Social Needs Screening in Hospitalized Pediatric Patients: A Randomized Controlled Trial Free

We conducted an RCT at a large, urban, quaternary care, freestanding children’s hospital to determine the preliminary efficacy of a hospital-based intervention for social needs. We recruited the mothers of pediatric patients 0 to 36 months of age admitted to the general pediatrics services on day 3 or later of hospitalization from June 13, 2016 to July 18, 2017. The study site has ∼8000 general pediatric admissions annually with a median length of stay of 2 days. In routine patient care, social work is not automatically consulted on all general inpatients. Financial counselors often self-consult for uninsured patients, and care managers assist with home health planning as needed. Our study focused on mothers, as has been done in previous studies,¹⁰  which allowed us to screen for maternal depression in the postpartum and early childhood period. The age range of 0 to 36 months was chosen given evidence of potential to improve neurocognitive and developmental outcomes by addressing poverty-related risk factors in young children.¹²  Caregivers were excluded if they were non-English-speaking because of limited study and interpreter service resources. Caregivers were also excluded if they lived outside of the greater Houston area (∼50-mile radius from study site), had previously participated in the study, or if the child was in Child Protective Services custody. Caregivers who met inclusion criteria were approached from ∼3 pm to 7 pm on weekdays and 9 am to 2 pm on weekends when the research team was available (137 days during the study period). Although we used a convenience sample for recruitment, we prospectively randomly assigned participants to an intervention and control group to evaluate the effects of that intervention on health-related biomedical outcomes, consistent with an RCT study design. Written informed consent was obtained from caregivers. The study was approved by the institutional review board and registered on clinicaltrials.gov (NCT03416712).

Study subjects were randomly assigned with a 1:1 allocation to control versus intervention groups immediately after consenting to study participation and before screening was conducted. Screening was done at the child’s bedside in the hospital room. All study subjects completed a baseline questionnaire that assessed social risks, which are individual-level adverse SDoH,¹³  including childcare, education, employment, food, health insurance, housing, PCP, and utilities coverage. Of note, social risk is distinct from social needs because social needs rely on caregivers’/patients’ desire for assistance.¹⁴  Study subjects were also screened for perceived child and caregiver health, caregiver mental health and safety (depressive symptoms, perceived stress, satisfaction with life, substance abuse, intimate partner violence [IPV]), sociodemographic characteristics, previous social needs screening by a doctor (education, employment, childcare, housing, food, and heating), and past and current receipt of public resources. The baseline survey was developed by Garg et al and based on Children’s Health Watch, as described in previous work.¹⁰ 

Participants in the control arm received standard of care, and social work was consulted when requested by the patient or determined necessary by the primary medical team or staff. Routine patient care did not involve standardized social needs screening, and the medical team was not aware of the baseline questionnaire results.

An outpatient SDoH guide was used during the planning of screening tools, strategies, and ethical considerations.¹⁵  Participants in the intervention arm received the WE CARE Houston clinical screening instrument. This tool was designed specifically for social needs screening, a family-centered method to intervene on individual adverse SDoH on the basis of parents’ priorities,¹³  and assessed caregiver’s receptiveness to assistance (ie, do you want help?). WE CARE Houston was adapted from WE CARE, which focused on 10 basic needs (alcohol, childcare, caregiver education, depression, drugs, employment, food, housing, IPV, smoking).¹⁶  WE CARE Houston expanded to focus on 20 basic needs on the basis of a literature review of validated social risk screening tools. A comprehensive list was chosen because the work was exploratory and the needs of the study population were unknown. It was also hypothesized that caregivers would have time to complete a longer survey during the hospitalization compared with the PCP setting. WE CARE Houston screened for alcohol, caregiver education, caregiver depression, childcare, drugs, employment, food security, health insurance, housing, immigration, IPV, legal assistance, learning English, literacy, parenting, paying for medications, PCP for a family member, tobacco, transportation, and utilities (Supplemental Fig 2). The survey was piloted with 32 caregivers. Semistructured interviews with 12 caregivers were completed to assess the face and construct validity of the survey, as previously described.¹⁷  Test–retest reliability with 8 caregivers demonstrated good reliability. The survey was written at a third-grade Flesch Kincaid level, was available in English, and required <10 minutes to complete. Caregivers were offered the option of completing the surveys on paper or having a research assistant read the surveys to them.

