Comparing health care use after hospital visit for SARS-CoV-2, respiratory syncytial virus, and other respiratory tract infections in children

Study sample

Our study sample included all children aged 1 month to 5 years who were residents in Norway and who had a hospital visit with a diagnosis of COVID-19, RSV infection, or other RTI, between February 1^st, 2017, and September 20^th, 2021. The children with a hospital visit were assumed to have a more severe disease course than children without hospital visits, i.e., hospitalization was regarded as a proxy for severe disease. Eligible hospital visits were identified using ICD-10 diagnostic codes, categorized into three mutually exclusive diagnosis categories: COVID-19, RSV infection, and other RTI including influenza (both primary and secondary diagnoses included, see Table S1-1 for full list of included ICD-10 codes). To define our sample, we included both inpatients (overnight hospitalizations and in need for specialist care) and outpatients (not requiring overnight hospital stay yet in need for specialist care), independent of duration and urgency. We used the terms admission and discharge to describe the first and last day of a hospital visit. Hospital visits with any mention of the included diagnostic codes that occurred with less than two calendar days in between, were coded as one hospital visit, i.e., with the same hospital visit ID and earliest date of admission and latest date of discharge.

We divided our sample into two age groups (1-12 months and 1-5 years old at date of admission), because infected infants (1-12 months) are likely to have higher risk of hospital contact and post-disease health care use than older children (1-5 years).^10-12 Children with more than one hospital visit during our study period reentered the study after a washout period of 6 months to allow for multiple hospital visits without conflating exposure with outcome.

Existing studies have reported relatively high rates of coinfection with SARS-CoV-2 and other respiratory pathogens among children.¹³ Coinfections (i.e., multiple mentions of different diagnosis categories within the same hospital visit) were very rare in our sample (1.4 % of all hospital visits). In these cases, we chose the most specific ICD-10 code (i.e., with known pathogen). No children in our sample had a hospital visit registered with both COVID-19 and RSV infection. However, our data did not contain information about testing for coinfections, and it is thus likely that many coinfections were missed.

Outcomes

We studied all-cause primary health care use (including consultations at general practitioner and emergency ward) and all-cause specialist health care use (overall and separated into inpatient and outpatient services) before and after each child’s hospital visit for COVID-19, RSV infection, or other RTIs. We constructed our outcome data in a relative-week panel structure (relative to the admission date and the discharge date of the hospital visit). All outcomes were measured twelve weeks prior to hospital admission and twelve weeks after hospital discharge. All children contributed with a minimum of four weeks follow-up time before and after the hospital visit, allowing us to include newborn infants from age 1 month. More specifically, we followed each child from January 1st, 2017 (study start), date of birth, or date of immigration, whichever came last. Similarly, all children were followed until October 18th, 2021 (study end), until the day they turned 6 years old, death, or emigration, whichever came first. With a maximum observation period of twelve weeks before and twelve weeks after the hospital visit for both age groups, we obtained panel data that were equally balanced (i.e., the similar amount of relative week data for both age groups allowed for valid comparison across the age groups).

Statistical analysis

First, we presented descriptive statistics on our study sample, as well as the crude rates of primary and specialist health care use before and after the hospital visit (calculated as the weekly percentage of children with at least one primary/specialist consultation), by age and diagnosis group. To estimate how much more, or less, frequent utilization of health care services was after the hospital contact for children with COVID-19 compared to 1) children with RSV infection, and 2) children with other RTIs, we used a difference-in-differences (DiD) design with separate estimates for each four-week period after hospital discharge. This allowed us to isolate the impact of being diagnosed with COVID-19 from being diagnosed with RSV infection or other RTIs on post-disease health care use. Hence, we extended the traditional DiD design from one to three post-treatment periods: 1-4 weeks; 5-8 weeks; and 9-12 weeks after discharge, compared to the baseline twelve weeks prior to admission. The statistical models were estimated using linear regression with robust standard errors on the individual level and adjusting for calendar month and year to account for any seasonal variation in health care utilization.^14-17

To assess whether results were driven by changed patterns of health care use and health-seeking behavior due to the pandemic, we conducted a post-hoc sensitivity analysis for each outcome including only hospital contacts from March 12^th, 2020, when national COVID-19 lockdown measures were first announced in Norway, to September 20^th, 2021, using the same approach as described above (see Supplementary material 2 for further details).

