Fluctuations in Pediatric Acute Otitis Media Burden During the First Two COVID-19 Years in Israel

Abstract

Background: 

To study pediatric acute otitis media (AOM) burden fluctuations before and during the first two COVID years, which were characterized by measures to reduce the spread of airborne diseases. We used urinary tract infection (UTI) as a comparison infection.

Methods: 

This was a cross-sectional study encompassing three pre-COVID years (March 1, 2017–February 29, 2020) and the first two COVID years (March 1, 2020–February 28, 2021, and March 1, 2021-February 28, 2022). Records were retrieved from the Clalit Health Services database, Israel’s largest healthcare maintenance organization. Children 0–15 years with AOM and UTI episodes were categorized according to age (1>, 1–4, 5–15 years). We collected demographics, seasonality, AOM complications, antibiotic prescriptions, and recent COVID-19 infections. The average AOM/UTI rates of the three pre-COVID years vs. two COVID years were used to calculate the incidence rate ratios (IRRs).

Results: 

We identified 1,102,826 AOM and 121,263 UTI episodes. The median age at AOM diagnosis was 2.0 years (IQR, 1.1–4.1). Male predominance, age at presentation, and the dominant age group of 1–4 years did not change during the COVID years. While UTI episode rates decreased during the COVID years (IRR 0.76, 95% CI, 0.68–0.84, P < 0.001), the reduction in AOM episode rates was >2-fold (IRR 0.46, 95% CI, 0.34–0.63, P < 0.001). The largest decrease was observed among children 1–4 years old during the first COVID year (β=−1,938 AOM episodes/100,00 children, 95% CI, −2,038 to −1,912, P < 0.001). Recent COVID-19 infection was associated with low AOM morbidity (IRR 0.05, 95% CI 0.05–0.05, P < 0.001).

Conclusions: 

AOM burden substantially decreased during the first COVID year but almost reached pre-pandemic levels during the second year.

INTRODUCTION

Acute otitis media (AOM) is a common pediatric infectious disease and is considered to be a complication of an upper respiratory tract viral infection.^1,2 Pediatric AOM epidemiology has been greatly affected by mass interventional campaigns, such as the introduction of pneumococcal conjugate vaccines and, to a lesser extent, immunization with seasonal influenza vaccines.^3,4 Common bacterial pathogens include Streptococcus pneumoniae, Haemophilus influenzae and Moraxella catarrahlis, and viral agents include respiratory syncytial virus (RSV), adenovirus, rhinovirus, and influenza.

The coronavirus disease-2019 (COVID-19) is a respiratory disease caused by the Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2).⁵ This novel strain of the coronavirus family is responsible for the COVID-19 pandemic that has spread globally within several weeks.⁶ During the first two COVID years, multiple measures of varying severities were imposed to limit the viral spread and infection and correlate with the number of affected individuals and hospitals’ capacity. Thus, the pediatric AOM burden would likely change again during the emergence of COVID due to measures aimed at diminishing the horizontal transmission of respiratory viral infections, which are the primary offending agents in AOM pathogenesis. Several reports from the first months of the COVID period showed a marked decrease in pediatric AOM burden worldwide.^7–10 However, data is still lacking regarding pediatric AOM epidemiology later in the course of the second COVID year, characterized by the relaxation of restrictions, the introduction of different COVID vaccines, approval of vaccine boosters, and the emergence of new COVID-19 variants.

In this study, we sought to analyze the nationwide pediatric AOM burden and the time trends of AOM episodes between the three pre-pandemic years and the first two COVID years.

METHODS

This cross-sectional, population-based study was approved by the Clalit Health Services (CHS) Research Ethics Committee (approved protocol number: 118-21). This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

First Two COVID Years in Israel

The first COVID-19 patient in Israel was confirmed on February 21, 2020. Due to a rapid accumulation of new cases, Israel began enforcing measures to limit the disease’s spread. There were three national lockdowns: March 19, 2020-May 7, 2020; September 18, 2020-October 18, 2020; and December 27, 2020–February 7, 2021. The number of pediatric COVID-19 cases remained low. From December 2020, vaccination with BNT162b2 was available for adults >16 years.

