Evaluating the Impact of a Pediatric Inpatient Social Care Program in a Community Hospital Free

Our social care program was implemented as a new standard of care at the community hospital, consisting of universal standardized social risk screening utilizing a structured tool and a social service referral system to facilitate family connections to community resources. Preintervention, patients’ social risks were probed informally through physicians’ social histories and optional questions throughout multiple nursing intake forms that were skipped if they were not thought to be relevant to the admission. Clinicians ordered social work consults ad hoc on the basis of their clinical judgment.

The specifications of our social care program were informed by our previous work, which engaged a stakeholder group of multidisciplinary clinicians (nurses, physicians, social workers, case managers) and parents of hospitalized children at the community hospital under study.^35,36  Building from the solicited clinician and parent preferences regarding inpatient social risk screening (eg, who performs screening, when during the hospitalization screening is performed), a smaller group of clinician and parent champions convened and rated potential implementation strategies on the basis of (1) fit within available resources, (2) fit within inpatient workflows, and (3) optimization of parent experiences. Through discussion and consensus with the stakeholder group, our implementation strategy was finalized consisting of person-to-person screening with parents of hospitalized children, performed by nurses during the admission intake process, and in-hospital social work referral to identify relevant resources for families who screen positively. Professional hospital interpreters would be used with families who preferred languages other than English. To facilitate in-hospital social work referrals, an automated prompt for a social work consult was built into the electronic health record (EHR), appearing in response to any positive screens. Clinicians could decline the consult order on the basis of clinical judgment or parent preferences. A best practice advisory was also built into the EHR, encouraging physicians to record identified social risks and subsequent assistance provided in their EHR documentation to facilitate ongoing care coordination.

We developed a social risk screening instrument consisting of 6 questions from clinically validated screening tools (Hunger Vital Sign, Housing Stability Vital Sign, Health Leads, Safe Environment for Every Kid [SEEK], and Protocol for Responding to and Assessing Patients’ Assets, Risks and Experiences [PRAPARE]) assessing 6 core social domains: food, housing, education, finances, transportation, and safety (Supplemental Information).^17,37–41  The 6 questions were drawn from a larger panel of existing questions available through the EHR. A multidisciplinary group of pediatricians, social workers, nurse managers, medical assistants, and representatives from the hospital’s Medicaid accountable care organization selected the 6 questions to prioritize social risks that directly affect children and are actionable in the inpatient environment while maintaining screening brevity for parents of hospitalized children. The screening instrument was thereby embedded into the community hospital’s EHR.

We used a retrospective cohort design. Our outcomes consisted of (1) program process measures and (2) preliminary effectiveness measures. Program process measures included (1) any social risk screening performed, (2) number of social domains (if any) screened, (3) comprehensive screening for ≥5 social domains (aligning with CMS recommendations), (4) screening for each of the individual social domains (eg, food, housing), (5) social work consultations completed, and (6) communication of social risks in discharge summaries. Preliminary effectiveness measures included (1) social risks identified in any domain (yes/no), (2) risks identified for each of the individual social domains, (3) timeliness of social work consultations, (4) community resource referral(s) made, and (5) hospital reutilization within 30 and 60 days of discharge.

Eligible subjects were patients <18 years old admitted to the pediatric inpatient unit at 1 university-affiliated community hospital between April 2021 and March 2022 (preimplementation cohort) and between April 2022 and March 2023 (postimplementation cohort), with no exclusions for non-English speakers. The hospital serves a racially diverse community with large Asian and Hispanic populations, with the majority publicly insured and nearly 25% living beneath the federal poverty line. The community hospital has ∼500 annual pediatric admissions, no pediatric ICU, and limited on-site subspecialty support. We excluded children who were in the custody of the Department of Children and Families or who were initially admitted to a different unit of the hospital (eg, a neonate admitted to the newborn nursery and later transitioned to the pediatric unit for more prolonged medical care) because parents of those children would not have experienced the admission intake procedures on the pediatric inpatient unit.

