Screening and Referral Care Delivery Services and Unmet Health-Related Social Needs: A Systematic Review

Introduction Unmet health-related social needs contribute to high patient morbidity and poor population health. A potential solution to improve population health includes the adoption of care delivery models that alleviate unmet needs through screening, referral, and tracking of patients in health care settings, yet the overall impact of such models has remained unexplored. This review addresses an existing gap in the literature regarding the effectiveness of these models and assesses their overall impact on outcomes related to experience of care, population health, and costs. Methods In March 2020, we searched for peer-reviewed articles published in PubMed over the past 10 years. Studies were included if they 1) used a screening tool for identifying unmet health-related social needs in a health care setting, 2) referred patients with positive screens to appropriate resources for addressing identified unmet health-related social needs, and 3) reported any outcomes related to patient experience of care, population health, or cost. Results Of 1,821 articles identified, 35 met the inclusion criteria. All but 1 study demonstrated a tendency toward high risk of bias. Improved outcomes related to experience of care (eg, change in social needs, patient satisfaction, n = 34), population health (eg, diet quality, blood cholesterol levels, n = 7), and cost (eg, program costs, cost-effectiveness, n = 3) were reported. In some studies (n = 5), improved outcomes were found among participants who received direct referrals or additional assistance with indirect referrals compared with those who received indirect referrals only. Conclusion Effective collaborations between health care organizations and community-based organizations are essential to facilitate necessary patient connection to resources for addressing their unmet needs. Although evidence indicated a positive influence of screening and referral programs on outcomes related to experience of care and population health, no definitive conclusions can be made on overall impact because of the potentially high risk of bias in the included studies.


Introduction
Unmet health-related social needs contribute to high patient morbidity and poor population health. A potential solution to improve population health includes the adoption of care delivery models that alleviate unmet needs through screening, referral, and tracking of patients in health care settings, yet the overall impact of such models has remained unexplored. This review addresses an existing gap in the literature regarding the effectiveness of these models and assesses their overall impact on outcomes related to experience of care, population health, and costs.

Methods
In March 2020, we searched for peer-reviewed articles published in PubMed over the past 10 years. Studies were included if they 1) used a screening tool for identifying unmet health-related social needs in a health care setting, 2) referred patients with positive screens to appropriate resources for addressing identified unmet health-related social needs, and 3) reported any outcomes related to patient experience of care, population health, or cost.

Introduction
Up to 80% of health outcomes can be attributed to social determinants of health (SDOH), the conditions in which we grow, live, and work (1,2). Adverse SDOH include food insecurity, housing instability, unemployment, and other unmet health-related social needs (3), which often contribute to negative health outcomes, in-cluding an increased risk for diabetes, hypertension, and heart disease (4)(5)(6)(7). Recently, higher unemployment rates and changes in health insurance coverage due to the ongoing COVID-19 pandemic have further compromised health care access and increased the number of people with unmet needs (8,9).
Health care organizations (HCOs) offer a natural setting for integration of clinical care, public health, and community-based services (10,11). The Centers for Medicare and Medicaid Services (CMS) has recognized the potential value in leveraging the infrastructure of HCOs for addressing health-related social needs. As part of the Accountable Health Communities initiative, CMS provides incentives for HCOs to consider solutions that address unmet needs by potentially improving population health and reducing system costs to drive overall performance (12). One common approach to the screening and referral-based care delivery model includes the identification of unmet needs through a screening questionnaire, followed by a referral component that addresses or mitigates unmet needs through referrals to appropriate resources, and subsequently evaluates the impact of this screening and referral program (12)(13)(14) (Figure 1). Although implementation of such screening and referral-based programs has increased in recent years (14), we found no review that summarized evidence on the impact of these programs on care outcomes. Therefore, in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (15), we answered the following population, intervention, comparison, and outcomes question (PICO): What is the impact of screening and referral programs targeting unmet healthrelated social needs in health care settings on outcomes related to experience of care, population health, and costs?

