Emergency Department Visits and Hospitalizations for Eating Disorders During the COVID-19 Pandemic

We extracted billing data for ED visits and inpatient admissions from PHIS for patients aged ≥10 years diagnosed with eating disorders. We included 38 PHIS participating hospitals that submitted complete data from January 2018 to June 2022. Patients with eating disorders were identified by principal discharge diagnosis code (International Classification of Diseases, 10th Revision; Supplemental Table 2). We also extracted aggregate total ED visit and inpatient admission volume for all patients over the study period for comparison.

Variables collected for ED visits and inpatient admissions included age at admission, sex, race, ethnicity, insurance payer, median household income by zip code, and principal diagnosis code. Inpatient admissions that originated in the ED or had any charges for psychiatric treatment unit during admission were identified by flags in PHIS. Median household income was divided into 3 groups on the basis of 2016 tertiles of income in the United States.²⁹  Principal diagnosis codes were used to group the primary eating disorder diagnoses (Supplemental Table 2).

Our primary dependent variable was aggregate monthly volume of eating disorder ED visits and inpatient admissions across all included hospitals. We also calculated eating disorder volume indexed to total patient volume (per 100 000 ED visits; per 10 000 inpatient admissions). Patients admitted via the ED were included in both ED visit and inpatient volumes, whereas inpatient volumes included all admitted patients, regardless of source of admission (eg, ED, transfer from another facility, or direct admission). We defined the pre–COVID-19 period as the 27 months from January 2018 to March 2020, and post–COVID-19 onset as the 27 months from April 2020 to June 2022. We include a reference line in figures corresponding to March 15, 2020, as the approximate onset date of the COVID-19 pandemic and the start of COVID-19–related restrictions (eg, school closures, stay-at-home orders) in the United States.

We compared patient sociodemographic characteristics and eating disorder diagnosis by visit type between pre– and post–COVID-19 onset periods using t tests (continuous age) and χ² tests for all other categorical variables. We reported frequency (percentage) for categorical variables and mean (SD) for age. We also examined length of stay (LOS) pre– and post–COVID-19 onset for inpatient admissions. Because of the skewed distribution of LOS, we reported median (interquartile range [IQR]) and tested for differences by period using the Wilcoxon rank-sum test. We also reported cumulative bed-days and tested for a pre- and postdifference using unadjusted Poisson regression analysis.

Our primary analysis used interrupted time series regression examining the pre– and post–COVID-19 onset monthly trends in aggregate eating disorder discharge volume overall and indexed to total volume for (1) ED visits, and (2) inpatient admissions across all hospitals. Visual inspection of the data indicated the potential for volumes leveling off or declining beginning in ∼1 year after the onset of the pandemic. We therefore divided the postonset period into 2 periods representing the first year since onset (12 months; April 2020–March 2021) and second year since onset and beyond (15 months; April 2021–June 2022) in our regression models. Models tested for a change over time before COVID-19 (preslope), an immediate shift postonset (postintercept), and a change over time in the first year postonset (first year postslope), as well as an immediate shift and a change over time in the second year postonset and beyond. For models examining eating disorder volume indexed to total volume, estimates were exponentiated and interpreted as a percentage change. Models did not adjust for additional covariates, such as individual demographic or clinical factors. As a balancing measure, we also examined the pre- and postmonthly trends in aggregate total volume of ED visits and inpatient admissions using interrupted time series regression. All analyses were performed in SAS (v9.4; Cary, NC) at an α-level of .05.

In the 27 months before the onset of the COVID-19 pandemic, there were a total of 2793 eating disorder ED visits versus 5217 in the 27 months postonset of the pandemic across 38 hospitals. Pre–COVID-19, the mean number of eating disorder-related ED visits per hospital was 73.5 (SD = 89.1; range 6–436) compared with 137.3 (SD = 178.5; range 0–975) post–COVID-19 onset. A total of 95% of hospitals had higher aggregate raw volumes and higher average monthly visit volumes in the postonset period. Sociodemographic factors and eating disorder diagnoses among ED visits pre- and post–COVID-19 are presented in Table 1. After the onset of the pandemic, a higher proportion of ED visits for patients with eating disorders were among patients of adolescent age (14–17 years), female sex, white race, privately insured, and from higher-median income zip codes. In addition, a slightly higher proportion of visits were among patients diagnosed with anorexia nervosa. There was no significant difference in percentage of eating disorder ED visits that resulted in inpatient admission pre– versus post–COVID-19 onset (n = 2057, 73.7% versus n = 3773, 72.3%, respectively; P = .20).