To determine the sample size required to demonstrate program preliminary effectiveness in the identification of social risks, we used a pilot chart review from the community hospital’s network which revealed that, without any standardized social risk screening, social risks were identified in 14% of pediatric inpatient charts. With the implementation of a pediatric inpatient social care program, Fortin et al identified social risks for 34% of patients.³²  Using a conservative approach with a smaller effect size (risks identified for only 25% postimplementation, α .05 and 90% power) we estimated that each study cohort would require at least 190 patients.

For all eligible patients, sociodemographic and clinical characteristics were extracted from the EHR in an automated fashion. For each cohort, we then used statistical software to create a random subset of ∼240 patients (∼20 patients/month). We then performed manual chart review, documenting any social risk screening performed with parents (defined as 1 or more standardized questions asked about social risks) and risks identified within 8 social domains (childcare, education, finances, food, housing, safety, transportation, and utilities), social work consults completed, documentation of social risks or lack of social risks in discharge summaries, time to social work consults (defined as the time in hours from admission to social work consult documentation), community resource referrals made, and hospital reutilization including all-cause emergency department revisits within 30 and 60 days of discharge, and all-cause hospital readmission within 30 and 60 days of discharge. With the increasing national focus on health-related social risk factors, we assessed childcare and utility domains not directly probed by the pediatric screening tool to look for potential secular trends in social risk identification that would have occurred independent of our social care program implementation. Two research assistants first independently abstracted a subset of patient charts while the Principal Investigator (PI) independently abstracted the same charts to ensure the reliability and accuracy of the data abstraction process, after which data abstraction was completed independently by the 2 research assistants. All data were recorded in a standardized REDCap data abstraction form. If, during manual data extraction, subjects were found to meet any exclusion criteria, they were subsequently removed from the analytic set.

We used descriptive statistics to describe the sociodemographic and clinical characteristics and rates or timeliness, as appropriate, of our program process and preliminary effectiveness outcomes. We compared outcomes between the pre- and post-implementation cohorts using χ-square or Wilcoxon rank tests, as indicated. To address potential confounding between the pre- and post-implementation period and the identification of any social risks, we created a multivariable model controlling for insurance type, race, ethnicity, primary language, and length of stay (LOS). We included the social constructs of race and ethnicity in our model because historically marginalized racial and ethnic groups are more likely to experience adverse social determinants of health when compared with nonmarginalized groups.^2,4,42 

Because the duration of time spent with families during hospitalizations has been cited as a potential facilitator for inpatient social care programs, we also used descriptive statistics among the post-implementation group to describe rates of comprehensive screening, social work consult completion, social risks identified, and community resource referrals made, stratified by LOS.^27,35  All statistical analyses were performed by using R version 4.2.1. This study received an exemption from the hospital’s institutional review board.

The final analytic dataset consisted of 232 patients in the pre-implementation cohort and 218 patients in the post-implementation cohort. Sociodemographic and clinical characteristics were similar between the groups, with nearly half of the patients being of white race and the majority being English-speaking and publicly insured (Table 1). Most children were admitted to a medical service and had short lengths of stay (0–1 days).

Any screening was performed more frequently preimplementation compared with postimplementation (Table 2), reflective of 1 domestic violence question asked nearly universally in the preimplementation timeframe. Screening for social domains beyond interpersonal safety was rare preimplementation, with hospital staff completing screening for a median of 1 social domain (25%- 75% 1–1, range 0–2) compared with a median of 6 postimplementation (P < .001). Although the majority of patients in the postimplementation group were screened comprehensively for ≥5 social domains, aligning with CMS recommendations, no patients in the preimplementation group had comprehensive screening. For each of the individual social domains except safety, screening rates were higher in the postimplementation group compared with the preimplementation group (P < .001).