Data sources
Because CMS only started implementing screening and referral-based care delivery models in 2016 (12), we searched PubMed to identify relevant peer-reviewed articles published over the past 10 years as of March 2020 to capture results from any pilot and demonstration projects before and after this time frame. Search terms were derived with the help of a subject librarian and included the following terms: ("social determinants of health" OR "social determinants" OR "social needs" OR food insecurity OR housing OR transportation OR employment) AND (screening OR needs assessment OR test) AND (referrals OR collaboration OR address needs) AND ("primary care" OR primary health care OR health services) NOT (biological OR psychology OR mental health). Our search terms did not contain an exhaustive list of all social determinants described in the literature. Specific healthrelated social needs (eg, food insecurity, housing) included in the search indicate the needs commonly addressed by current screening and referral programs. Additionally, we scanned the bibliographies of all articles that met the inclusion criteria and other literature reviews (16,17). To maximize our final article yield, older studies published before January 1, 2010, obtained from bibliographies, were included if they met the inclusion criteria.

Study selection
Articles were included if they were written in English and described an intervention in a health care setting that 1) used a screening tool to identify unmet health-related social needs, 2) referred screened patients with positive results (or positive screens) to resources offering assistance (eg, on-site provision of food or referral to a food bank), and 3) reported any care outcomes resulting from the screening and referral components described in 1) and 2), beginning with program recruitment or referral uptake. After the study selection, all outcomes were categorized into experience of care, population health, and cost-related based on the Institute for Healthcare Improvement (IHI) Triple Aim framework (18), which targets 3 dimensions for optimizing performance in HCOs: 1) improving the patient experience of care through quality and satisfaction; 2) improving health of the patient population, and 3) reducing the per capita cost of care. related social needs was performed or if solely process-related, descriptive screening outcomes (eg, number of screenings, number of referrals) were reported.
Screening of titles and abstracts was carried out by 2 reviewers (E.R.E., S.P.) using Microsoft Excel (Microsoft Corporation). Once relevant articles were independently identified, each reviewer completed a full-text review of the selected articles. We planned to resolve discrepancies during the article selection process by using consensus among the authors (E.R.E., S.P., C.B.), but no discrepancies occurred.

Data extraction
From each eligible article, we extracted the following: author name(s), year of publication, place of origin, health care setting(s), target population, study design, sample size, screening tool used, targeted unmet health-related social need(s), referral approach, referral site, outcome(s) assessed, and study results.

Risk of bias assessment and data analysis
Valid and complementary assessment tools for randomized (19) and nonrandomized studies (20) were used to examine risk of bias. For randomized clinical trials, we used the Cochrane tool (19) to make critical assessments (low risk, high risk, and unclear risk) of included studies in 6 domains: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and "other sources of bias." For nonrandomized studies, we made similar critical assessments (low risk, high risk, and unclear risk) using the RoBANS tool (Risk of Bias Assessment for Nonrandomized Studies) (20) for a slightly different set of 6 domains: selection of participants, confounding variables, measurement of exposure, blinding of outcomes assessment, incomplete outcome data, and selective outcome reporting. For both randomized and nonrandomized studies, the final assessment within and across studies was based on the responses to individual domains.
A qualitative synthesis of results across studies was performed. Meta-analysis was not performed because of heterogeneity in the study populations, interventions, and outcomes of included studies.

Results
A total of 1,821 articles were identified from the PubMed database search ( Figure 2). After applying the PICO question and our inclusion criteria, 42 articles were selected for full-text review, of which 18 met the inclusion criteria. An additional 17 articles were included from bibliographies, bringing the total to 35 articles in the final review. Seven (20%) studies were randomized control tri-als, 6 (17%) were observational studies that compared outcomes within the intervention group to a nonintervention comparison group, and the rest examined outcomes within an intervention group only (n = 22; 63%).

Risk of bias assessment
Randomized studies demonstrated a potentially high risk (n = 6) or unclear risk of bias (n = 1) ( Table 1). Insufficient or lack of information about blinding of participants, personnel, or outcomes indicated that potential selection, performance, and detection biases were present. Additionally, all nonrandomized studies (n = 28) were assessed as having a potentially high risk of bias ( Table  2). The most common domains demonstrating high risk were blinding of outcomes assessment (n = 28), confounding variables (n = 19), and participant selection (n = 13).

