Impact of COVID-19 on Admissions and Outcomes for Children With Complex Chronic Conditions

We performed a multicenter, cross-sectional study of inpatient and observation status stays among children with CCCs using the Pediatric Health Information System (PHIS) database. The PHIS database contains detailed patient-level data from 49 children’s hospitals affiliated with Children’s Hospital Association (Lenexa, KS). Eight hospitals were excluded for failure to report billing data throughout the study period (7 hospitals) and for length of stay (LOS) data not reported in hours (1 hospital).

We included children with 1 or more CCCs,¹⁷  who were 0 to 18 years of age with an index admission (medical or surgical) at a PHIS-participating hospital. We used Feudtner’s description to identify the number and types of CCCs among children in our study.¹⁷  We excluded birth admissions for healthy newborns (all patient refined diagnosis related groups [APR-DRGs], 626 and 640) and maternal obstetrical admissions, because children’s hospitals have variable admission practices for these indications and most admissions are not related to acute illness. We excluded transfers, those who left against medical advice, and hospitals and hospitalizations with incomplete data (eg, delinquent data submissions, incomplete billing data through the study period, and LOS not in hours).

We present overall trends in admissions for children with CCCs between January 1, 2020 and March 15, 2021 compared to the average number of admissions during this same period in the preceding 3 years. We used a 3-year average before 2020 to account for year-to-year differences in admissions. To understand relative changes in admissions for children with CCCs during the pandemic, we also included a contemporaneous control group of children without CCCs. In examinations of clinical and financial outcomes, we examined a 1-year COVID-19 period from March 16, 2020 to March 15, 2021 and a corresponding pre-COVID-19 period of March 16, 2017 to March 15, 2020. For these analyses we divided the data into 3 subgroups on the basis of trends during the COVID-19 period as follows: COVID-19 period 1 (March 16, 2020 to May 31, 2020; corresponding approximately to the US declaration of a national emergency to the end of the academic year for many school districts); COVID-19 period 2 (June 1, 2020 to August 31, 2020; corresponding approximately to when inpatient admissions and surgical procedures began to stabilize after a period of pronounced reduction¹⁸  and the start of summer vacation for many school districts); and COVID-19 period 3 (September 1, 2020 to March 15, 2021; corresponding approximately to when hospital admissions typically increase with the onset of respiratory season and the beginning of the academic year for many school districts). The end date of March 15, 2021 was used to allow a full 30-day readmission window into April for all encounters. To allow for direct comparisons over years, we normalized all dates using 7-day increments (ie, day of the week was not considered in describing weeks of the year).

We describe overall trends in admissions for children with CCCs and for the 12 most prevalent APR-DRGs (version 36, 3M Corporation, St. Paul, MN) on the basis of admission trends before 2020. APR-DRGs are a patient classification scheme that group patients on the basis of the principal discharge diagnosis. This allows for assessment of changes on the basis of the indication for admission. For ease of interpretation, we collapsed other neonatal APR-DRG codes together and labeled them as “complex neonates” (Supplemental Table 3).

We also examined clinical and financial outcomes overall and by type of CCC. Clinical outcomes included LOS measured in hours, same-cause readmission rates at 3-, 7-, 14-, and 30-day intervals, ICU use, mortality, and use of mechanical ventilation. We examined LOS for the index admission and the total episode of care (index and readmission) and describe the days between index admissions and readmissions. In examinations of readmissions, we counted only the first readmission (if 1 occurred) for each readmission window. We examined ICU use at any time during the index admission. Financial outcomes included standardized costs in US dollars for the index admission and the total episode of care. We used previously described methods for calculating annual standardized costs using billing data and inflated standardized costs from 2017 to 2019 to 2020 dollars using the medical component of the consumer price index.^19,20 

We examined demographic characteristics including age, sex, primary payor, race and ethnicity, and median household income quartile by zip code. We included race and ethnicity as a covariate as multiple studies have reported the differential impacts of the COVID-19 pandemic on the basis of race and ethnicity.^21–24  We examined patient characteristics including planned procedures²⁵  and mean hospitalization resource intensity scores for kids (H-RISK),²⁶  H-RISK serves as a surrogate for severity of illness and is calculated by assigning relative weights to each APR-DRG and severity of illness level facilitating comparison across APR-DRG groups. We also examined hospital characteristics including hospital region and source of admission (through the ED versus direct admission).

Descriptive statistics described demographic, patient, and hospital characteristics. A χ² test compared differences in categorical characteristics and unadjusted outcomes between pre-COVID-19 and COVID-19 periods. A Wilcoxon rank sum test compared nonnormally distributed continuous variables between the COVID-19 and pre-COVID-19 eras and an independent Student’s t test compared normally distributed continuous variables. Generalized linear mixed models (GLMMs) examined the association of COVID-19 period and outcomes adjusting for age, sex, primary payor, race and ethnicity, number of CCCs, median household income, admission through the ED, hospital region, and H-RISK. GLMMs for readmission assumed an underlying binomial distribution and a logit link; GLMMs for LOS and cost assumed underlying log-linear distribution. All GLMMs included a random hospital effect to account for clustering. In a subanalysis, we examined differences in outcomes for children with CCCs over time comparing the 3 COVID-19 periods to each other. All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC), and P values <.001 were considered statistically significant (unless otherwise indicated) because of the inclusion of large sample sizes and multiple comparisons. This study, using deidentified data, was deemed exempt from institutional board review.