The monthly trends in eating disorder visit volume overall and indexed to total ED visits are presented in Fig 1. Before the pandemic, ED visit volume for eating disorders was increasing slightly over time (β = 1.50 per month; 95% confidence interval [CI]: 0.46–2.55; P = .006; Fig 1). Post–COVID-19 onset, there was an immediate decline in the number of visits (β = −48.7; 95% CI: −79.1 to −18.2; P = .002), followed by a significant increase over time (β = 12.9 per month; 95% CI: 9.2–16.6; P < .001) through the first year in March 2021. In the second year postonset through the end of the study period (April 2021–June 2022), visit volume decreased over time (β = −6.3 per month; 95% CI: −9.0 to −3.5; P < .001).

Indexing to total ED volume indicated a slight increase over time in eating disorder visits per 100 000 total ED visits before COVID-19 (β = 1.5% per month; 95% CI: 0.8%–2.1%; P < .001; Fig 1), followed by an immediate shift postonset (β = 29%; 95% CI: 8.1%–55.0%; P = .006), and an increase over time through the first year postonset (β = 3.9% per month; 95% CI: 1.6%–6.1%; P = .001). In the second year postonset through the end of the study period, there was a decrease over time (β = −4.2% per month; 95% CI: −5.8% to −2.7%; P < .001).

The monthly trend in total ED visit volume for all patients pre- and postonset of the COVID-19 pandemic is presented in Fig 2. After a steep drop immediately postonset, total patient ED visit volume increased significantly over time through the first year of the pandemic, followed by a leveling-off in the second year and beyond with stable volumes over time similar to prepandemic volumes.

Across 38 hospitals, there were a total of 3570 inpatient admissions for patients with eating disorders in the 27 months pre–COVID-19 vs 5732 in the 27 months postonset of the COVID-19 pandemic. In the pre–COVID-19 period, the mean number of eating disorder-related admissions per hospital was 93.9 (SD = 116.6; range 3–514) compared with 150.8 (SD = 192.9; range 5–992) post–COVID-19 onset. A total of 87% of hospitals had higher raw admission volumes and higher average monthly volumes during the postonset period. Patient sociodemographic factors and eating disorder diagnosis among inpatient admissions pre– and post–COVID-19 are presented in Table 1. After the onset of the pandemic, a higher proportion of inpatient admissions for eating disorders were among patients of adolescent age (14–17 years), female sex, white race, non-Hispanic, privately insured, and from higher-median income zip codes. Postonset, a slightly higher proportion of admissions were among patients diagnosed with anorexia nervosa. Pre–COVID-19, 2057 (57.6%) patients were admitted via the ED compared with 3773 (65.8%) post–COVID-19 (P < .001). Pre–COVID-19, 535 (15.0%) of inpatient admissions had any psychiatric treatment unit utilization during admission compared with 674 (11.8%) post–COVID-19 onset (P < .001). Median LOS before COVID-19 was 8.0 days (IQR = 8.0) vs 9.0 days (IQR = 8.0) postonset (P = .018). In the 27 months before COVID-19, there were a total of 40 933 cumulative bed-days (mean = 1516 per month) for patients with eating disorders compared with 67 907 bed-days in the 27 months post–COVID-19 onset (mean = 2515 per month; P < .001).

The monthly trend in eating disorder admission volume overall and indexed to total inpatient volume are presented in Fig 3. Pre–COVID-19, inpatient eating disorder volume was increasing slightly over time (β = 1.70 per month; 95% CI: 0.70–2.71; P = .001; Fig 3). Post–COVID-19 onset, inpatient volume immediately declined (β = −46.3; 95% CI: −75.6 to −17.0; P = .003), followed by a significant increase over time (β = 11.9 per month; 95% CI: 8.4–15.5; P < .001) through the first year in March 2021. In the second year postonset through the end of the study period (April 2021–June 2022), inpatient volume decreased over time (β = −7.6 per month; 95% CI: −10.2 to −4.9; P < .001).