Social work providers completed consults 2 times more frequently for the postimplementation group compared with the preimplementation group (13.8% vs 5.6% [P < .01], respectively). Written communication regarding social risks (or lack of social risks) was also greater postimplementation, with documentation present in discharge summaries for Primary Care Providers (PCPs) of nearly half of the postimplementation group compared with <10% of the preimplementation group (P < .001).

Patients in the postimplementation group had more than double the rate of any risks identified compared with the pre-implementation group (17.4% vs 7.8% [P < .01], respectively; Table 2). Similarly, after controlling for confounding in multivariable analysis, patients in the postimplementation group had 2.9 times the odds of any social risks identified compared with those in the preimplementation group (Table 3).

For each individual social domain except education, risks were identified more frequently for postimplementation patients compared with preimplementation patients, with differences reaching statistical significance for food, financial, and transportation risks (Table 2). For social domains not directly probed by the pediatric screening tool (childcare and utilities), risks were identified infrequently, and there were no statistically significant differences in risks identified between the groups.

Social work consultations were completed earlier in the course of inpatient care for the post-implementation group, with a median time to completion of 19 hours compared with 25 hours in the preimplementation group (P = .03, Table 2). The postimplementation group also had double the rate of community resource referrals (proportionate to the higher rate of social risks identified) compared with the pre-implementation group; however, this difference did not reach statistical significance. In both groups, nearly half of those with social risks received community resource referrals.

Hospital readmission and ED revisit rates were low overall, and there were no statistically significant differences between the pre and postimplementation groups.

Among the postimplementation group, rates of comprehensive screening, social work consult completion, social risks identified, and community resource referrals all increased as the duration of time spent in the hospital increased. Comparing patients with short LOS (0–1 day) versus intermediate LOS (2–5 days) versus long LOS (>5 days), comprehensive screening rates increased from 60.1% to 84.4% to 100%, social work consult rates increased from 7.2% to 23.4% to 66.7%, social risks identified increased from 11.6% to 24.7% to 100%, and community resource referral rates increased from 3.6% to 13% to 33.3% (Supplemental Table 4).

Although previous pediatric inpatient social risk screening studies have been conducted in children’s hospitals, we aimed to evaluate the impact of a standardized social risk screening and referral program within a community hospital setting as a new standard of care.^{24–28,31,32}  Community hospitals are the most common site of hospitalization for children, but they are typically relatively resource-constrained compared with freestanding children’s hospitals or pediatric services within large academic medical centers.³⁴  Our study reveals that within routine medical care, the implementation of a stakeholder-informed social care program in a community hospital may help to identify more families who could benefit from social resources, increase family connections with hospital social service providers and community resource referrals, and improve the timeliness of social work consultations and discharge communication of social risks for PCPs. In addition to these benefits, our study also illustrates some challenges associated with the implementation of social care in a community hospital, with shorter lengths of stay presenting a potential barrier to the completion of program processes and community resource referrals. Our findings may provide guidance for other community hospitals to engage stakeholders to develop their own social care programs tailored to their specific institutional contexts to leverage facilitators while avoiding barriers to optimize the likelihood of successful implementation and support for families with social risks.

Our study provides support for some findings demonstrated within children’s hospitals, with structured inpatient social care programs increasing social risks identified, connections to in-hospital social service providers, and community resource referrals.^{24,25,27,31,32}  Our findings are unlikely to be caused by secular trends because we demonstrate no change in risks identified over time for social domains not included in our screening tool. Beyond the existing literature, our study reveals that the implementation of an inpatient social care program can improve the timeliness of social work consultations for families and enhance the discharge communication of social risks for patients’ PCPs to facilitate longer-term care and follow-up. Also, unlike previous studies, ours was conducted in a community hospital setting within a routine standard of care, providing vital proof of concept for the expansion of social care programs to more community hospital environments to potentially reach more families in need.