Settings, populations, and unmet health-related social needs
All included studies (n = 35) had a screening and referral component and originated in the US (Tables 1 and 2

Referral component
Studies featured HCOs that partnered with various communitybased organizations (CBOs) or expanded their internal resources to include assistance programs that addressed immediate unmet needs. Five studies reported on providing on-site social assistance services including CBO eligibility applications (32,45,49,53) and distribution of food supplies (30,53). Although descriptions of community collaboration were sparse, referral sites included CBOs such as food banks, nutrition programs, intimate partner violence agencies, housing programs, and early childhood education programs. Additionally, we found 3 referral approaches: indirect, direct, and warm handoff. In an indirect referral, health care providers simply hand over information about the referral site(s) to the patient (eg, distribute a list of local food banks and their contact information to patients who have a positive screen for food insecurity). A direct referral approach is when the HCO directly forwards a patient's contact information to a referral site contingent on the patient's consent and is often administered through health information exchange tools. The referral site then follows up with the patient to assist in the patient's application or enrollment in programs to alleviate unmet needs. A warm handoff is when an on-site intermediary person in the HCO (eg, community health worker, social worker) assists patients who have a positive screen with connecting to the referral site.

Qualitative synthesis of outcomes
Most studies (n = 25) (21,23-25,27,29,32-41,43-45,47, 49,50,53-55) reported only outcomes related to experience of care (Table 3), which included referral uptake (ie, participants who connected with or used necessary resources expressed as a percentage or ratio of all participants who had a positive screen or consented to a referral) and patient-reported outcomes (ie, selfreported changes in social needs, diet, health, and patient satisfaction). Other studies reported population health outcomes (n = 7) (22,26,30,31,46,48,52), which included changes in indicators of patient health such as systolic and diastolic blood pressure, glycosylated hemoglobin levels (HbA 1c ), body mass index (BMI), low-density lipoprotein cholesterol, medication adherence, appointment adherence, violent injury recidivism rates, and preventive care outcomes (ie, completing lead tests, developmental screens, infant immunization schedules, or well-infant visit sets). Only 3 studies reported cost-related outcomes (28,42,51), including evaluation of program costs, changes in health care utilization, or cost-effectiveness.

Experience of care outcomes
Referral uptake. Data on participants or participating families who connected with or used necessary resources were expressed either as a percentage of all participants who had a positive screen or as a percentage of those who consented to/received a referral (referral uptake). Although most studies (n = 30) reported some information on patient connection to the referral site, the reported results were highly contextual and varied widely from study to study. For example, some studies reported connection rates as low as 3% (47) while others reported rates as high as 75% (54).
Patient-reported outcomes. Nine studies (21,24,28,36, 38,40,41,44,48) reported positive findings on patient-reported outcomes. For example, 1 study interviewing patients with unmet needs (41) reported participants being able to make concrete changes in their lives as a result of screening and referral, including resolving immediate social needs, a healthier diet, or physical and mental/emotional benefits; another study (28) found that participants' self-reported food insecurity decreased by 94.1%.
Three studies (36,38,48) that investigated patient satisfaction reported positive feedback on referral sites and program tools. More than 80% of patients found their referral sites helpful in 1 study (36), and more than 90% of parents enrolled in a communitybased resource reported being "very" or "somewhat" satisfied in a different study (38). Similarly, participants with diabetes in another study reported high acceptability across multiple survey items in their program's resource tool (eg, 93% "learned a lot," 98% found "topics relevant") (48).

Health outcomes
Seven (22,26,30,31,46,48,52) studies that examined outcomes related to population health found some positive changes in health indicators. After addressing income insecurity, 7% fewer patients (P = .03) reported skipping doses of medicines because of financial concerns (48). Another study found a drop in the violent injury recidivism rate from an initial 16% to 4.5% by the end of the program (52). Other studies found improved preventive care outcomes, including faster completion of lead tests, developmental screens, infant immunization schedules among participants (77% vs 63% completed by age 7 months, P = .002 and 88% vs 78% by age 8 months, P = .008) (26), and greater likelihood of completing a full set of well-infant visits by 14 months (42% vs 28.7%; P < .001) (30). Changes reported in intervention participants enrolled in screening and referral programs compared with those who did not receive a referral included improvements in systolic blood pressure and diastolic blood pressure trends during pregnancy (46) (P = .004), and a modest differential change in systolic blood pressure of −1.2 mm Hg (95% CI, −2.1 to −0.4), diastolic blood pressure of −1.0 mm Hg (95% CI, −1.5 to −0.5), and improved lowdensity lipoprotein cholesterol (differential change −3.7 mg/dL; 95% CI, −6.7 to −0.6) among participants with diabetes (31).