The second COVID year was characterized by two major outbreaks; The first in August-November 2021 and the second during December 2021-February 2022 – both with high infection rates among young children (5-11 years old).¹¹ Unlike the first COVID year, there were no strict nationwide lockdowns, because a significant part of the population had already been vaccinated or recovered. Nevertheless, preventive measures were still enforced.

COVID-19 (BNT162b2) Vaccination in Israel

The BNT162b2 vaccine was approved in December 2020 and was administered to the entire adult population (>16 years). In July 2021, Israel launched a third (booster) vaccination campaign, and during February 2022 (the “Omicron outbreak”), a fourth booster vaccination was approved for at-risk populations. As of March 2022, 70.5%, 64.2%, 47.3%, and 8.4% of the Israeli population were vaccinated with first, second, third, and fourth vaccination doses, accordingly.¹¹

Children >12 years were eligible to be vaccinated from June 2021. Due to the accumulating shreds of evidence of COVID-19 complications among young children (5-11 years), and especially Pediatric Inflammatory Multisystem Syndrome (PIMS),^12,13 health authorities worldwide encouraged vaccinating this age group. After a significant public debate, young children (5-11 years) were allowed to vaccinate in Israel from November 2021.

Database

We retrieved de-identified records of patients insured by Israel’s largest healthcare organization, Clalit Health Services (CHS), which insures 4.7 million patients (51.6% of the Israeli population), of them ~1.37 million individual children <15 years (data obtained from the Israeli National Insurance Institute archives, www.btl.gov.il, Hebrew). Data were extracted using the MDClone platform, which derives a novel, synthetic dataset specifically computed to preserve the statistical properties while containing none of the individuals from the original data (www.mdclone.com).

AOM Episodes

An AOM episode was defined as a visit in which AOM diagnosis or acute mastoiditis was coded with the following International Code of Disease (ICD)-9 diagnosis codes: 382.0, otitis media (OM), acute, suppurative; 382.9: OM, suppurative and unspecified; 382.4: OM, suppurative otitis media, unspecified; 383.00; mastoiditis, acute and subacute, and 383.9: mastoiditis, unspecified.

UTI Episodes

We chose urinary tract infection (UTI) as a reference infection (ICD-9 codes: infection of the kidney, 590.X; cystitis, 595.X; UTI, 599.0; infection, urinary tract 771.82; urinary complications, not elsewhere classified, 997.5). UTI is also prevalent in children but unrelated to a respiratory viral etiology, and it is usually caused by bacteria such as Escherichia coli, Klebsiella pneumoniae, and Proteus mirabilis. Therefore, UTIs would not be dramatically affected by measures aimed to reduce airborne diseases as enforced during the COVID pandemic.

Study Population

Inclusion criteria included all AOM and UTI visits and admissions of children <15 years at clinics and hospitals across Israel from March 1, 2017, to February 28, 2022. The three pre-COVID years were defined as follows: March 1, 2017 to February 28, 2018 (first pre-COVID year), March 1, 2018-February 28, 2019 (second pre-COVID year), and March 1, 2019-February 29, 2020 (third pre-COVID year). We chose to include 3 pre-pandemic years and pool them together, because they were not characterized by any events that may have influenced AOM burdens, such as new vaccines or unusual RSV or influenza seasons. The average of the pre-COVID years served as a reference for incidence rate ratio (IRR) calculations with the first two COVID years, defined as March 1, 2020–February 28, 2021 (first COVID year) and March 1, 2021-February 28, 2022 (second COVID year). Seasons were defined as summer (June-August), fall (September–November), winter (December-February), and spring (March-May). Children were categorized into three age groups: <1 year, 1–4 years, and 5–15 years. This arbitrary age classification was based on the National Insurance Institute system.

