Follow-up After Pediatric Mental Health Emergency Visits

We conducted a retrospective cohort study of 28 551 children aged 6 to 17 years enrolled in Medicaid with at least 1 MH ED discharge from January 2018 to June 2019. We used the IBM Watson MarketScan Medicaid Database, which contains patient-level demographic, enrollment, and health care claims data for Medicaid enrollees in 11 geographically dispersed and deidentified states.¹⁸  We excluded children with no MH and substance use coverage (4.6%), children without continuous Medicaid enrollment for 6 months after index ED visit (7.4%), and children with only MH ED visits resulting in admission, defined as a hospitalization claim on the same or next calendar day as the ED visit (4.5%) (Supplemental Fig 2).^7,19 

The child’s first MH ED discharge from January 2018 to June 2019 was classified as the index ED discharge. The date of the index ED discharge served as the anchor point to identify past-year MH service use, follow-up outpatient MH care within 7 and 30 days of the index ED discharge, and return MH acute care encounters (ED visits or hospitalizations) during the 6 months after the index ED discharge. We presumed that additional ED claims within 1 calendar day of the index ED visit mainly represented ED-to-ED transfers; thus, we analyzed these ED claims together with index ED visit claims rather than as return visits.

We defined MH encounters following specifications in the Healthcare Effectiveness Data and Information Set (HEDIS) quality measure definition for MH follow-up after ED visits.⁶  Primary MH diagnoses were identified using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) HEDIS diagnosis code sets for “mental health illness” or “intentional self-harm.” The MH encounter location (outpatient visit, ED visit, or hospitalization) was determined by procedure and place of service (POS) codes. We identified MH ED visits using POS code 23 and CPT evaluation and management codes 98281 through 98285, and MH hospitalizations based on POS codes for inpatient hospitals and inpatient psychiatric facilities (codes 21 and 51, respectively). We defined outpatient MH visits based on POS codes for outpatient settings (eg, school, home, office, health clinic) or outpatient visit procedure codes (eg, CPT 99211 “Office or other outpatient visit for E/M of an established patient”), which included care provided by MH specialists and nonspecialists.

We used a systematic method to assign each ED discharge, which may have multiple claims and ICD-10-CM diagnosis codes, to a single MH diagnosis group. We first considered ED discharges with any claims matching the HEDIS “intentional self-harm” code set as their own MH diagnosis group. For the remaining ED discharges, we used the Child and Adolescent Mental Health Disorders Classification System (CAMHD-CS) to assign a MH diagnosis group.^20,21  If the ICD-10-CM codes for an ED discharge corresponded to multiple CAMHD-CS groups, we assigned the ED discharge to the most prevalent matching MH diagnosis group in the overall sample. We then removed ED discharges with assigned MH diagnosis groups and repeated the process until <20% of all ED discharges remained, which were then categorized as “other.” This process resulted in assignment of each ED discharge to 1 of 5 MH diagnosis groups: intentional self-harm; depressive disorders; disruptive, impulse control, and conduct disorders; trauma and stressor-related disorders; and other.

The sociodemographic variables included were age group (6 to 11 or 12 to 17 years), sex, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other, missing), and insurance type (fee-for-service or managed care). We assessed race and ethnicity using a health equity framework, considering both as social constructs rather than biologic determinants.²²  As markers of the complexity of the index ED discharge, we included the number of distinct CAMHD-CS MH diagnosis groups identified from ICD-10-CM codes during the index ED discharge, and the number of calendar days in the ED during the index ED discharge encounter.^7,11,20,21  To define non-MH comorbid conditions, we used non-MH body systems from the Pediatric Medical Complexity Algorithm to group patients into 3 categories using ICD-10-CM codes: no chronic conditions, noncomplex chronic conditions, or complex chronic conditions.^11,23,24  As markers of previous MH service use, we included the number of MH outpatient visits, MH ED visits, and MH hospitalizations in the year preceding the index ED visit.

We described outpatient follow-up rates after MH ED discharges by sociodemographic and clinical characteristics and used adjusted multivariable logistic regression to determine characteristics associated with outpatient follow-up within 7 or 30 days. We used χ² tests to determine socioeconomic and clinical characteristics associated with having a return MH acute care encounter. To assess for engagement in outpatient MH care, we determined the number of follow-up visits between 8 days and 6 months for children who followed up within 7 days, and the number of follow-up visits between 31 days and 6 months for children who followed up within 30 days.