Indexing to total inpatient volume indicated an increase in eating disorder discharges per 10 000 total discharges over time pre–COVID-19 (β = 1.3% per month; 95% CI: 0.8%–1.9%; P < .001; Fig 3), with no evidence of an immediate shift post–COVID-19 onset (β = 0.4%; 95% CI: −14.1% to 17.3%; P = .96), and evidence of an increase over time through the first year postonset (β = 3.1% per month; 95% CI: 1.2%–5.1%; P = .002). In the second year postonset through the end of the study period, there was a significant decrease over time (β = −3.7%; 95% CI: −5.0% to −2.3%; P < .001).

The monthly trend in total inpatient admission volume for all patients pre- and postonset of the pandemic is presented in Fig 2. After an immediate steep drop postonset, total inpatient admission volume increased significantly over time through the first year, followed by a leveling off and return to prepandemic levels in the second year and beyond.

In a diverse cohort of patients from tertiary care pediatric hospitals across the United States, we found a twofold increase in both ED visits and inpatient admissions for patients with eating disorders, in the first year postonset of the pandemic, followed by a decline in the second year and beyond. Despite this decline in the second year, absolute volumes were still elevated relative to prepandemic levels 27 months after pandemic onset, particularly for ED visits. In addition, we found eating disorder inpatient admissions were slightly longer postonset of the pandemic, which, in combination with increased volume, resulted in a nearly 66% increase in monthly average cumulative bed-days. We found small but significant differences in patient sociodemographic factors and eating disorder diagnosis comparing those seen pre– and post–COVID-19 onset; importantly, however, these differences were too small to likely be clinically meaningful, though they are indicative of ongoing disparities in access to quality eating disorder care by insurance and sociodemographic factors.

The COVID-19 pandemic has led to a mental health crisis, particularly for children and adolescents whose daily routines and support systems were completely disrupted as a result of school closings.^2,6,7  Previous studies have found increasing incidence and worsening symptomatology for patients with eating disorders after the onset of the pandemic, which has resulted in an unprecedented increase in the volume of patients seeking eating disorder care across multiple settings.^{12,15–17,20,28}  Similar to previously reported findings,^14–16  we report an approximate doubling in emergency visits and inpatient admissions to pediatric hospitals in the first year after onset of the pandemic. Though we were unable to examine severity in eating disorder presentation in this study, previous studies have indicated that patients presenting for eating disorder care during the COVID-19 pandemic are higher acuity compared with those prepandemic.^19–21,26  Thus, there is marked strain on emergency and inpatient resources in caring for a population of children and adolescents with eating disorders.

Care for patients with eating disorders is complex, and providing that care in ED settings is particularly difficult.^24,25  Providing quality eating disorder care in ED settings requires intensive care coordination and specialized medical care from providers often unfamiliar with management and treatment strategies.^24,25  In addition, patients presenting for emergency care since the pandemic have a significantly increased medical and psychiatric burden, with increased signs of dehydration, increased bradycardia, and increased concurrent mental illness.^20,26  These changes in severity of illness means additional workup and consultation must be performed in the ED, further straining an environment already facing challenges in managing the current mental and behavioral health crisis.²⁷  Additional studies are needed to quantify the resource utilization in eating disorder populations relative to other populations cared for in EDs, as well as to identify variation across hospitals that may identify best practices in an effort to improve care.