The authors of some previous studies have identified longer duration of hospitalizations relative to shorter ambulatory care visits as a hypothetical facilitator for inpatient social care programs, and our study is the first that we are aware of that revealed variability in program process measure completion and outcomes relative to hospital LOS.^27,35  Our findings also highlight that LOS’s effect is polarized, and although longer LOS may indeed be a facilitator for inpatient social care programs, shorter LOS, more often experienced by otherwise healthy children and in nontertiary settings, may instead be a barrier that we must take into consideration to overcome with our inpatient social care implementation strategies. It is also important to note that, in addition to short lengths of stay, some program process measures may not have been completed because of families’ declination of in-hospital social work assessments, potentially due to preexisting relationships with social service providers in the outpatient setting or fear of stigma associated with the use of social services. Additionally, although few of our study participants had prolonged lengths of stay, social risks were identified for all patients in this group. This could illustrate the additional time required to prepare safe and effective discharges in the setting of social risks, such as housing instability or transportation difficulties. These findings may suggest that addressing families’ social risks could decrease hospital resource utilization.

We identified a lower rate of social risk in our study population compared with some previous studies in both inpatient and outpatient settings.^13,25,32,33  Although this may be reflective of variability in social risks between different patient populations, some studies have revealed improved disclosure rates by using survey methods compared with person-to-person screening^43,44  and parent preference for screening later in the course of admission after acuity has “settled”.³⁶  Therefore, it remains plausible that our study participants disclosed risks less frequently because of the in-person nature and timing of our screening program.

In the wake of multiple standard-setting organizations’ recent recommendations for inpatient social risk screening, our study may provide insights for other institutions, particularly community hospitals, seeking to implement their own inpatient social care programs.²²  The engagement of key stakeholders is paramount at all stages of implementation research and vital in the development phase of a social care program to identify facilitators and barriers to take into consideration to optimize the likelihood of successful implementation and benefit to families.⁴⁵  However, our study reveals that, even with such stakeholder engagement, additional real-world barriers will inevitably be encountered, highlighting the vitality of the continual reevaluation and evolution of inpatient social care programs. For example, in our institution, shorter lengths of stay likely impeded the completion of some in-hospital social service interventions and community resource referrals. Inpatient units often lack infrastructure for the longer-term follow-up and management that is more typical in the outpatient sector. Therefore, hospital personnel’s inability to complete inpatient social care program processes may be more likely to cause both clinician and family frustration, potentially undermining therapeutic relationships.⁴⁶  As more hospitals look to implement social risk screening while avoiding barriers and unintended consequences, they may consider the importance of including less time-intensive but effective interventions, such as premade resource packets tailored to the local community, or interventions that incorporate linkages to outpatient providers for more longitudinal follow-up into their programs.^13,30,31  Beyond individual hospital efforts, it will be vital for standard-setting organizations and policymakers to provide support not just for screening but for response to screening, bolstering hospitals’ social service programs and providing recommendations and tools for hospitals to address identified risks with effective social care interventions.

Our findings must be considered within the context of limitations. Our study was conducted at a single community hospital with its own individual culture and context; therefore, our findings may not be generalizable to other institutions. Given the single-center setting and smaller sample size, we were underpowered to detect differences between the groups for some outcomes, including resource referrals. Our pre–post test study design may reflect secular trends and increased attention to social risk factors. However, our finding that domains not included in the screening tool did not have higher rates of risks identified provides evidence against this limitation. Our standardized screening tool assesses more severe social risks; therefore, it is possible that families had less severe social risks that were not captured by our program. Finally, we were unable to assess longer-term effectiveness outcomes such as family receipt of community resources, health-related quality of life, and child health outcomes.

Our study revealed that the implementation of a stakeholder-informed social risk screening and referral program within a community hospital setting identified more families with social risks who might benefit from community resources, improved the timeliness of social work consultations, and increased the discharge communication of social risks for PCPs. Building from these findings, future studies could include prospective multi-community hospital effectiveness implementation trials evaluating longer-term outcomes and mixed methods to assess stakeholder perspectives on the feasibility, acceptability, appropriateness, and sustainability of social care programs within a community hospital environment.