The association of having a follow-up visit within 7 and 30 days and risk of return MH acute care encounters was assessed using Cox proportional hazards multivariable models, adjusted for socioeconomic and clinical characteristics. Days between the index ED discharge and first-return acute care encounter contributed to the Cox models, at which point contributions to the model stopped and further outpatient follow-up visits or acute care encounters were not considered in the analysis. For example, if a child had an index ED discharge on day 0, a return ED visit on day 4, and a follow-up outpatient visit on day 6, the child would be counted in the model as not having had an outpatient follow-up visit from days 0 to 4, their contributions to the Cox model would stop at day 4, and their follow-up outpatient visit on day 6 would not be included in the analysis.

In our initial models, the proportional hazard assumption failed, as we found that the hazard ratio (representing the likelihood of return among children with versus without outpatient follow-up) was not constant over time. To overcome this, we examined hazard ratios by day and empirically stratified the study period into 2 groups, such that each would have constant hazard ratios. We identified these groups as ≤5 days after index ED discharge and >5 days after index ED discharge, and we stratified each analysis into 2 models based on these periods to fulfill the proportional hazard assumption. All analyses were performed using SAS 9.4 (SAS Institute; Cary, NC). This study was deemed exempt by the Lurie Children’s Hospital institutional review board.

Of 28 551 children aged 6 to 17 years old with an MH ED discharge, three-fourths (75.5%) were aged 12 to 17 years, 51.6% female, 57.0% non-Hispanic White, and 31.7% non-Hispanic Black (Table 1). The most common MH diagnoses were depressive disorders (39.1%); disruptive, impulse control, and conduct disorders (25.0%); and trauma and stressor-related disorders (14.2%). After the index ED discharge, 31.2% of children had follow-up within 7 days and 55.8% had follow-up within 30 days. Of patients who followed up within 7 days, a median of 8 additional outpatient visits occurred (interquartile range 3, 16) between 8 days and 6 months. Of patients who followed up within 30 days, a median of 6 additional outpatient visits occurred (interquartile range 2, 12) between 31 days and 6 months.

Non-Hispanic Black children were less likely to have follow-up within 7 days (adjusted odds ratio [aOR], 0.89; 95% confidence interval [CI], 0.84-0.94) and within 30 days (aOR, 0.78; 95% CI, 0.74-0.83) than non-Hispanic White children. Compared with children with capitated insurance, children with fee-for-service insurance were less likely to have follow-up within 7 days (aOR, 0.72; 95% CI, 0.68-0.77) and within 30 days (aOR, 0.83; 95% CI, 0.79-0.88). Follow-up visit rates increased as the number of previous-year MH outpatient visits increased. Children with 14 or more previous-year MH outpatient visits had 5.37 times higher adjusted odds of follow-up within 7 days (95% CI, 4.96-5.82) and 9.53 higher adjusted odds of follow-up within 30 days (95% CI, 8.75-10.38) compared with children with no previous-year outpatient MH visits.

After the index ED MH discharge, 6.5% of children had a return acute care encounter within 7 days, 12.8% within 30 days, and 26.5% within 6 months (Table 2). The distribution of timing of return MH acute care encounters is illustrated in Fig 1. Of return MH acute care encounters, 49.7% were ED discharges (ie, treat and release) and 50.3% were hospitalizations (with or without an associated ED visit). Rates of return MH acute care encounters varied significantly by socioeconomic and clinical characteristics including race/ethnicity, insurance plan type, number of comorbid MH conditions, and previous-year MH ED visits and hospitalizations.

Having an outpatient follow-up visit within 7 days was associated with a 27% decreased risk of having a return MH acute care encounter within 5 days of the index ED discharge (hazard ratio [HR], 0.73; 95% CI, 0.62-0.86) and an increased hazard for a return MH acute care encounter after 5 days of the index ED discharge (HR, 1.07; 95% CI, 1.02-1.13). Having an outpatient follow-up visit within 30 days was associated with a 26% decreased risk for a return MH acute care encounter within 5 days of the index ED discharge (HR, 0.74; 95% CI, 0.63-0.91) and an increased hazard for a return MH acute care encounter after 5 days of the index ED discharge (HR, 1.20; 95% CI, 1.14-1.27).

In a large sample of Medicaid-enrolled children discharged from the ED for an MH condition, fewer than one-third of children had outpatient MH follow-up within 7 days of discharge and less than 60% had follow-up within 30 days. The odds for follow-up care were lower for non-Hispanic Black children and children with fee-for-service insurance, whereas the odds for follow-up care progressively increased with the number of previous-year MH outpatient visits. Timely follow-up was associated with a lower risk of return for 5 days after the index ED discharge, but an increased risk thereafter.