In total, only 25 children aged 18 and younger were registered with codes indicative of Multiorgan Inflammatory Syndrome after COVID-19 (MIS-C) in Norway by June 2021, and no further analyses of this subgroup was undertaken due to the even lower number that can be expected when restricting to our age group of interest.¹⁸

Statistical software R v. 4.0.2 was used for data preparation and statistical analyses.

Ethics

The establishment of an emergency preparedness register forms part of the legally mandated responsibilities of The Norwegian Institute of Public Health (NIPH) during epidemics. Institutional board review was conducted, and The Ethics Committee of South-East Norway confirmed (June 4^th, 2020, #153204) that external ethical board review was not required.

Crude all-cause primary and specialist health care use

Infants aged 1-11 months had similar levels of primary health care use in the twelve weeks prior to the hospital visit for all diagnoses, while pre-admission use of primary care was slightly higher for children aged 1-5 years with RSV infection than for same-aged children with COVID-19 or other RTIs (Table S1-3). The lowest average level of primary health care use prior to the hospital visit was found for children aged 1-5 years with COVID-19 (Table S1-3). Use of primary health care increased for all diagnosis groups and both age groups in the seven days before the hospital visit (Figure 3). This is to be expected, because hospital care depends on referral from primary care at least when the degree of urgency is low. Within the first four weeks of discharge, children with RSV infection and other RTIs had resumed to pre-admission levels of primary care use, while it remained elevated until weeks 5-8 for the children with COVID-19 (Figure 3; Table S1-3).

• Download figure
• Open in new tab

Figure 3. Crude share (%) of primary and specialist health care use in the weeks before and after hospitalization.

Figure 3 note: 95 % confidence intervals (CI) calculated using the Wilson method for binomial proportions. For each week (7-day period) before and after hospitalization, the denominator equaled the number of children who could be observed the given week and the numerator equaled the number of children who had at least one health care contact (calculated separately for age and diagnosis groups).

The average weekly utilization of specialist care prior to admission was similar to the level of primary health care use for all groups, except for children aged 1-5 years with RSV infection or other RTI (Figure 3; Table S1-3). For most children, we observed an increase in the crude percentage that utilized specialist health care the week after discharge (Figure 3). Like primary health care use, utilization of specialist health care resumed to pre-admission levels within 5-8 weeks of discharge for most diagnosis and age groups (Table S1-3).

Differences-in-differences analyses, all-cause primary care use

Children aged 1-12 months with COVID-19 had a short-term increase in primary care use after the hospital visit compared to same-aged children with RSV infection (0.064 percentage points, 0.52 % relative increase), corresponding to an average weekly increase of approximately 6 consultations per 10,000 children with a hospital visit for COVID-19 (Table 1). Similarly, children aged 1-5 years also had an increase in utilization of primary health care the first four weeks after discharge, when compared to children with RSV infection (0.068 percentage points, 0.53 % relative increase) or other RTIs (0.046 percentage points, 0.45 % relative increase) (Table 1). These differences corresponded to average weekly increases of approximately 7 and 5 consultations per 10,000 children with COVID-19, respectively. After 5 weeks of discharge, there were no differences in utilization of primary health care for either age group. The elevated primary care use among infants was confirmed in our sensitivity analysis (0.086 percentage points, 0.73% relative increase), where we restricted the comparison groups to children with hospital contact during the COVID-19 pandemic (Table S2-1). We found no differences in primary care use for children aged 1-5 years in the sensitivity analysis.

Differences-in-differences analyses, all-cause specialist care use

Children with COVID-19 had no increased use of all-cause specialist care services (inpatients and outpatients studied together) after discharge when compared to children with RSV infections or other RTIs (Table 1). However, when restricting to use of inpatient services, we observed that children with COVID-19 had an increase in post-discharge utilization both for weeks 1-4 (0.044 percentage points, 0.95 % relative increase), 5-8 (0.037 percentage points, 0.80 % relative increase) and 9-12 (0.038 percentage points, 0.82 % relative increase) when compared to children with RSV infection, yet only for the youngest children (1-12 months) (Table S1-4). This corresponds to a weekly average increase of approximately 4 inpatients per 10,000 children. No increase was observed for outpatient care (Table S1-4). These findings were largely confirmed in our sensitivity analyses (Tables S2-1, S2-2). In addition, the sensitivity analyses showed elevated use of specialist care for infants aged 1-12 months with COVID-19 when compared to children with RSV infection (0.096 percentage points, 0.83% relative increase) (Table S2-1). When studying inpatient care during the pandemic, the differences in utilization after COVID-19 and RSV infection hospital visit among infants dissipated after 8 weeks, but we also found slightly elevated use of inpatient care when comparing hospital contact for COVID-19 to other RTIs in weeks 1-4 (0.029 percentage points, 0.45% relative increase) and in weeks 9-12 (0.035 percentage points, 0.54% relative increase) (Table S2-2).