Cost-related outcomes
Only 3 studies examined cost-related outcomes. One study targeting food insecurity with a food prescription program (28) found program costs to be $12.20 per person per redemption. Additionally, participants reported an average $57 savings per week on grocery bills. Another study targeting multiple social needs (51) reported a decreased likelihood of health care utilization among the intervention group compared with the control group. Also, a cost-effectiveness study (42) of a hospital-based violent injury prevention intervention (52) yielded 25.6 quality-adjusted life years (QALYs, a standardized measure of disease burden in costeffectiveness evaluations that typically combines both survival and health-related quality of life to guide decisions on the distribution of limited health care resources) versus 25.3 for the non violent injury prevention group, with net costs of $5,892 per patient versus $5,923 for the non violent injury prevention group.

Impact of referral approach on all outcomes
Studies comparing direct (27,54) to indirect referral reported greater referral uptakes with direct referral. One study (27) found that intervention participants receiving direct referrals reported a greater percentage of children who connected with an education resource (41% vs 18%) and actively attended the development program (25% vs 11%) than intervention participants who received an indirect referral. Another study (54) reported patient connection to referral sites increasing from 5% to 75% when the approach was changed from indirect to direct referral.
Some studies compared indirect referrals paired with additional services (eg, on-site assistance) in the intervention to a control group that received indirect referrals only. One such study (32) found a similar percentage of participants using the referral resource (21.4% vs 17.4%; P = .43) as the control group. However, a greater percentage of intervention participants than control participants connected with the on-site advocate (32.8% vs 4.4%; P < .001). Similarly, another study (23) reported that intervention participants had greater odds than control participants of being employed or enrolled in a job training program (aOR = 44.4), receiving childcare support (aOR = 6.3) and fuel assistance (aOR = 11.9), and lower odds of being in a homeless shelter (aOR = 0.2).
Two studies (22,24) compared outcomes in patients receiving a warm handoff with patients who received an indirect referral. The intervention group receiving warm handoff (22) had decreased HbA 1c levels (mean differences of −3.09 vs −1.66; P = .012), improved STC (Starting the Conversation)-Diet scale (62) (mean differences of 2.47 vs 0.06; P = .001), but no difference in BMI (mean differences of -0.17 vs 0.84; P = .43) compared with control participants. Similarly, intervention participants in another PREVENTING CHRONIC DISEASE www.cdc.gov/pcd/issues/2021/20_0569.htm • Centers for Disease Control and Prevention study (24) reported fewer unmet social needs (mean change of −0.39 vs 0.22; P < .001) and greater improvement in their child's health than control participants (mean change of −0.36 vs 0.12; P < .001).