Hospital admissions began to decline around the 11th week of 2020 (March 11- to March 17 2020), with the greatest declines in admissions observed during the 14th week (–47.6%, April 1 to April 7, 2020) (Fig 1). We observed a sharp decline in weekly admissions from week 11 to week 14 followed by a slower recovery toward pre-COVID-19 admission numbers. Beginning week 22, children with CCCs experienced a period of stabilization in admissions near to pre-COVID-19 levels (–12.6%, May 27 to June 2, 2020). Children with CCCs experienced less of a decline and a more rapid return to pre-COVID-19 admission numbers compared to children without CCCs (Supplemental Fig 4).

The COVID-19 pandemic impacted admissions for children with CCCs based on their APR-DRG category (Fig 2, Supplemental Table 4, Supplemental Fig 5). Among the 12 most prevalent prepandemic APR-DRGs, we observed that children with CCCs hospitalized for acute respiratory infections (ie, bronchiolitis, pneumonia, and upper respiratory infections) experienced the greatest declines (−44.7% to –57.7%) in admissions during the COVID-19 period, whereas patients with diabetes experienced the greatest increases in admissions (+21.2%) during the COVID-19 pandemic. We also observed differences in admission trends over time for these APR-DRGs (Fig 2). Conditions such as “other GI diagnoses” and seizures demonstrated an early sharp decline in admissions with return toward pre-COVID-19 numbers by week 22. In comparison, children with CCCs hospitalized with acute respiratory infections experienced sharp early declines that were sustained throughout the study period, whereas admissions for “other hematologic and immunologic diagnoses” and “chemotherapy” overall remained steady throughout.

We identified a total of 780 184 admissions for 431 883 children with CCCs, of which 165 033 admissions occurred in the COVID-19 period from March 16, 2020 to March 15, 2021 and 615 151 admissions (mean: 205 050 per year) occurred in the corresponding pre-COVID-19 period (Table 1, Fig 3, and Supplemental Table 5). Compared to the mean annual number of admissions in the pre-COVID-19 period, the total number of admissions for children with CCCs declined 19.5% across the entire COVID-19 period. We observed small differences in the distributions of admissions by age, payor, median household income, and numbers and types of CCCs between COVID-19 and pre-COVID-19 periods. The mean H-RISK score was 6.1% greater among children with CCCs presenting in the COVID-19 period relative to the pre-COVID-19 period. Of included children, 1910 received a diagnosis of COVID-19 during their hospitalization, with the greatest percentage of cases observed between November 2020 to January 2021. Compared to periods 2 and 3, children with CCCs presenting in period 1 had higher proportions of 3 or more CCCs and higher H-RISK scores (Supplemental Table 6).

In adjusted analyses, the differences in overall LOS and costs of index admissions and the total episode of care, although statistically significant, were overall similar during the COVID-19 period compared to the pre-COVID-19 period (Table 2). Readmission rates were lower during the COVID-19 period compared to the pre-COVID-19 period. ICU use and mortality remained similar between pre-COVID-19 and COVID-19 periods; however, the use of mechanical ventilation for children with CCCs was lower in the COVID-19 period. Unadjusted outcomes are presented in Supplemental Table 7. In a subanalysis of the 3 COVID-19 periods, children with CCCs presenting in period 1 of the COVID-19 pandemic experienced longer LOS, had higher readmission rates, ICU use, and rates of mechanical ventilation though mortality rates were similar compared to the other 2 COVID-19 periods (Supplemental Table 8).

Outcomes remained similar in adjusted analyses for children with CCCs regardless of the type of underlying medical complexity with some small differences (Supplemental Table 9). For example, children with underlying respiratory complexity had shorter lengths of stay, decreased 30-day readmission rates, costs, and ICU use during the COVID-19 period, whereas children with underlying metabolic complexity had longer lengths of stay, higher costs, and ICU use during the COVID-19 period. Children with tech dependency or increased numbers of CCCs experienced overall shorter LOS and had reduced ICU use in the COVID-19 period.

During the COVID-19 pandemic, admissions for children with CCCs declined nearly 20% compared to the pre-COVID-19 period. Children with CCCs experienced the most substantial declines in early April 2020 with return toward, but not full return, to pre-COVID-19 admission numbers by late May of 2020. The pandemic affected the number of admissions and the timing of admissions for children with CCCs differently on the basis of admission indication. Overall adjusted outcomes including LOS, costs, ICU use, and mortality were similar during the COVID-19 period compared to the pre-COVID-19 period; although children with CCCs admitted during the early pandemic experienced increased resource use compared to those presenting later during the pandemic. These findings suggest that social distancing policies and fears regarding COVID-19 exposures may have affected access to care and and/or health-seeking behaviors for children with CCCs, especially within the early pandemic, although future research exploring health-seeking behaviors for this population is needed.

