuss the positive screener 0 0% 1 Not enough time in the visit 0 0% 0 I didn't know I needed to discuss the screening result 0 0% 0 Other 0 0% 0 75% 0% 25% 100% 0% 0% 100% 0% 0% 0% If my patient screens positive, I provide food insecurity resources. Yes 4 No 0 Sometimes 0 100% 0% 0% 4 0 0 100% 0% 0% Which resources do you provide? (Select all that apply) UICSL’s printed list of food resources 3 Other 1 75% 25% 4 1 100% 25% 33 Please specify Other “Food bank” “The updated food resources provided by Emily” 34 Table 3 Barriers and facilitators to screening for and addressing food insecurity Pre (N=4) % Question N Barriers to screening for food insecurity (Select all that apply) 50% Lack of training 2 25% Lack of time to screen 1 Lack of support staff to administer screener 0 0% Lack of comfort in asking about food insecurity 0 0% Lack of comfort in addressing food insecurity 1 25% 25% Other 1 Please specify Other “Resources for food insecurity” Post (N=4) N % 2 3 50% 75% 3 75% 0 0% 0 0% 25% 1 “Patients might have other acute concerns that need to be addressed first” Barriers to addressing food insecurity during visit (Select all that apply) Lack of time during visit to discuss screening results 0 0% 2 50% 0% 25% Lack of time to provide resources 0 1 Lack of resources to address food insecurity once it is identified 1 25% 1 25% 0% 0% Lack of knowledge about food insecurity 0 0 Lack of knowledge about resources for food insecure patients 3 75% 0 0% 25% 25% Other 1 1 “N/A; we have a handout “Sometimes not given forms Please specify other for food resources” during check-in” Which would increase access to food insecurity resources? (Select all that apply) Providing a page of resources in exam rooms 4 100% 2 Providing a resource handout at the end of the visit 0 0% 3 More time to discuss resources with patients 0 0% 2 More training on addressing food insecurity with patients 3 75% 1 0% Other 0 2 Please specify other 50% 75% 50% 25% 50% “I think UICSL is doing a good job addressing food insecurity. Increasing knowledge of resources in the greater AI/AN community would be 35 helpful, an education campaign could reach a wider audience” Table 4 Knowledge and comfort level with food insecurity N Pre (N=4) % 3 3 2 3 3 75% 75% 50% 75% 75% 4 4 4 4 3 100% 100% 100% 100% 75% I know what to do if my patient screens positive for food insecurity. Disagree Somewhat agree Agree 0 4 0 0% 100% 0% 0 0 4 0% 0% 100% Knowledge Check: When should your refer a patient to the dietician for food insecurity? (Select all that apply) Patients with hypertension Patients with diabetes Patients with obesity Anyone who screens positive for food insecurity 0 0 0 4 0% 0% 0% 100% 2 2 2 4 50% 50% 50% 100% I am familiar with clinic resources to address food insecurity. Disagree Somewhat agree Agree 0 2 2 0% 50% 50% 0 1 3 0% 25% 75% Comfort level with interpreting positive screener Not at all comfortable Somewhat comfortable Comfortable 0 2 2 0% 50% 50% 0 1 3 0% 25% 75% Comfort level with discussing a positive screener with patients Not at all comfortable Somewhat comfortable Comfortable 0 2 2 0% 50% 50% 0 0 4 0% 0% 100% Question Knowledge Check: Which of the following are health implications of food insecurity? (Select all that apply) Depression Elevated BMI Elevated Blood Pressure Greater sleep disturbance Substance abuse Post (N=4) % N 36 Comfort level with discussing food insecurity resources with patients Not at all comfortable Somewhat comfortable Comfortable 0% 75% 25% 0 3 1 0 0 4 Table 5 Accuracy of screener, feasibility, and provider satisfaction Question Frequency (N=4) % N Do you think the Hunger Vital Sign screener identifies food insecure patients? Yes Sometimes No 4 0 0 100% 0% 0% Based on your experience, how feasible is it to implement the toolkit into standard practice? Very feasible Somewhat feasible Neutral Somewhat infeasible Very infeasible 4 0 0 0 0 100% 0% 0% 0% 0% Would you recommend this toolkit be used in clinic practice to address food insecurity? Yes No 4 0 100% 0% 0% 0% 100% 37 Table 6 Provider commentary about the toolkit Question What did you like about the food insecurity toolkit? Frequency (N=4) N % 1 25% Theme Convenience of resources Quote(s) “How easy it was to grab resources specific to the patients location” Thorough and accurate resources “How thorough the resource sheet was with up-to-date information & included details to help the patient decide what resource would be best for them” 1 25% Fast documentation “I like the dot phrases; [they] made it quick to document about patient food insecurity treatment” 1 25% Ease of use “It was easy to follow” 1 25% Increased resources “Provides options” 1 25% What would you change about the toolkit? No desire to change contents of toolkit “N/A” or “nothing currently” 4 100% Briefly describe any other information that you think would be valuable to know for this QI project No other information to add “Nothing to add” “Unsure” 3 75% Liked resources and dietician consult “I really like that the food resources handouts are available for providers to give out to patients and families if they screen positive. It is also nice to have dietitian consult with patients who screen positive.” 1 25% 38 Appendix A Pre Implementation Survey 39 40 41 42 Appendix B Post Implementation Survey 43 44 45 46 Appendix C Food Insecurity Toolkit and Resources 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 Appendix D University of Utah IRB Exemption 78 79 80 81 Appendix E