Discussion
The body of evidence on the relationship between unmet healthrelated social needs and poor patient outcomes has continued to grow in recent years. In response, screening and referral programs have expanded to mitigate unmet health-related social needs among patients in health care settings (12,14).
This review found 35 studies on screening and referral delivery services that reported outcomes related to patients' experience of care, population health, and cost. The delivery service targeted patients with different chronic conditions and demographic characteristics, aiming to mitigate different health-related unmet needs.
We found some indication that screening and referral programs had a generally positive impact on outcomes related to experience of care, population health, and cost. Patient connection to referral sites and patient-reported outcomes such as self-reported diet intake, resolution of unmet health-related social needs, overall wellbeing, and patient satisfaction increased. Indicators of health such as blood pressure trends, low-density lipoprotein cholesterol, and medication adherence improved. Additionally, results indicated an improvement in QALYs, decreased likelihood of health care utilization, and modest savings associated with these programs.
Overall, included studies revealed a high risk of bias for elements related to study design and evaluation. Thus, we were unable to draw any definitive conclusions about the impact of screening and delivery services on any outcome.
The linkage of patients to resources seemed to be influenced by the type of referral and the degree of navigation within the HCOs and collaboration between the HCOs and the CBOs involved. Results suggested that patients were more successful in connecting with resources when partnered CBOs are more directly involved (ie, direct referral), or if referral efforts were made through on-site intermediaries such as community health workers to direct patients in contacting or applying to referral sites (ie, warm handoff).
Studies have indicated (37,41,47) that referral uptake was influenced by accessibility to referral sites, including patient eligibility criteria and intensity of time and labor required to access resources. For instance, the application process for SNAP (Supplemental Nutrition Assistance Program) is lengthy and complex (63). Such barriers, as speculated in the literature (64), can explain why, despite participating in screening and referral pro-grams, patients can have difficulties in accessing some resources. In short, the degree of referral uptake can be subject to various program characteristics including referral approach, on-site assistance, and accessibility of referral sites.
This review serves as a call to action for policy makers, advocates, and care providers to facilitate screening and referral delivery services through strong collaborations among health care, public health, and community sectors to address unmet health-related and social needs. Such programs can offer a comprehensive solution for health care administrators and insurers looking to improve the health of their patient population, reduce system costs, and optimize overall performance by addressing social determinants of health in their patient populations and delivering high-quality person-centered care. Research with stronger study designs and rigorous evaluation methodologies is needed to establish a strong evidence base of the effectiveness of screening and referral delivery services. Future studies can further explore social-needs screening in mental/behavioral health settings that target individual behavior-related determinants of health (eg, smoking, alcohol abuse) along with social determinants.
To our knowledge, this review is the first study to provide an overview of the impact of screening and referral programs on outcomes related to experience of patient care, population health, and costs. Although our search for articles was performed in accordance with PRISMA guidelines for a systematic review (15), our study was exploratory. We limited our search to peer-reviewed articles and 1 database, which might have excluded other results reported in the gray literature or in other databases.
In summary, literature on the impact of screening and referral programs in HCOs had a tendency toward high risk of bias. Although the evidence indicated promising changes in patient connection to resources, patient-reported outcomes, patient satisfaction, and some health indicators, no definitive conclusions could be made about the impact of such programs on outcomes related to experience of care, population health, and cost. This study can inform public health professionals, administrators, and policy makers about the impact of implementing screening and referral care delivery services in health care settings, paving the way for the expansion of such programs to improve population health.        a Reported as the total number of participants who underwent screening. If the study did not report number of screenings, the number of referrals was reported as the sample size. b A referral approach in which health care providers simply hand over information about relevant referral sites to the patient (eg, a list of local food banks and their contact information). c Additional on-site services may include assistance with applying to community-based resources or connection to other resources through a helpdesk, and/or onsite provision of supplies. d Refers to participants who connected to necessary resources expressed as a percentage or ratio of all participants who had a positive screen or those who consented to a referral. e A referral approach that requires the patient's consent to forward their contact information to the corresponding internal or external resource. The referral site then directly contacts the patient. f A referral approach in which patients are introduced to an on-site intermediary person in the health care organization (eg, community health worker, case manager) who works to connect them to referral sites.
(continued on next page) PREVENTING CHRONIC DISEASE   84% (475 of 562) of women with a positive screen consented to meeting with an on-site advocate, of whom 54% (258 of 475) then agreed to meet with a case manager. At follow-up, 24% (127 of the 528) of women reported they no longer believed they were at risk for violence from their abuser.  32% (19 of 59) of parents that used the onsite Help Desk reported enrolling in at least 1 community program. 21% (4 of the 19) enrolled in ≥2 community programs. More than 90% of parents who enrolled in a community resource were very or somewhat satisfied.