Among children with CCCs presenting during the COVID-19 pandemic there was a significant decline in admissions compared to previous years, especially in the early months of the pandemic. The overall 19.5% reduction in admissions that we observed for children with CCCs was less severe and not as prolonged as the reductions observed among generally healthy children presenting during the pandemic (overall 35.1% reductions;⁵  Supplemental Fig 1). Nearly half of participants with chronic diseases in a previous study reported that the pandemic impacted their health, with 53% of participants reporting missed routine health care visits.²⁷  Our observations likely represent a combination of the overall medical complexity of individual patients, caregivers’ ability to identify acute worsening and to escalate care at home, potential disruption of services related to school closures, and the impact that social distancing policies and stay at home orders had on health-seeking behaviors for this fragile population. Developing mitigation strategies, such as robust telemedicine programs, may help to further address disruptions of in-person health care delivery for children with CCCs within future large-scale disasters limiting health care access.^28,29 

Acute respiratory infections were among conditions with the greatest reductions in admissions for children with CCCs, whereas conditions such as diabetes were among a very few conditions that experienced increases in admissions during COVID-19. Studies examining ED and inpatient resource use for children during COVID-19 similarly observed significant reductions in admissions for respiratory illnesses.^30–32  Among children with CCCs presenting to a pediatric ED in Italy, Brisca et al observed that those with respiratory medical complexity experienced declines in ED visits from 15.6% in 2019% to 4.8% in 2020.⁷  The sustained reductions in hospitalizations for acute respiratory infections among children with CCCs in our study likely reflect the impact that social distancing practices, masking, and school closures had on community transmission of infections. These findings highlight how public health measures may mitigate some medical illnesses among children with CCCs and underscore a need to reevaluate infection control practices for this population. However, any practice change would need to be counterbalanced against the potential negative psychological and social impacts that these measures may have for some individuals. Previous studies highlighted delays in diagnosis of new onset diabetes and increased incidence of diabetic ketoacidosis during the pandemic.³³  Although we observed increased admissions for diabetes among children with CCCs, we are unable to determine if these hospitalizations represent new a diagnosis versus complications of chronic disease. Future studies focused on understanding the increased admissions for diabetes among children with CCCs may provide further context for these admissions and identify potential unmet needs specific to this population.

Studies of the COVID-19 pandemic in children have reported delayed presentations leading to worsening health outcomes.^1–3  In our study, children with CCCs presenting during the COVID-19 pandemic, especially during the early COVID-19 period, had higher H-RISK scores and increased number of CCCs. With some exceptions (eg, reduced LOS, 30-day readmissions, costs, and ICU use among those with respiratory medical complexity), outcomes were overall similar between pre-COVID-19 and COVID-19 periods regardless of the type of underlying medical complexity. However, in a subanalysis of children with CCCs by period of presentation, we observed increased resource use (eg, increased LOS, higher readmission rates) among children with CCCs presenting in the early pandemic. Our observations of potentially worsened health outcomes early in the pandemic may represent changes in health-seeking behavior secondary to COVID-19 policies and pandemic fears, whereas our observations later in the pandemic, including a return toward baseline outcomes for children with CCCs, may be secondary to the easing of restrictions and/or a natural progression back to previous health-seeking behavior. Alternative explanations for these findings may include changes in hospital case mix early in the pandemic secondary to admission deferral of less severe cases and/or prolonged lengths of stay related to difficulties coordinating and arranging outpatient services in the setting of lockdowns and school closures. Regardless, ensuring hospitals are perceived as a safe environment to caregivers and access to care is not limited during a national event such as a pandemic is especially important for children with CCCs.

There are several important limitations to our study. First, we used administrative data, and any data quality issues (eg, incomplete or missing data) could impact or bias our results. The PHIS database and contributing hospitals rigorously monitor data quality, so data quality concerns are expected to be small. Second, our analysis focused on admissions to children’s hospitals, which frequently provide care to children with CCCs because of the concentration of subspecialty services in these settings; however, inpatient admissions represent only 1 metric of health care access, and with lockdowns and social distancing measures, it is possible that children with CCCs may have accessed telehealth or sought care in alternative settings (eg, smaller community hospitals, ambulatory settings). Third, individual communities were impacted differently by the COVID-19 pandemic over time, which we are unable to account for, and differences in local prevalence of infections and local and regional closures and lockdowns could have contributed to some of our observations. Finally, previous reports describe variability in the impact of the COVID-19 pandemic on the basis of socioeconomic factors, and our adjustments for sociodemographic and clinical factors, including payor and race and ethnicity, may have adjusted away some of the indirect effects of the pandemic by biasing our results toward the null.

Children with CCCs experienced overall declines of nearly 20% in admissions to US children’s hospitals during the COVID-19 pandemic compared to the same timeframe in the previous 3 years. The COVID-19 pandemic differentially impacted children with CCCs on the basis of the indication for admission. Although statistically different, the overall adjusted outcomes including LOS, costs, and ICU use were similar to the pre-COVID-19 period. Future studies focused on understanding health-seeking behavior and the impact on outcomes for this vulnerable population may shed further light on our observations.