WE CARE Houston community resource handouts (“One-Pagers”) were developed for each of the needs screened (eg, Supplemental Fig 3). They contained the resource program name, a brief description of the program, and resource contact information. They were provided by the research assistant to the intervention group on the basis of the caregivers’ screening results and receptiveness to resources on the WE CARE Houston survey by using an algorithm (Supplemental Fig 4). If a caregiver answered “yes” or “maybe later” to wanting help, they were considered receptive to resources. Housing concerns, IPV, mental health concerns, and substance abuse were considered “sensitive issues,” triggering additional surveys by the study team and social work consult. For drug use, alcohol abuse, depression, and IPV, additional screening was completed by using currently validated tools for these needs (“CRAFFT,”¹⁸  “CAGE,”¹⁹  “PHQ-8,”²⁰  and “HITS,”²¹  respectively), a resource page was provided for drug use, alcohol abuse, and depression, and social work was consulted. For positive depression screen results on WE CARE Houston, the Edinburgh Postnatal Depression Scale²²  was also completed with the mothers of infants <12 months of age. For housing concerns, details were outlined by the research assistant, and social work was consulted. Social work was also consulted if the family requested assistance with any other need or if a social work consult was determined necessary by the primary medical team or staff.

For caregivers in the intervention group, a 2-week follow-up call was performed to confirm receipt of resource pages and to offer a new copy via e-mail, if needed.

Study feasibility outcomes were participant recruitment (percentage of surveys successfully completed, percentage of caregivers who declined), participant engagement (percentage of caregivers receptive to resources), and participant retention (percentage of follow-up assessments completed).

Follow-up calls were conducted with all study participants 6 months posthospital discharge to assess self-reported resource enrollment, barriers to enrollment, and child health indicators (well-child check, ER visits, hospitalizations, immunizations, and perceived child health). Medical records were reviewed for the number of social work consults.

The primary study outcome was enrollment in a new community-based resource at the time of the 6-month follow-up call. Secondary outcomes were feasibility and baseline prevalence of unmet needs.

We estimated that each arm required a minimum of 160 caregivers to detect an absolute difference of at least 10% in the overall rate of enrollment in resources with a 2-tailed error of 0.05 and power of 80%. We conservatively estimated that the control and intervention groups would have 5% and 15% enrollment in community resources, respectively, on the basis of previous research that revealed an 18% increase in referrals (2% control versus 20% intervention) after a similar intervention in the outpatient setting.¹⁶  We estimated a 25% loss to follow-up and, therefore, targeted 200 caregivers per group. An intention-to-treat approach was used to compare study outcomes between the control and intervention groups.

We calculated summary statistics for baseline and WE CARE Houston survey responses. Univariate unadjusted analyses were performed to identify associations between the intervention and control group for baseline unmet needs, 6-month enrollment in resources, and 6-month self-reported health indicators. Categorical variables were compared by using Fisher’s exact or χ² tests. We compared continuous variables using the student’s t test and Kruskal-Wallis test. Univariate unadjusted analyses were also conducted to determine factors associated with caregiver enrollment in resources across the study population. Factors with a P value of <.20 were controlled for in our multivariable logistic regression models (WE CARE Houston intervention arm, caregiver demographics [race/ethnicity, country of birth, marital status], food security status, current Supplemental Nutrition Assistance Program [SNAP] use, current case worker, asked about food needs by a doctor previously, and perceived stress). Odds ratios (OR) and confidence intervals (CI) were determined from multivariable logistic regression models. All statistical analyses were performed by using SAS 9.4 for Windows (SAS Institute, Inc, Cary, NC). A P value ≤.05 was considered significant.

There were 413 of 417 consented caregivers who completed the baseline questionnaire (Fig 1). Overall, 26.7% of eligible caregivers who were approached declined.