Our findings also reveal the impact the pandemic had temporally for children seeking emergency and inpatient care in general, as well as for eating disorders specifically. For example, the visit rate to EDs for eating disorders was increasing slightly before the pandemic, indicating that, even before the pandemic, this population was seeking care more frequently. With the onset of the pandemic, the volume of visits for patients with eating disorders seeking emergency care returned and surpassed prepandemic volumes within ∼3 months, whereas the overall ED volume took ∼12 to 18 months to return to previous levels. Similarly, inpatient admissions for eating disorders were increasing over time before the pandemic, but surpassed prepandemic volumes within a few months, whereas the overall inpatient volume took until mid-2021 to rebound to prepandemic levels. By the end of our study period, inpatient volumes had declined and were more reflective of prepandemic volumes. This may indicate a decline in numbers needing care or may reflect reduced capacity for patients with eating disorders. Interestingly, we found there was no difference in the percentage of ED visits for eating disorder care that resulted in inpatient admission pre- versus postonset of the pandemic. However, the percentage of inpatients with eating disorders admitted from the ED was significantly higher after the onset of the pandemic versus prepandemic. This likely reflects a combination of factors, including capacity issues in the ED and hospital, and increased availability of telemedicine allowing more frequent and rapid follow-up with discharged patients. The increase in admissions from the ED may represent a decline in direct admissions from outpatient clinics, which could be a result of worsening severity and a resultant increase in those emergently seeking care rather than waiting for an outpatient visit. In addition, we found that psychiatric treatment unit utilization among inpatient admissions was lower post–COVID-19 onset, likely reflecting limited psychiatric unit capacity rather than a decline in the mental health needs of this population, though further study is needed to clarify this finding. Because the potential for additional COVID-19 variants causing substantial disease remains high, it is vital to realize the rapid impact on patients with mental and behavioral health disorders, especially those with eating disorders. Future research should prioritize ongoing surveillance, continuing to examine changes over time in patient volume and visit rates for this vulnerable population.

Our findings and others demonstrating increasing patient volume and severity underscore the importance of upstream interventions occurring before the ED visit, such as eating disorder prevention and early detection, providing benefit to both patients and the health care system in general.^13,14,16  Specialty eating disorder programs at every level, from outpatient to residential to inpatient, have seen increased demands similar to our findings among emergency and inpatient medical settings.^15–17  As a result, our experience is that more eating disorder care has remained in primary care settings while patients await access to specialized care. Supports have been established in an ad hoc manner at our institution (eg, urgent outpatient consultations, primary care trainings) but may not be enough to sustain the needs. Innovative strategies to both increase comfort and skill levels of primary care clinicians while simultaneously increasing the eating disorder-specialized workforce are needed to prevent further burdening ED and inpatient medical settings. Disparity in access to eating disorder care, particularly private, nonhospital programs, by insurance and other factors is another important consideration and an issue likely exacerbated by the pandemic. Urgent insurance reform is needed, incentivizing payers to provide more-robust payments to specialized centers for residential or partial hospitalization treatment options, thereby improving access and shifting some of the burden from emergency and inpatient settings.

This study has several limitations. First, pediatric tertiary care hospitals in PHIS may not be generalizable to other care settings (eg, community hospitals, adult tertiary care hospitals, or general EDs), though, generally, the most severe pediatric patients with eating disorders are typically seen in settings with specialty care provided by pediatric specialists. We were also unable to examine visit or admission rates among the population of patients with eating disorders or a broader population-level denominator because this information was not available. Second, we identified patients with eating disorders by principal discharge diagnosis, which may underestimate the volume of patients seeking care for eating disorders because secondary diagnoses were not included. Third, we approximated the start of the COVID-19 pandemic in late March 2020 and considered April 2020 as the first postpandemic month, which does not account for geographical variation in either the onset of the pandemic or pandemic-related restrictions and school closures. However, stay-at-home guidelines were fairly universally enacted at the beginning of the pandemic. PHIS does not release zip codes or any additional geographic information code to link school closure data, which may be an important factor associated with eating disorder volume. Finally, in this study, we examined total unadjusted eating disorder visit volume rather than sex- or diagnosis-specific changes; though consistent with previous studies, most visits were among females with anorexia nervosa.¹⁶  Future studies should examine changes in volume over time adjusted for and/or stratified by sex, diagnosis, and other sociodemographic and clinical factors to determine which patient-level factors are driving increases in volume.

Since the onset of the COVID-19 pandemic, the volume of patients seeking emergency and inpatient care for eating disorders at pediatric hospitals has increased dramatically. Our findings have important implications for pediatric hospital staffing and capacity, as well as outpatient or residential treatment accessibility, particularly if the demand for eating disorder care continues to remain above prepandemic levels.

CI

confidence interval

COVID-19

coronavirus disease 2019

ED

emergency department

IQR

interquartile range

LOS

length of stay

PHIS

Pediatric Health Information System