Classification of SARS-CoV-2 Pediatric Hospitalizations: Delta vs Omicron Variant

This was a retrospective cohort study characterizing the reason for hospitalization in pediatric patients ≤18 years of age who tested positive for COVID-19 at a tertiary care pediatric hospital in South Florida. Patients were admitted to the hospital from July 1, 2021, to February 3, 2022, and tested positive for SARS-CoV-2 via nasopharyngeal testing (antigen and/or polymerase chain reaction). During the entire period of this study, universal testing for COVID-19 was performed on every patient before hospitalization. Hospitalizations were classified as admitted either “for COVID” because of symptomatic COVID-19 or “with COVID,” signifying incidental positivity but admitted for a reason unrelated to COVID-19. This classification was determined on review of the electronic medical record based on the reason for admission as obtained from the admission history and physical note. Symptomatic COVID-19 was determined by admission reasons secondary to respiratory symptoms, sepsis, dehydration related to emesis or diarrhea, or neonatal fever, for example. If patients developed symptoms of COVID-19 requiring medical attention during admission but were initially admitted for an unrelated reason, they remained classified as “with COVID.” If we could not determine the relationship between SARS-CoV-2 infection and cause of admission, for example, in the case of admissions for a brief resolved unexplained event, increased seizure frequency, or new-onset diabetes mellitus, we classified these patients as “for COVID” on careful exclusion of alternative causes of symptoms in the medical record. Interrater reliability testing was performed between 2 raters on 100/492 (20%) of participants classified as “for COVID” or “with COVID” with determination of the κ statistic.

Patients were further separated into groups based on presumed SARS-CoV-2 infection variant type (delta variant versus omicron variant), which was assumed based on national variant predominance during admission time frames. Admissions from July 1, 2021, through September 31, 2021, were assumed to be due to the delta variant; admissions from December 11, 2021, through February 15, 2022, were assumed to be due to the omicron variant. Admissions in October and November 2021 were excluded to account for overlap of variant types. Disease severity was determined based on the requirement of PICU level care, receipt of remdesivir, and LOS. The study was reviewed by our institution’s institutional review board and determined to be exempt.

Descriptive statistics were used to review demographic information and clinical characteristics as obtained from the medical record, which were summarized with mean and standard deviation (SD) for continuous variables and counts and column percentages for categorial variables. A χ² test was used to analyze the association between admission reason and assumed variant type as well as the association between variant type with PICU admission, receipt of remdesivir, and LOS separately. P values were determined for each association separately with cutoffs set at .05, and estimates were reported with 95% confidence intervals (CIs).

A stratified analysis was performed for the association between admission reason and assumed variant type and the separate associations between variant type and PICU admission, receipt of remdesivir, and LOS, all stratified by age (<1 year, <5 years, and ≥5 years) using the Cochran–Mantel–Haenszel test.

There were 492 patients admitted to the hospital from July 2021 to February 2022 who tested positive for COVID-19. Overall, 303/492 (62%) were admitted “for COVID” and 189/492 (38%) were admitted “with COVID.” There were 185 patients admitted during the delta variant predominant time frame and 307 patients admitted during the omicron variant predominant time frame. Some of the reasons for admission in the group admitted “with COVID” included suicidal ideation, aggression, abnormal movements, skin and soft-tissue infection, urinary tract infection/pyelonephritis, and appendicitis. Suicidal ideation and appendicitis were the 2 most common. Interrater reliability testing on a subset of patients provided a κ statistic of 0.718, indicating substantial agreement on determination of the reason for admission.⁷ 

Among patients admitted with SARS-CoV-2 infection likely because of the delta variant, there was a mean age of 8.0 years (SD, 6.6 years) with 48% female and 52% male. Average LOS was 7.2 days (SD, 12 days). There were 138/185 (75%) admitted “for COVID” and 47/185 (25%) admitted “with COVID.” Thirty-eight (21%) patients required PICU admission and 50 (27%) patients received remdesivir (Table 1). Among patients admitted with SARS-CoV-2 infection likely from the omicron variant, there was a mean age of 7.4 years (SD, 6.6 years) with 47% female and 53% male. Average LOS was 3.4 days (SD, 11 days). There were 165/307 (54%) admitted “for COVID” and 142/307 (46%) admitted “with COVID.” Forty-four (14%) patients required PICU admission and 42 (14%) patients received remdesivir (Table 1). Most patients who were vaccine eligible at the time of admission remained unvaccinated against COVID-19.

The association between patients admitted “for COVID” versus “with COVID” during each variant predominant time frame was statistically significant (P < .001), with more patients admitted “for COVID” during the delta variant predominant time frame (75%) in contrast to almost equal numbers of patients admitted “for COVID” (54%) and “with COVID” (46%) during the omicron variant predominant time frame.

Disease severity was significantly increased during the delta variant predominant time compared with the omicron variant predominance based on PICU admissions, receipt of remdesivir, and LOS. There was a greater percentage of PICU admissions occurring at any point during hospitalization (21% vs 14%; P < .001), a greater percentage of patients who received remdesivir (27% vs 14%; P < .001), and a longer average LOS (7.2 days vs 3.4 days; P < .001) during the delta variant predominance.

When patients were separated into age groups <5 years and ≥5 years, most admissions occurred in patients ≥5 years for both variant types. During the delta variant predominant time frame, there were more admissions “for COVID” in both age groups compared with the omicron predominant time frame (Table 2). However, there was an increased odds of admission “for COVID” in patients <5 years old during the delta predominant time frame (P = .013) (odds ratio [OR], 2.6; 95% CI, 1.2–5.7) (Table 2). During the omicron variant predominant time frame, there was also an increased odds and significantly greater number of patients admitted “for COVID” who were <5 years (78%) compared with older children ≥5 years (39%) (P < .001) (OR, 5.7; 95% CI, 3.4–9.4) (Table 2). The increased odds of admission “for COVID” in patients <5 years remained statistically significant when stratified and adjusted for presumed variant type (P < .001) (OR, 4.4; 95% CI, 2.9–6.8). The association between age and receipt of remdesivir, PICU admission, and LOS for both assumed variant types was not significant.