For each unique AOM episode, we extracted age, sex, setting (ambulatory outpatient clinics or hospitals), type of visit (office or telehealth contact), complications (i.e., acute mastoiditis, subperiosteal abscess), COVID-19 status (if tested, by nasopharyngeal/oropharyngeal swab), and type of antibiotic treatment. Telehealth visits for AOM were performed by using the TytoCare™ device, which offers remote physical examination, including otoscopy, performed by caregivers that was sent online to a health provider. Positive COVID-19 status was considered if a PCR test was positive within 28 days before the diagnosis of AOM.

Study Hypothesis

We hypothesized that AOM visit rates would significantly decrease during the COVID years.

Statistical Analysis

We calculated AOM and UTI episode rates per 100,000 children for each age group and then compared monthly disease-specific for each study year. We then modeled seasonal IRRs using interrupted time-series analyses, with variation in time examined as trends, season, and cyclical patterns using trigonometrical functions. Pre-intervention period was from 03/2017, and the post-intervention period was from 03/2020 Based on quasi-Poisson regression modeling, intervention impact was estimated by comparing estimates in the post-intervention period to expected estimates if the lockdowns did not occur. The validity of the quasi-Poisson regression model was assessed by visual inspection of the correlograms (Auto Correlation and Partial Auto Correlation Functions) and residuals analysis. All statistical tests were two-sided, and we considered a result “significant” when the p-value was <0.05. All statistical analysis involved using R v.3.6.1 (http://www.R-project.org).

RESULTS

AOM Episodes

A total of 1,102,826 AOM unique episodes were identified. The median age at diagnosis was 2.0 years (IQR, 1.1–4.1), and most children were between 1-4 years (59%). Male predominance, age at presentation, and dominant age group did not change between pre-COVID and COVID years (Table 1).

TABLE 1. - Demographics of AOM Episodes, According to Study Year

	1st pre-COVID year N = 248,930	2nd pre-COVID year N = 254,623	3rd pre-COVID year N = 252,484	1st COVID year N = 110,465	2nd COVID year N = 236,324	P-value
Age, years						<0.001
Mean ± SD	3.11 ± 2.98	3.11 ± 2.95	3.15 ± 2.95	3.12 ± 3.12	3.07 ± 2.83
Median (IQR)	1.96 (1.08, 4.05)	1.98 (1.09, 4.04)	2.03 (1.12, 4.12)	1.83 (1.04, 3.97)	2.05 (1.16, 3.90)
Range	0.01, 15.00	0.02, 15.00	0.04, 15.00	0.03, 15.00	0.03, 15.00
Age subgroups, n (%)						<0.001
<1 year	54,272 (22%)	54,766 (22%)	51,118 (20%)	25,740 (23%)	45,292 (19%)
1–4 years	147,968 (59%)	152,227 (60%)	153,246 (61%)	63,510 (57%)	150,100 (64%)
5–15 years	46,690 (19%)	47,630 (19%)	48,120 (19%)	21,215 (19%)	40,932 (17%)
Gender, n (%)						<0.001
Boys	133,340 (54%)	135,660 (53%)	134,240 (53%)	59,859 (54%)	125,895 (53%)
Telehealth contact, n (%)	2,433 (1.0%)	2,542 (1.0%)	3,174 (1.3%)	5,827 (5.4%)	4,994 (2.6%)	<0.001

AOM, acute otitis media; COVID, coronavirus.

While during the three pre-COVID years the annual average of AOM episodes was 252,012 per year (19,646/100,000 children), there were only 110,465 (8,319/100,000) episodes during the first COVID year (58% decrease). During the second COVID year, there were 236,324 (18,555/100,000 children) AOM episodes (a 6% decrease from the pre-COVID years). The AOM IRR for the entire population before and after COVID-19 emergence was 0.46 (95% CI, 0.34–0.63, P < 0.001) (Table 2). The largest decrease in AOM episodes was observed among children 1–4 years of age during the first COVID year, with 16,224/100,000 children, compared to 39,480/100,000 children during the pre-COVID years (59% decrease). During the second COVID year, there was a statistically significant increase in AOM episode rates in all age groups compared to the first COVID year, nearly reaching the pre-pandemic levels. The increase was most prominent in children 1–4 years old, from 16,224/100,000 to 38,341/100,000 AOM episodes (136% increase).