Sociodemographic variables did not significantly differ by study arms, with the exception of caregiver marital status (Table 1). Overall, the mean child age was 11 months. Forty-five percent of the children were Hispanic, 25% were white, 20% were Black, and 66% of the patients were publicly insured. The majority of patients had a PCP (95%). One-fifth of caregivers were born outside of the United States. More than 25% of caregivers reported being uninsured, and 29% had an annual household income of <$30 000. At discharge, there was not a significant difference in social work consults by study arm.

Overall, 85% of all study participants had ≥1 social risk with a median of 2 (range of 0–9). The most prevalent risks were unemployment (46%) and childcare (37%; Table 2). Low and very low household food security was reported by 13% of caregivers.

The majority of the study cohort (93%) reported moderate and high perceived stress. Approximately 27% and 13% of caregivers had mild and moderate to severe depression symptoms, respectively. Ten percent of participants were at-risk drinkers, and 9 caregivers (2%) screened positive for IPV. A minority of caregivers (10%) reported receiving social risks screening by a doctor in the past, which could include inpatient or outpatient doctors.

The most used public resources were food resources, with 49% of caregivers enrolled in Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and 40% enrolled in SNAP.

On WE CARE Houston, 83.7% of caregivers had ≥1 social needs and 75.8% were receptive to resources (Table 3). By prevalence, the most common needs were caregiver employment, health insurance and PCP for someone in the family, caregiver depression, childcare, smoking, and food resources, and caregivers were most receptive to transportation, housing, childcare, parenting, utilities, and reading programs.

Across the full cohort, 72.4% of caregivers completed the 6-month follow-up call (73.3% control, 71.4% intervention; Table 4). At follow-up, 52.5% of all caregivers reached had enrolled in a new resource (47.8% control, 56.6% intervention, P value = .17). There were no differences in enrollment in resources by study arm. The most common resources enrolled were food, health insurance, PCP, and childcare. There were no differences in caregiver-reported health indicators by study arm.

In multivariable logistic regression, caregiver birth outside of the United States was associated with a 1.97 higher odds of successful resource enrollment (95% CI: 1.024–3.79). Patients in the WE CARE Houston intervention arm had 1.40 greater odds of enrolling in resources than the control arm (95% CI: 0.84–2.35), but this was not significant. Additionally, other factors controlled for in the models were not significantly associated with caregiver enrollment in resources.

This RCT provides an overview of a comprehensive screening and intervention program for the social needs of the caregivers of hospitalized pediatric patients. Our study revealed that social needs screening in the inpatient hospital setting was feasible and resulted in the identification of social needs. We also found a high prevalence of social needs, with 85% of the WE CARE Houston cohort reporting at least 1 social need. This is similar to previous work by Fritz et al⁷  and Vaz et al⁸  in hospitalized patients. In addition, work by Colvin et al revealed that caregivers were receptive to social needs screening²³  and that a systematic intervention could increase the detection of social needs with the iHELP tool.⁶  Most patients in our study had not been screened for social needs in the past, including by their child’s pediatrician. The results of our study support that systematic screening programs in the hospital can be used to detect social needs.

Among the risks identified, it was notable that 40% of caregivers screened positive for depression. There are limited data on the mental health of caregivers of hospitalized children, but depressive/anxiety symptoms have previously been reported in 25% of caregivers of hospitalized children.⁸  More is known on perinatal and postpartum depression (PPD) diagnoses, which are estimated to affect 11% to 19% of mothers.²⁴  Previous work examining PPD in the mothers of hospitalized infants, revealed that 28% of mothers screened positive for postpartum depression.²⁵  Manifestations of PPD may include sleep disturbance, anxiety, irritability, feeling overwhelmed, and preoccupation with the baby’s health and feeding,²⁶  and it is possible that these symptoms may be exacerbated during hospitalizations when caregivers are under increased stress. High perceived stress levels were noted by caregivers in our study, and it is unknown whether the high prevalence of depression is related to acute hospitalization or longer-standing conditions. Notably, caregivers in our study were highly receptive to mental health resources, thus revealing an opportunity to intervene in the inpatient setting.