Comparison to previous studies

Few similar previous studies exist that allow for a valid comparison of findings. However, in a study of caregivers/parents reports about children’s health (age 30 months or less) following RSV discharge (N=45), poorer health was observed for up to 60 days after discharge when compared to age-matched controls (N=46).⁸ Here, we could compare the rather serious hospital visit for RSV with hospital visit for COVID-19 and other RTIs in 34,214 children aged 0-5 years using health care records. In this way, we could put the consequences of RSV infection and COVID-19 into context, showing they may be more or less equally severe in terms of post-discharge health and health care use, despite that the average duration of the initial visit was longer for children with RSV infection, indicating more severe disease. The slight increase in primary health care use after hospital visit for COVID-19 compared to after RSV infection or other RTIs is in line with results from our recent study showing that children with mild COVID-19 had increased utilization of primary health care after infection, when compared to same-aged non-infected children.¹⁴ The findings in these two studies do not necessarily indicate that children with COVID-19 have more severe post-disease symptoms than children with or without other diagnoses. Rather, it may reflect differences in health-seeking behavior or the behavior of health care services when facing the new SARS-CoV-2 virus compared to the more well-known infections due to e.g., RSV. Still, health care use may be regarded as a proxy for poor health. Although we did not aim to shed light on long-term complaints or clinical consequences in any of these studies, the etiological mechanisms for potentially worse complaints and/or more health-seeking behavior following COVID-19 among children than among healthy children or children ill with other diseases, deserve further research attention in clinical studies.

Clinical implications

Our study is representative for all countries with equal access to health care services. Because hospital visits for RSV infection and other RTIs is considerably more common than hospital visits for COVID-19 among young children, our observations indicate that targeted interventions against RSV and other RTI should receive priority over COVID-19 when it comes to reducing the burden on the health services. Furthermore, there is evidence that the more contagious Omicron variant of the SARS-CoV-2 virus leads to milder disease than previous virus variants, also among children.^19-21 Whether our findings, which are based on the Wuhan, Alpha, and Delta virus variants, over- or underestimates, or correctly estimates, post-discharge health care use following COVID-19 in children should thus be addressed in future studies.

Limitations

Some limitations to our data and results should be noted. First, COVID-19 requiring inpatient or outpatient specialist care is extremely rare in children, which is reflected in the sample size and statistical strength even in our nation-wide population. Along this line, due to the suppressed incidence of hospital contacts for RSV and other RTI because of disease control measures during 2020 and 2021, we compared the incidence of the very few hospital contacts for COVID-19 during these years, with the much higher incidence of RSV infection and other RTI during 2017-2019. We handled this potential limitation by controlling for seasonal variation (calendar month and year) in our models and by running sensitivity analyses where we included RSV and other RTI hospital contacts only from March 2020 and onwards, with similar findings. Our similar observations across the main and sensitivity analyses may indicate that potential violations to the common trend assumption was not an issue in our study. Thus, we regard any potential difference in health-seeking behavior prior to and after the implementation of lockdown restrictions in March 2020 as unlikely to have impacted on our findings, at least among infants. A second limitation may be that for children in most age and diagnosis groups, we found similar utilization levels for specialist health care as for primary health care in the weeks prior to hospital admission. This may indicate that children requiring hospital care due to the included diseases generally have poorer health and thus different patterns of health-seeking behavior than the general population. Thus, hospital care may not serve as a proxy for disease severity, but rather reflect the underlying health of the child. However, both inpatient and outpatient hospital contact generally require referral from a general practitioner or emergency ward. We therefore studied those potentially more severely ill than infected children not in contact with any health care services or only in contact with primary care, as they were judged by a physician to require specialist examination or treatment. Lastly, our 12-week observation period may be too short to capture post-viral conditions, such as RSV followed by repeated episodes of bronchopulmonal obstructions.

Finally, there was no general guideline on mandatory post-COVID follow-up examinations in Norway, as has been performed in some studies.^22,23 However, we cannot exclude that the small increase in care visits observed for children hospitalized with COVID-19 was due to the caregivers requesting a closer follow-up, which was not requested for children hospitalized with other RTI.