TABLE 2. - IRRs for AOM and UTI, COVID Years Vs. Pre-COVID Years, According to Age Group

	All ages (0-15 years)			<1 year			1–4 years			5–15 years
	IRR	95%CI	p-value	IRR	95%CI	p-value	IRR	95%CI	p-value	IRR	95% CI	P-value
AOM	0.46	0.34, 0.63	<0.001	0.55	0.43, 0.71	<0.001	0.44	0.31, 0.62	<0.001	0.46	0.33, 0.64	<0.001
UTI	0.79	0.72, 0.86	<0.001	0.84	0.75, 0.95	0.008	0.81	0.73, 0.90	<0.001	0.76	0.68, 0.84	<0.001

AOM, acute otitis media; CI, confidence interval; IRR, incidence rate ratio; UTI, urinary tract infection.

Figure 1A shows AOM episodes during the study period. We found that monthly IRRs for AOM episodes were significantly lower in all age groups during the COVID years compared to the pre-COVID years. After seasonal adjustment, we noticed a significantly higher percentage of AOM episodes during the summer months in both COVID years compared to the pre-COVID years (28% and 27% vs. 19% on average, respectively, P < 0.001), with a concurrent decrease during the winter months (26% and 29% vs. 35% average, respectively, P < 0.001) (Figure 2). This “seasonal switch” coincided well with the timeline of COVID outbreaks and national lockdowns.

FIGURE 1.: (A) AOM Episodes Fluctuations Timeline. AOM, acute otitis media. The blue line represents the average episode number; the shaded area indicates the 95% confidence intervals. The dashed line represents the emergence of COVID-19 infection in Israel. (B) UTI Episodes Fluctuations Timeline. UTI, urinary tract infection. The blue line represents the average episode number; the shaded area indicates the 95% confidence intervals. The dashed line represents the emergence of COVID-19 infection in Israel. 

FIGURE 2.: (A) Seasonality of AOM, according to COVID period. (B) Seasonality of UTI, according to COVID period. AOM, acute otitis media; COVID, coronavirus; UTI, urinary tract infection. 

UTI Episodes

We identified 121,263 unique UTI episodes. The median age was 4.7 years (IQR 2.2–8.3), and the majority (48%) occurred in children 5-15 years of age. No significant changes were observed in age distribution and sex partition between the pre-COVID and COVID years (Figure 1B). While during the three pre-COVID years the annual average of UTI episodes was 25,795 per year (2,014/100,000 children), there were only 20,384 (1,535/100,000) UTI episodes during the first COVID year (24% decrease), and 23,492 (1,741/100,000 children) episodes during the second COVID year (13% decrease). The UTI IRR for the entire study population before and after COVID-19 emergence was 0.76 (95% CI, 0.68–0.84, P < 0.001, Table 2). Unlike AOM, we did not observe significant seasonal changes in the distribution of UTI episodes during the COVID years.

Diagnosis and Treatment

98% of AOM cases were diagnosed in community outpatient clinics (mainly by pediatricians), and the minority were diagnosed in secondary and tertiary medical centers. Telehealth visits rates for AOM diagnosis increased dramatically from a pre-COVID average of 1.0% to 5.4% and 2.6% during the first and second COVID years, respectively (P < 0.001). Treatment for AOM was consistent, with amoxicillin the most prescribed antibiotic treatment (79–81% of cases), followed by amoxicillin/clavulanic acid (10–12%) and azithromycin (6.4–7.1%).

AOM Complications

Significantly higher rates of acute mastoiditis and subperiosteal abscess were documented during the first COVID year compared with the pre-COVID years (0.5% and 0.2% vs. 0.3% and 0.1% per year, respectively, P < 0.001). There were no noticeable differences in other AOM complications before and after COVID-19 emergence. During the second COVID year, there was a decrease in AOM complications compared to the first COVID year. AOM hospitalization rates were significantly lower during the COVID years compared with pre-COVID years (0.7% and 0.6% vs. an 0.8% average, respectively, P < 0.001) (Table 3).