In our study, 57% of the intervention group versus 48% of the control group had enrolled in a new resource at follow-up. Although likely clinically relevant, this was not a significant difference in enrollment in community resources. The absolute difference between the 2 groups is similar to resource enrollment in the 2015 WE CARE study in which 39% of caregivers in the intervention group enrolled in at least 1 new resource compared with 24% in the control group,¹⁰  but overall enrollment in resources was higher in our study than what we had expected on the basis of previous literature.¹⁶  It is possible that the higher rate of enrollment in resources in our study is due to caregivers recognizing their social needs as part of being in the study and seeking assistance within the hospital or in their community. Another possible cause of the higher enrollment in our study is that we screened for more resources than authors of previous studies. Finally, it is possible that the hospital setting is already uniquely positioned to address social needs at baseline. Over the course of hospitalization, social needs may be revealed as high priority or as relevant to the child’s illness, chronic care management, or discharge planning. Although hospital team members do not have longstanding relationships with caregivers, it can be an intense time in which caregivers may recognize and reveal social needs to one of the staff members with whom they develop rapport. Additionally, the new resources caregivers enrolled in most in our study were often food (including WIC, SNAP, and food pantries), health insurance for someone in the family, and a new PCP for someone in the family. These are domains that may have been addressed during routine care, such as by meeting with the financial counselor about insurance, which likely contributed to our high background enrollment in resources. In the future, hospital-based screening programs could build from their existing areas of strength and complement medical home efforts to meet the social needs of the patients they both serve.

The hospital setting provides several advantages for SDoH intervention work, including more time for screening and referrals and the availability of social work on site to provide consults as needed for sensitive needs. Our study intervention of providing resource information and phone numbers (indirect referrals) in the hospital setting is likely not optimal. Instead, direct or active referrals are another potential future intervention. Direct referrals are those in which patients’ contact information is provided to a referral site, often through health information exchange tools, and contingent on the patient’s consent. Evidence suggests that direct referrals are generally more effective than indirect referrals.²⁷  Similarly, in work by Garg et al, the research staff actively referred caregivers with resource applications on site. Additionally, some hospitals have partnered with community-based organization referral sites, such as “findhelp,”²⁸  which allow for documentation of screening results, direct referrals, and closed-loop communication to follow up on the results of the referral. Further research is needed on the effectiveness of direct or active referrals within the inpatient setting, as well as coordination with outpatient services to ensure optimal referral follow-up.

Our study had generalizability limitations and was restricted in scope because it was focused on English-speaking mothers (missing single/dual father and other caregiver households) at 1 quaternary care hospital located at an urban site in the South. However, we retained a diverse population by race/ethnicity, country of birth, education, and income, with a predominance of publicly insured patients. Additionally, our study was exploratory in the feasibility of recruiting in the hospital setting, and we focused on admissions on day 3 or later, which limits the generalizability of our work in shorter hospitalizations. Approximately one-quarter of caregivers (Fig 1) declined participation in the study, and the social risk of those who declined is unknown. It is also possible that control group caregivers requested social work consults or other assistance from hospital staff more often as a result of being in this study, contributing to our high baseline rate of enrollment in resources. Although we found an almost 10% difference in enrollment between the WE CARE Houston and control groups, which we believe is clinically significant, this did not reach statistical significance because of a lack of study power, given that we did not expect such high enrollment rates for both groups a priori. We hope our study data can inform future investigators about their sample size and power calculations. Finally, we were also limited by the recruitment of a convenience sample and the inability to blind participants, given the nature of the study.

Expansion of social needs screening across the health care system is needed to address adverse SDoH and improve health equity. Current research has focused largely on outpatient settings, but we found that screening for social needs in the hospital is feasible and can result in the identification of social needs and the receipt of resources. As hospitals increasingly implement social needs screening programs, more research is needed to develop evidence-based approaches and successful referral mechanisms that improve child and parent health, well-being, and health care utilization outcomes.

We would like to thank all of the families who participated in the study as well as Dr Ricardo Quiñonez, our Division Chief for Hospital Medicine in the BCM Department of Pediatrics, for his support. We would also like to acknowledge Drs Sara Curtis and Daniel Wang and medical students Kevin McKenzie and Sydney Garrett, for their contributions to the study.