Infants <1 year of age followed a similar trend with the majority admitted “for COVID” in both the delta and omicron variant predominant time frames (Table 2). There were similar percentages of patients who required remdesivir and only 1 PICU admission during each time frame. Infants admitted during the delta variant predominant time frame had longer LOS than those admitted during the omicron predominant time. When compared with patients <5 years of age, there was an increased odds of admission “for COVID” in infants for both the delta and omicron predominant time frames, although this was not significant (P = .127) (OR, 1.4; 95% CI, 0.36–1.36).

Since the onset of the COVID-19 pandemic, about 19% of all SARS-CoV-2 infections in the United States have affected the pediatric population.⁸  Children were initially considered to be largely asymptomatic with minimal impacts on health care utilization, which is not necessarily the case with the emergence of new SARS-CoV-2 variants.^9–11  Our findings suggest that although pediatric hospitalization rates increased during subsequent stages of the pandemic, hospitalization rates in children may serve as an overestimation of true disease severity. We discovered that during the peak of national COVID-19 pediatric hospitalizations in the winter of 2021–2022 during the omicron variant predominant time frame, nearly half of the SARS-CoV-2 positive admissions were either asymptomatic or minimally symptomatic with an unrelated cause for admission.

Hospitalization rates in children seem to mirror community prevalence rather than the disease severity itself. This may be the result of higher infectivity of the omicron variant compared with prior variants.¹¹  It is estimated that the incidence of the omicron variant is 6 times higher with a higher effective reproduction number (Rt) than that of the delta variant.^11,12  A recently published study noted higher SARS-CoV-2 positivity in January 2022 during the omicron variant predominant time frame compared with other variants, which maintains the belief that this variant has higher infectivity.¹³  This could account for the increase in number of hospitalizations with COVID-19 during the winter months of 2021–2022.

The increase in hospitalization rates in the United States during the omicron variant predominance was most prevalent in patients <5 years of age per Centers for Disease Control and Prevention data, almost doubling the hospitalization rate in this age group before the omicron variant predominance.^14,15  Although our institution had similar hospitalization rates among patients <5 years old for both presumed variants, we interestingly found that this age group was more likely to be admitted with symptomatic COVID-19 regardless of the variant type. We postulated this could be the result of vaccination in older patients who were vaccine eligible at the time of admission, although we did not have enough vaccinated patients in our study to make this determination.

Infant hospitalization rates from COVID-19 have been on the rise, particularly in those <6 months of age with increased likelihood of hospitalization due to COVID-19 compared with other children <5 years old.^15,16  We obtained similar findings for infants, with a greater percentage of admissions with symptomatic COVID-19 during both the delta and omicron variant predominant time frames compared with older patients, but PICU admissions were scarce. Increased hospitalization rates could be the result of emergency department provider reluctance to discharge a febrile infant <6 to 8 weeks of age or the development of bronchiolitis in response to COVID-19, for example. More studies are needed to determine the effects of COVID-19 infection on infants.

Overall, there was a higher proportion of symptomatic COVID-19 hospitalizations during the delta variant predominance, and hospitalized patients had a higher disease severity with increased PICU admissions, increased receipt of remdesivir, and longer LOS than their counterparts during the omicron variant predominance. This may indicate a greater disease severity during early periods of the pandemic despite lower hospitalization rates. At our institution, we do not have a PICU stepdown unit and are able to provide oxygen supplementation via high-flow nasal canula for bronchiolitis on the medical/surgical floors. Thus, many of the PICU admissions secondary to COVID-19 required escalation of oxygen therapy beyond high-flow nasal canula, aligning with higher disease severity.

A limitation of our study is the single center, retrospective design that may influence generalizability. Nevertheless, as previously mentioned, our findings mirror those of other investigations suggesting that the ratio of symptomatic to asymptomatic individuals with COVID-19 may align with national patterns.^9,10  The reason for admission was determined subjectively with reliance on chart review, which left room for differences in reviewer interpretation. To mitigate this, we completed interrater reliability testing, with results indicating substantial reviewer agreement. Nonetheless, the admission reason was difficult to assess and differed between reviewers in cases in which symptoms were nonspecific or could be attributed to simultaneous coinfection or another disease process. SARS-CoV-2 positivity was determined by either antigen or polymerase chain reaction positivity on admission. Because antigen results are less accurate, some patients included in this study may have had false-positive results for COVID-19. Furthermore, recent studies have suggested that appendicitis may be an intestinal reaction to SARS-CoV-2 infection and thus a symptom of COVID, but we classified these patients as admitted “with COVID.”^17,18  This could have resulted in an underestimation of symptomatic COVID-19–related hospitalizations. More research is needed to determine the causality of this relationship between SARS-CoV-2 and appendicitis. Last, we did not review medical complexity or risk factors for severe COVID-19 disease including neurologic, cardiopulmonary, or immunocompromise.

Overall, our findings suggest that caution should be taken when using pediatric hospitalization rates to calculate disease burden or severity in this population. More studies are needed to confirm the long-term effects of COVID-19 on this patient population and whether they differ based on variant type or age at presentation. Additionally, studies showing how vaccination status impacts symptom development would be useful to determine vaccine efficacy and its direct effects on health care utilization.