TABLE 3. - AOM Hospitalizations and Complications, According to Study Year

	1st pre-COVID year N = 248,930	2nd pre-COVID year N = 254,623	3rd pre-COVID year N = 252,484	1st COVID year N =110,465	2nd COVID year N = 236,324	P-value
Hospitalization, n (%)	1,773 (0.7%)	1,962 (0.8%)	2,004 (0.8%)	733 (0.7%)	1,402 (0.6%)	<0.001
Hospitalization days mean ± SD (n)	2.93 ± 3.30 (1,773)	3.01 ± 5.24 (1,962)	3.02 ± 3.83 (2,004)	2.86 ± 2.83 (733)	2.72 ± 2.40 (1,402)	0.12
SPA, n (%)	415 (0.2%)	368 (0.1%)	307 (0.1%)	249 (0.2%)	353 (0.1%)	<0.001
AM, n (%)	742 (0.3%)	801 (0.3%)	799 (0.3%)	572 (0.5%)	738 (0.3%)	<0.001
Facial palsy, n (%)	28 (<0.1%)	19 (<0.1%)	24 (<0.1%)	16 (<0.1%)	23 (<0.1%)	0.4

AM, acute mastoiditis; AOM, acute otitis media; COVID, coronavirus; SPA, subperiosteal abscess.

Concurrent AOM and COVID-19

In our cohort, 235/110,465 (0.2%) children with AOM had a recent (<28 days) COVID-19 infection during the first COVID year, and none were vaccinated. During the second COVID year, 3,892/236,324 (1.6%) children with AOM had a recent COVID-19 infection, and 3,553 (1.5%) children with AOM were vaccinated against COVID-19. Children who had a recently confirmed COVID-19 infection were less likely to present with an AOM episode when compared to children who did not have COVID-19 (OR 0.004, 95% CI, 0.004–0.0046, P < 0.0001). The majority (60%) of recently recovered COVID-19 children in our cohort were between 1–4 years (2,352/3,892) (see table, Supplemental Digital Content 1, https://links.lww.com/INF/E863).

DISCUSSION

In this population-based study, we showed a temporary sharp decrease in AOM burden during the first two COVID years in children <15 years. A sharp drop in AOM episodes characterized the first COVID year, followed by a statistically significant >two-fold increase in AOM episodes during the second COVID year, though still not reaching the pre-COVID rates. Notably, there was a “seasonal switch” during the two COVID years, despite the increased use of telehealth services. While AOM complications increased during the first COVID year, hospitalization rates decreased. A recent COVID-19 infection was not associated with a higher AOM morbidity.

Our findings align well with other reports showing a substantial reduction in AOM burden during the first COVID year,^14–18 but we showed an increase in AOM episode rate during the second COVID year, which has not been reported so far. Nevertheless, the AOM burden during the second COVID year did not reach pre-pandemic levels. We also found a statistically significant reduction in UTI episode rates during the first two COVID years, which is also in line with other reports.¹⁹ However, the decrease in AOM burden was more than two-fold. We showed that antibiotic prescription rates for AOM during the COVID era remained unchanged with a similar distribution of antibiotic categories when compared to the pre-COVID era, as in a similar Israeli study.²⁰

