Potential Bias in Social Work Consultations in the Pediatric Inpatient Setting Free

We conducted a retrospective cohort study at 1 academic quaternary care children’s hospital in an urban setting. We included patients <6 years of age discharged from the PHM service between July 1, 2012 and June 30, 2020 with admission diagnoses of FTT, BRUE, accidental ingestion, and drowning. Diagnoses were identified using International Classification of Diseases, Ninth Revision (ICD-9) and International Classification of Diseases, 10th Revision (ICD-10) codes (Supplemental Information). For patients with multiple encounters, only the first was included. We excluded patients on the PHM complex care service because all receive SW consultations. Patient age at admission, sex, race, ethnicity, primary insurance, secondary insurance, length-of-stay, and presence of SW consultation orders and SW notes were obtained from the electronic health record.

Included patients were those seen by social workers after consult order placement by the medical care team. At our institution, SW consultations are performed by the social worker assigned to the unit on weekdays and by a covering social worker on nights and weekends; SW consultation is not protocolized for the selected diagnoses.

This study was deemed exempt by the institutional review board.

Our primary dependent variable was SW consultation, defined as having a consultation order placed or having a SW note documented during the admission. Our primary independent variable was patient race, considered a proxy for racism. Per our institutional standard, race is documented according to family self-report; up to 3 races can be selected. We used categories of Black, multiracial (≥2 races selected), white, and other, a category created for analytic purposes, which included those identified as American Indian or Alaska Native, Asian, and Native Hawaiian or Pacific Islander, each <2% of our study population. Ethnicity was classified as Hispanic and non-Hispanic, categorized separately from race in our system. Language was defined as English, Spanish, or other. Insurance was defined as public, private, or unknown.

We selected and reviewed 10% of the included charts with SW consultation to identify reasons for consultation. We first grouped charts by diagnosis, race, and insurance; then, we randomly selected charts for review within each grouping, proportional to our total population. Reviewed charts encompassed all study years. We reviewed SW notes, other team member notes, and consultation orders to identify consultation reason(s). We identified 3 categories, including safety, abuse, or neglect concern, resource needs, and other, and >1 category could be selected.

We used χ² or Fisher’s exact tests to evaluate bivariate associations between SW consultation and race, diagnosis, and relevant covariates. We used multivariable logistic regression to assess for independent associations between SW consultation and race, adjusting for diagnosis, ethnicity, language, and insurance. In a post hoc analysis, we tested a race-by-insurance interaction term within the model. R software was used for statistical analyses.¹² 

A total of 1341 patients met the initial inclusion criteria. After excluding 109 repeat encounters and 33 patients admitted to our complex care service, 1199 encounters remained for analysis, 682 with SW consultations (57%), as follows: 806 FTT, 345 BRUE, 26 accidental ingestion, and 22 drowning. Twenty-two percent of patients were identified as Black, 64% were identified as white, and 7% were identified as multiracial (Table 1). Fifty-four percent of patients were male and <5% identified as Hispanic; 96% of families had a primary language of English and 77% were publicly insured.

There were significant bivariate associations between SW consultation and patient race across all assessed diagnoses; 72% of Black patients received SW consultation compared with 64% of multiracial patients and 52% of white patients (P < .001). Similar patterns were noted for each specific diagnosis. For FTT, 83% of Black patients received a SW consultation, compared with 70% for multiracial patients and 61% for white patients (P < .001). For BRUE, SW consultation was requested for 36% of Black patients, 42% of multiracial patients, and 27% of white patients (P < .05). For accidental ingestions and drownings, all Black and multiracial patients received SW consultations, whereas not all white patients received them; however, this finding did not reach statistical significance.

The adjusted odds of SW consultation for Black patients were 1.61 times higher than for white patients (95% confidence interval 1.14–2.29; Table 2). The odds of SW consultation for those with public insurance were 6.1 times higher than for those with private insurance (95% confidence interval 4.28–8.80). We observed no association between SW consultation and ethnicity or language; however, just 4.5% of included patients were Hispanic, and <4% spoke a language other than English. When race-by-insurance was analyzed, the interaction of race and insurance was not significant.

We reviewed 67 patient charts, 10% of those with SW consultations (Fig 1). Consultations for abuse, neglect, or safety concerns were more common among Black compared with white patients (68% vs 50%). However, SW consultations for resource needs were similar between Black and white patients (41% vs 48%). Compared with privately insured patients, those who were publicly-insured were more likely to have referrals placed for abuse, neglect, or safety (58% vs 38%) and less likely to have referrals placed for resource needs (44% vs 63%).

Among children admitted for common PHM conditions in which abuse or neglect are considered as underlying etiologies, Black patients and publicly insured patients were more likely to receive SW consultations than their white and privately insured counterparts, respectively. We found associations between SW consultation and patient race for FTT and BRUE; these diagnoses often require further workup or investigation to determine underlying etiologies. As part of this workup, care team biases may affect how frequently teams consider abuse or neglect.

Our findings align with previous studies revealing racial and ethnic disparities in abuse evaluations.^10,13,14  Additionally, SW consultation rates were high (>80%) across all racial groups for patients hospitalized for accidental ingestion and drowning. Associations were not significant for these diagnoses, although, it was notable that SW consultation was obtained for all Black and multiracial patients but not all white patients. Thus, provider bias may still be at play. The potential implications, including the overdiagnosis of abuse or neglect in Black patients or underdiagnosis of such conditions in white patients (or rarer pursuit of safety assessments), cannot be ignored.

Care teams engage social workers to investigate suspected abuse or neglect. Social workers play a critical part in addressing social needs and supporting family mental health. We acknowledge that SW consultation across these common diagnoses may be requested for multiple reasons. However, we noted Black patients and publicly insured patients had SW consultations more often for safety, abuse, or neglect concerns. This supports our hypothesis that provider bias may affect such workup for these concerns.

Black and low-income children are overrepresented at every stage of the child welfare system and evaluations of suspected abuse in emergency settings are likely biased.¹⁵  Our study provides insight into potential contributing factors. Care teams’ heightened suspicions of abuse or neglect in Black and socioeconomically disadvantaged children hospitalized with common pediatric diagnoses may perpetuate inequitable rates of SW consultation, which may lead to increases in referrals to child protective services.⁷  To provide equitable care, we should consider standardized SW involvement (eg, automatic referrals, utilization of standardized screening tools to prompt SW referrals). In addition, the promotion of increased representation across the medical field and additional implicit bias training for all care team members could promote the provision of more equitable care. Standardization and reducing variability at every stage of patient workup and treatment of these diagnoses may facilitate early identification of children at risk for abuse or neglect, lead to more prompt intervention, and provide resources and support to families in need.^16–18 

Our study has limitations. First, our data are from a single children’s hospital; results may not be generalizable. Second, this was a retrospective study; thus, we are unable to assess causality between dependent and independent variables. Our chart review, however, provided support for our conclusions. Future qualitative insights from care team members and families would help to further contextualize our findings. Finally, we did not explore the demographics of consulting providers, which has been shown elsewhere to affect patient experience and outcomes.^19–21 

Our findings indicate the potential overassessment of abuse or neglect for Black children and publicly insured children or an underassessment for white children and privately insured children during admissions for common pediatric diagnoses. Standardized approaches to SW consultation may facilitate more equitable, less biased care.