Rates of timely follow-up among non-Hispanic Black children were particularly low, with 10% fewer receiving follow-up within 30 days compared with non-Hispanic White children. This is consistent with previous literature demonstrating that Black children are less likely than White children to receive outpatient MH treatment and underscores the need to remove barriers to MH care for Black children.²⁵  Strategies to address this disparity may include reducing stigma to seeking MH care, improving diversity in the MH workforce, and increasing availability of community and school-based MH services.^25–27  In addition, children with fee-for-service Medicaid were less likely to have outpatient follow-up than children enrolled in a managed care Medicaid program. Although characteristics of enrollees in these plans may differ, managed care payment structures may also work to promote population health management and incentivize performance on quality measures.²⁸  The strongest predictor of outpatient-follow up was having previous-year MH outpatient visits, which may represent having an established MH provider, access to a usual source of ambulatory MH care, or having regular outpatient MH visits already scheduled before the ED visit.⁷  Interventions to improve follow-up after MH ED visits should focus on children with new MH diagnoses who have not previously engaged in outpatient MH care.

More than one-quarter of children with MH ED discharges in our sample had a return MH ED visit or hospitalization within 6 months, which is consistent with previous estimates.^13,29  These high return rates suggest that EDs may not be effective sources of care for management of MH crises for children and adolescents. We found that return acute care encounters were much more common among children with previous ED visits or hospitalizations within the past year. These may be markers of increased clinical severity or other unmeasured markers of a family’s likelihood to seek care and may indicate an opportunity for targeted intervention.¹³ 

We found timely follow-up was associated with increased returns after 5 days from the index ED discharge. This aligns with previous studies examining follow-up after MH hospitalization, which paradoxically demonstrated an increased risk of readmission among children who established outpatient follow-up.^11,30–32  Similarly, in a Canadian study of ED visits for anxiety or acute stress reactions, children who had follow-up outpatient MH care had a shorter ED return time.¹⁴  Unaccounted clinical severity may partially explain these findings because children with greater clinical severity may be scheduled for more frequent follow-up visits or families may be more likely to attend visits. Timely follow-up may be a marker for having an established MH provider, decreased stigma in seeking MH care, or increased family resources.⁷  Outpatient follow-up may also directly result in increased acute care utilization, if symptoms are recognized during follow-up visits (such as worsening of suicidal ideation) that prompt appropriate referrals for acute MH care. Poor engagement in outpatient MH care does not appear to explain the increased risk of return, as children who followed up within 7 and 30 days had a median of 8 and 6 outpatient MH visits within 6 months, respectively.

This study has several limitations. We defined MH ED and outpatient follow-up visits using administrative HEDIS code sets, which may result in misclassification of some visits. Misclassification may have also occurred in our assignment of MH diagnosis groups to visits, and we may have incompletely adjusted for the severity or complexity of illness. Past-year MH care may have been underestimated because we only required continuous enrollment for 6 months after the ED discharge. We assumed that visits with additional ED claims within 24 hours represented transfers, but this approach may have excluded some visits resulting in discharge to home and return within 24 hours. We could not determine which outpatient MH visits were scheduled specifically in response to the ED visit versus which had been previously scheduled. Although we included data from 11 states across US regions, the results may not be generalizable to all states. Finally, although we identified contact with outpatient follow-up care, we could not assess the quality of care provided, which may influence return rates. To address these limitations, prospective studies are needed to understand how MH service utilization after MH ED discharges varies based on predisposing factors, enabling factors, and the need for ongoing MH care (including illness severity), and to assess how the type and quality of follow-up care may influence return rates. Future research should also assess specific interventions to promote outpatient follow-up, such as care coordination and use of telemedicine, to determine if they reduce return visits or promote cost savings.^33–35 

Among Medicaid-enrolled children, rates of outpatient follow-up after a MH ED discharge are low, particularly for children who are Black, with fee-for-service insurance, and without previous MH outpatient visits. After an MH ED discharge, more than one-quarter of children return to the ED or hospital within 6 months. In addition, timely follow-up was associated with reduced returns within 5 days of the ED discharge, but an increased risk of return thereafter. Interventions that promote connection to follow-up care after a MH ED discharge should target linkage to outpatient services within 5 days to maximize opportunities to reduce return visits.

aOR

adjusted odds ratio

CAMHD-CS

Child and Adolescent Mental Health Disorders Classification System

CI

confidence interval

ED

emergency department

HEDIS

Healthcare Effectiveness Data and Information Set

HR

hazard ratio

ICD-10-CM

International Classification of Diseases, Tenth Revision, Clinical Modification

MH

mental health

POS

place of service