Abbreviations: BMI, body mass index; CBO, community-based organization; DBP, diastolic blood pressure in mm Hg; EHR, electronic health record; HbA 1c , hemoglobin A 1c ; FAMNEEDS, Family Needs Screening Program; HeLP Program, Health Law Partnership; IPV, intimate partner violence; KIND, Keeping Infants Nourished and Developing; LDL-C, low-density lipoprotein cholesterol in mg/dL; NRT, nicotine replacement therapy; QALYs, quality-adjusted life years; SBP, systolic blood pressure in mmHg; SNAP, Supplemental Nutrition Assistance Program; USDA US HFSS, US Department of Agriculture US Household Food Security Survey; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children. a Reported as the total number of participants who underwent screening. If the study did not report number of screenings, the number of referrals was reported as the sample size. b A referral approach in which health care providers simply hand over information about relevant referral sites to the patient (eg, a list of local food banks and their contact information). c Additional on-site services may include assistance with applying to community-based resources or connection to other resources through a helpdesk, and/or onsite provision of supplies. d Refers to participants who connected to necessary resources expressed as a percentage or ratio of all participants who had a positive screen or those who consented to a referral. e A referral approach that requires the patient's consent to forward their contact information to the corresponding internal or external resource. The referral site then directly contacts the patient. f A referral approach in which patients are introduced to an on-site intermediary person in the health care organization (eg, community health worker, case manager) who works to connect them to referral sites.   Postintervention, the intervention group had fewer families that filed child protective services reports (13.3% vs 19.2%; P = .03), and fewer instances of possible medical neglect including nonadherence (4.6% vs 8.4%; P = .05) and delayed immunizations (3.3% vs 9.6%; P = .002) than the control group. Control group had more parentreported harsh punishment (P = .04). Intervention mothers reported less psychological aggression initially and 12 months later (initial effect size P = .006; 12month effect size P = .047) and fewer minor physical assaults (initial effect size P = .02; 12-month effect size P = .04) than control.
Abbreviations: BMI, body mass index; CBO, community-based organization; DBP, diastolic blood pressure in mm Hg; EHR, electronic health record; HbA 1c , hemoglobin A 1c ; FAMNEEDS, Family Needs Screening Program; HeLP Program, Health Law Partnership; IPV, intimate partner violence; KIND, Keeping Infants Nourished and Developing; LDL-C, low-density lipoprotein cholesterol in mg/dL; NRT, nicotine replacement therapy; QALYs, quality-adjusted life years; SBP, systolic blood pressure in mmHg; SNAP, Supplemental Nutrition Assistance Program; USDA US HFSS, US Department of Agriculture US Household Food Security Survey; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children. a Reported as the total number of participants who underwent screening. If the study did not report number of screenings, the number of referrals was reported as the sample size. b A referral approach in which health care providers simply hand over information about relevant referral sites to the patient (eg, a list of local food banks and their contact information). c Additional on-site services may include assistance with applying to community-based resources or connection to other resources through a helpdesk, and/or onsite provision of supplies. d Refers to participants who connected to necessary resources expressed as a percentage or ratio of all participants who had a positive screen or those who consented to a referral. e A referral approach that requires the patient's consent to forward their contact information to the corresponding internal or external resource. The referral site then directly contacts the patient. f A referral approach in which patients are introduced to an on-site intermediary person in the health care organization (eg, community health worker, case manager) who works to connect them to referral sites.   Experience of care: More participants were using low-cost resources at 2-month followup compared with baseline, such as online diabetes education (40% vs 29%; P = .05) and assistance programs related to blood glucose supplies (40% vs 16%; P = .03). Participants found the resource tool highly acceptable across 15 indicators (eg, 93% Abbreviations: BMI, body mass index; CBO, community-based organization; DBP, diastolic blood pressure in mm Hg; EHR, electronic health record; HbA 1c , hemoglobin A 1c ; FAMNEEDS, Family Needs Screening Program; HeLP Program, Health Law Partnership; IPV, intimate partner violence; KIND, Keeping Infants Nourished and Developing; LDL-C, low-density lipoprotein cholesterol in mg/dL; NRT, nicotine replacement therapy; QALYs, quality-adjusted life years; SBP, systolic blood pressure in mmHg; SNAP, Supplemental Nutrition Assistance Program; USDA US HFSS, US Department of Agriculture US Household Food Security Survey; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children. a Reported as the total number of participants who underwent screening. If the study did not report number of screenings, the number of referrals was reported as the sample size. b A referral approach in which health care providers simply hand over information about relevant referral sites to the patient (eg, a list of local food banks and their contact information). c Additional on-site services may include assistance with applying to community-based resources or connection to other resources through a helpdesk, and/or onsite provision of supplies. d Refers to participants who connected to necessary resources expressed as a percentage or ratio of all participants who had a positive screen or those who consented to a referral. e A referral approach that requires the patient's consent to forward their contact information to the corresponding internal or external resource. The referral site then directly contacts the patient. f A referral approach in which patients are introduced to an on-site intermediary person in the health care organization (eg, community health worker, case manager) who works to connect them to referral sites.    contact information). c Additional on-site services may include assistance with applying to community-based resources or connection to other resources through a helpdesk, and/or onsite provision of supplies. d Refers to participants who connected to necessary resources expressed as a percentage or ratio of all participants who had a positive screen or those who consented to a referral. e A referral approach that requires the patient's consent to forward their contact information to the corresponding internal or external resource. The referral site then directly contacts the patient. f A referral approach in which patients are introduced to an on-site intermediary person in the health care organization (eg, community health worker, case manager) who works to connect them to referral sites.