We suggest several explanations for our findings: (1) the enforced social distancing and restrictions measures were associated with a lower rate of respiratory viral diseases, representing true lower AOM rates during the COVID years. Several reports support this theory by reporting on a significant reduction in isolations of human metapneumovirus, parainfluenza, rhinovirus, respiratory syncytial virus, and influenza virus during the first COVID year^21–23; (2) The fluctuations in AOM burden during the COVID years coincided with the alternating lockdown/relaxation periods in Israel. The “seasonal switch” observed during the first two COVID years, with higher AOM incidence rates during the summer and lower incidence rates during the winter, can be attributed to the major lockdowns and educational institutions closure during the winter months in both COVID years. This phenomenon did not occur in UTIs during the study period; (3) public fear and avoidance of using medical services, especially during the first COVID year, may have led to a dramatic decrease in AOM burden, despite an increase in telehealth services. Telehealth encounters may be associated with AOM under-diagnosis, thus giving a pseudoreduction impression of AOM burden. Inaccurate AOM diagnosis is still common with current technologies, including at-home oto-endoscopes.^24,25 However, if indeed the avoidance of medical counseling was the main factor, one would expect a substantial parallel decrease in UTI burden as well, which was not the case; (4) The higher rate of complications during the first COVID year, especially acute mastoiditis (0.5% vs. 0.3% in pre-COVID years, P < 0.001), does not coincide with recent UK data.²⁶ This could be explained by a decrease in the total AOM episodes, therefore not reflecting a truly increased incidence of acute mastoiditis. Another explanation can be attributed to a higher threshold of seeking medical care during the first COVID period, leading to a selection bias. The fact that we did not observe different complication rates during the second COVID year may support this theory, (5) the “seasonal switch” observation could also be due to changes in care-seeking behavior due to lockdowns during the winter months, and viral interference was likely a major cause for AOM burden reduction during COVID. With how widespread SARS-CoV-2 infections were, this may have led to viral interference leading to fewer infections with other common respiratory viruses that cause AOM.

The association between COVID-19 infection and AOM is yet to be determined. We report on >27,000 children with a concurrent recent COVID-19 infection and AOM, but they contributed <8% of the total AOM episodes during the COVID years. Our numbers outnumber the single US case report on concurrent COVID-19 and AOM infection.²⁷ As an inhabitant of the nasopharynx, COVID-19 may enter the Eustachian tube and subsequently to the middle ear, as both middle ear and nasal epithelial cells show relatively high expression of angiotensin-converting enzyme 2 (ACE2) receptors, which are required for SARS-CoV-2 entry.²⁸ We cannot report on middle ear culture results (bacteriology or PCR for COVID, if performed). Thus, the observation of decreased AOM morbidity in COVID-19-recovered children still needs to be explored. To date, only one adult patient tested positive for COVID-19 from his middle ear fluid.²⁹

Strengths of our study include: (1) our study population represents the majority of the Israeli pediatric population in many health and socio-economic aspects. Therefore, we believe that our observations reflect the national AOM burden fluctuations; (2) we reported and analyzed AOM burden fluctuations during the first two COVID years, unlike previous reports covering only the first COVID year; and (3) a powered database encompassing a large population of over one million children. Limitations of our study include: (1) AOM episodes were retrieved using administrative codes; thus, we cannot confirm their accuracy because chart review was impossible. However, this is common for all extensive database studies. Regardless of the accuracy of AOM diagnosis in the retrieved visits, we saw a consistent pattern of high antibiotics prescription rate, which is in line with many AOM encounters worldwide; (2) we cannot exclude a possible reduction in AOM burden due to a home-care approach, avoidance of seeking medical attention due to quarantines, and fear of contacting COVID-19 infection. However, we believe that if any AOM complication had occurred, patients would have been referred to hospital clinics with all the cautious measures; and (3) we did not collect any parameters indicating disease severity, such as temperature, blood count, inflammatory indices (i.e., C-reactive protein), because it was out of the scope of this study.

CONCLUSION

COVID-19 emergence was associated with a decrease in the AOM burden in the Israeli pediatric population, especially in those 1–4 years of age. This decrease was more evident during the first COVID year, with more AOM episodes occurring during the summer, in line with the temporary relaxation of lockdowns imposed during the winter. These changes reflect the role of close contact and viral infection in the pathogenesis of AOM in children. The increase in AOM episodes during the second COVID year may suggest a possible change in the nature of the pandemic, but future studies are needed to clarify this trend.

ACKNOWLEDGMENTS

We thank the workers of Clalit Research Institute: Bracha Cohen, Michal Gordon, Eyal Dviri, Liya Bieber, Ye’ela Bartov, and Shir Zekaria. This study was conducted as part of the requirements for the M.D. degree of Mr. Omer Yaakov, from the Joyce and Irvin Goldman Medical School at the Faculty of Health Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel.