Trends in Prevalence of Bacterial Infections in Febrile Infants During the COVID-19 Pandemic

We conducted a multicenter cross-sectional study as part of the AAP’s Reducing Excessive Variability in Infant Sepsis Evaluation II (REVISE II) quality improvement initiative that included 97 hospitals in the United States and Canada. Infants were eligible for inclusion if they had an ED visit or hospital admission at a participating site between November 1, 2020 and March 31, 2022. This study was reviewed and deemed exempt by the AAP’s Institutional Review Board.

We included full-term, previously healthy, well-appearing infants 8 to 60 days old with a temperature of ≥38°C (100.4°F) in the ED, at home, or a clinic in the preceding 24 hours.⁴  We excluded infants with bronchiolitis, specific bacterial infections (eg, cellulitis), receipt of antibiotics or immunizations in the preceding 48 hours, and ill-appearing infants,⁴  based on an algorithm used previously to define ill-appearance.⁹  We also excluded infants who were transferred from a referring hospital if laboratory tests were obtained before transfer. Participating sites entered data on eligible infants monthly with an option to limit to 20 infants a month (for sites with >20 eligible infants per month), although not every site entered infants into the data collection system for all months and individual sites could also choose to enter more than 20 eligible infants per month. Overall, 4% of site months (the number of months with data entered multiplied by the number of sites) had exactly 20 infants entered, whereas 3% had >20 infants. Data were entered in a standardized data collection form and included demographics, the results of urine, blood, and cerebrospinal fluid (CSF) cultures (if obtained), if the infant received a treatment course for UTI or IBI, COVID-19 testing results, whether the infant had a return visit within 7 days, and if a UTI or IBI was diagnosed at the return visit.

Our primary outcome was the prevalence of UTI and IBI, overall and by CPG age group (8 to 21 days, 22 to 28 days, and 29 to 60 days old). Our secondary outcomes were changes in the odds of UTI, bacteremia without meningitis, bacterial meningitis, and IBI overall over time, prevalence according to region (Northeast, Midwest, South, West, and Canada), and whether the odds of an infant having each type of bacterial infection was associated with COVID-19 prevalence in the month of presentation. A UTI was defined by the presence of a positive urinalysis (defined as positive leukocyte esterase, positive nitrites, or >5 white blood cells per high powered field [HPF] on microscopy) with ≥10 000 colony forming units per HPF of a urinary pathogen on urine culture obtained by catheterization or suprapubic aspiration,⁴  or from a bagged sample if the infant had concomitant bacteremia (1 infant).^10,11  An IBI was defined as the growth of an a priori defined pathogen (Supplemental Table 3) in blood (bacteremia) or CSF (bacterial meningitis) culture and for which the infant received a treatment course for IBI.^4,9  Additionally, for known pathogens in the blood or CSF (eg, Group B Streptococcus, Escherichia coli) but for whom the site did not list that a treatment course for IBI was given, 2 investigators (P.L.A., and C.E.M.) reviewed the infant’s data and classified the organism as IBI or contaminant based on consensus, with a third investigator (M.E.W.) resolving discrepancies.

We calculated χ² tests to compare categorical variables and a P value < .05 was considered statistically significant. To calculate prevalence of each bacterial infection type and of COVID-19, we used the denominator of all eligible febrile infants within each categorical group who were entered into the data collection system. Logistic regression evaluating the association of infection status with month of presentation was used to determine whether the odds of each bacterial infection type (UTI, bacteremia without meningitis, bacterial meningitis, and IBI overall) varied over the course of the study period, adjusting for age. The independent variables for the model were study month and infant age in days, and the dependent variable was the log odds of each bacterial infection type. The coefficient estimate for study month was exponentiated to produce odds ratios comparing the odds of each bacterial infection type in a given study month to the month prior. These models were stratified separately by region to determine if these trends were modified by this covariate. We also evaluated the monthly trend in the average number of bacterial infections (UTI or IBI) per site (for site months in which an eligible infant was entered into the data collection system) using linear regression.

The association of COVID-19 prevalence with the odds of each bacterial infection type was tested using logistic regression models equating the infant’s outcome for each type of infection (ie, UTI, bacteremia without meningitis, bacterial meningitis, and IBI overall) to the prevalence of COVID-19 among submitted charts in the month in which the infant presented to the ED or hospital, adjusting for age. The independent variables for each model were COVID-19 prevalence in the study month and infant age in days, and the dependent variable was the log odds of each bacterial infection type. The coefficient estimate was exponentiated to produce odds ratios comparing the odds of each bacterial infection type in each month to months with a COVID-19 prevalence of 5% less. These models were also stratified separately by region to test for effect modification of this association. The 95% confidence intervals (CIs) for analyses were calculated using the standard error across the entire sample and subgroup-specific samples (eg, infants with COVID-19). All analyses were conducted using R version 4.0.4 (Vienna, Austria).

Over the 17 month study period, 9112 infants met inclusion criteria. Of these, 18% (N = 1600) were 8 to 21 days, 12% (N = 1130) were 22 to 28 days, and 70% (N = 6382) were 29 to 60 days old. Although the South was the most represented census region (38% of infants), all US census regions were represented in the sample and 5% of infants were from Canada. Overall, 61% (N = 5563) were evaluated in freestanding children’s hospitals, 29% (N = 2671) were evaluated in operationally dependent, nonfreestanding children’s hospitals, and 10% (N = 878) in general hospitals with or without a dedicated pediatric unit (Table 1).

Of the 9112 infants, 603 (6.6%; 95% CI: 6.1% to 7.2%) had a UTI, 163 (1.8%; 95% CI: 1.5% to 2.1%) had bacteremia without meningitis, and 43 (0.5%; 95% CI: 0.3% to 0.6%) had bacterial meningitis. The prevalence of each infection type decreased with increasing age, though the prevalence of bacteremia was similar among infants 8 to 21 days and 22 to 28 days old (Table 1). The overall prevalence of UTI and/or IBI decreased from 14.9% in infants 8 to 21 days to 11.2% in infants 22 to 28 days to 5.6% in infants 29 to 60 days old (Table 2). The most common pathogen for UTI and bacteremia without meningitis was Escherichia coli, at 87.4% and 62.0%, respectively, whereas Group B Streptococcus was the most common cause of bacterial meningitis (44.2%) followed by E. coli (30.2%) (Supplemental Table 3).

Temporal trends over time demonstrated a significant downtrend each month in the age-adjusted odds of an infant having UTI (OR 0.93; 95% CI: 0.91 to 0.94), bacteremia without meningitis (OR 0.90; 95% CI: 0.87 to 0.93), bacterial meningitis (OR 0.91; 95% CI: 0.86 to 0.97) and IBI overall (OR 0.90; 95% CI: 0.87 to 0.93) from November 2020 through March 2022. The downward trend in odds of UTI and IBI across months was not modified by region.

The prevalence of UTI was highest in November 2020 (11.2%) and decreased to a low of 3.0% in January 2022. The prevalence of bacteremia without meningitis was highest in November 2020 (4.5%) and decreased to a low of 0.3% in August 2021. The prevalence of bacterial meningitis was highest in February 2021 (1.7%) and decreased to a low of 0% in March 2022. IBI prevalence overall was highest in February 2021 (6.1%) and decreased to 0.4% in January 2022. In March 2022, the prevalence of UTI and IBI were 4.1% and 1.4%, respectively (Fig 1). The average number of infants with UTI or IBI per site each month decreased slightly during the study period (−0.02; 95% CI: −0.03 to −0.13).

Monthly prevalence of UTI and IBI was higher overall in infants 8 to 21 days and 22 to 28 days old than in infants 29 to 60 days old (Fig 2). Regional trends in UTI and IBI prevalence were similar to the overall trends (Fig 3).

Overall, 2102 infants (23.1%) tested positive for COVID-19. Of these infants, 16 (0.8%; 95% CI: 0.5% to 1.3%) had UTI and 8 (0.4%; 95% CI: 0.2% to 0.8%) had IBI, including 7 (0.3%; 95% CI: 0.1% to 0.7%) with bacteremia without meningitis and 1 (0.05%; 95% CI: 0.002% to 0.3%) with bacterial meningitis. For every 5% increase in the prevalence of COVID-19 among included infants in the month that a given infant presented, the age-adjusted odds of the infant having UTI (OR 0.97; 95% CI: 0.96 to 0.98), bacteremia without meningitis (OR 0.94; 95% CI: 0.88 to 0.99), and IBI overall (OR 0.94; 95% CI: 0.89 to 0.98) decreased. COVID-19 prevalence in the month of presentation was not associated with the odds of an infant having bacterial meningitis. These associations with COVID-19 prevalence were not modified by region.

In this multicenter cross-sectional study that included a convenience sample of 9112 full-term, previously healthy, well-appearing febrile infants 8 to 60 days old evaluated in the ED or admitted to the hospital over a 17 month period during the COVID-19 pandemic, the prevalence of UTI and IBI was highest in late 2020 and early 2021, and the prevalence of each infection type decreased over time to prepandemic levels through March 2022. In addition, the odds of an infant having UTI or bacteremia without meningitis were lower in months with higher COVID-19 prevalence. These prevalence estimates can help inform decision-making about LP and hospitalization in different waves of the COVID-19 pandemic.

During the first year of the COVID-19 pandemic, pediatric ED visits for febrile infants decreased,^5,6  presumably because of a lack of circulating viruses.¹²  Preliminary studies also demonstrated that the prevalence of UTIs and IBIs among infants evaluated in the ED was proportionately higher during this time. These findings raised the concern that extra caution needed to be applied in the evaluation and management of febrile infants during the pandemic on account of the known relationship between prevalence and negative and positive predictive values.^6,7  In a single-center study conducted within a tertiary care pediatric ED in Montreal, Quebec that included 324 febrile infants ≤90 days old evaluated from March 2020 to March 2021, the prevalence of UTI or IBI was 20.4% among febrile infants ≤90 days old and the prevalence of IBI alone was 3.4%, increased from 10% and 1% prepandemic, respectively.⁶  At 6 mid-Atlantic community hospitals affiliated with an academic children’s hospital, the prevalence of UTI, bacteremia, and bacterial meningitis among 95 febrile infants ≤56 days old who presented between March 2020 and February 2021 were 19%, 14%, and 4%, respectively, increased from 9%, 2%, and 0.2% in the prepandemic years.⁷  The higher prevalence of UTIs and IBIs was postulated to be secondary to a lower incidence of non-COVID-19 respiratory viral infections during lockdown and school closures.^7,13  Our data from a large, geographically diverse sample of febrile infants showed that the prevalence of UTI and IBI was similar to the reported prevalence before the COVID-19 pandemic (∼7% for UTI and ∼2% for IBI, historically),^2,3  and that prevalence was even lower at the end of the study period in March 2022 (4.1% and 1.4%, respectively). The decrease in prevalence later in the COVID-19 pandemic may be because of a higher incidence of viral respiratory infections rebounding in 2021 and early 2022,^14,15  as our data did not demonstrate a clinically significant decrease in the average number of monthly bacterial infections per site.

The AAP CPG for well-appearing febrile infants emphasizes that the clinician’s and parent’s risk tolerance should inform decision-making, specifically for LP and hospitalization decisions for infants 22 to 28 days with normal inflammatory markers and infants 29 to 60 days with abnormal inflammatory markers. For these groups of infants, the AAP CPG recommends that clinicians use SDM with parents.⁴  Shared decision-making is a process whereby the clinician explains that there are 2 or more reasonable management options and then presents the potential risks and benefits of each option. The parents then express their values and preferences, and a joint decision is made.¹⁶  To present the potential risks of performing versus not performing an LP or of hospitalization versus ED discharge for febrile infants, clinicians must know the prevalence estimates of UTIs and particularly IBIs. The prevalence estimates from our study demonstrate that the risk of UTI, bacteremia, and bacterial meningitis among febrile infants is higher in time periods when respiratory viral infections are likely of low prevalence, potentially because of pandemic mitigation measures, but then return to baseline risk, or lower, when these measures end. The changing prevalence over time may impact clinicians’ and parents’ risk tolerance about LP and hospitalization in infants for whom SDM is recommended by the AAP CPG.

We also found that higher monthly COVID-19 prevalence was associated with lower odds of an infant having UTI or bacteremia without meningitis. This finding is likely caused by the lifting of mitigation measures during our study period, resulting in a higher proportion of febrile infants presenting to care with COVID-19 in periods of higher community COVID-19 prevalence (eg, in January 2022 during the Omicron wave), and a low prevalence of UTI and bacteremia among febrile infants who tested positive for COVID-19. This association may help inform risk discussions during SDM with parents about LP and hospitalization during subsequent waves of the COVID-19 pandemic. An as example, the positive predictive value of abnormal inflammatory markers for bacteremia without meningitis in a 29 to 60 day old infant may be lower in months when the prevalence of COVID-19 is higher and the odds of bacteremia is lower, which might impact the clinician’s or parent’s risk tolerance in deciding whether to hospitalize the infant. Although the odds of an infant having bacterial meningitis did not vary with monthly COVID-19 prevalence, this finding is likely caused by the rarity of bacterial meningitis among febrile infants,^3,4  regardless of COVID-19 prevalence. Further study is needed to inform the individual risk assessment of febrile infants with COVID-19.

Our study has several limitations. First, determination of exclusion criteria, such as ill-appearance and the presence of bronchiolitis, may be limited when assessed retrospectively. Second, participating sites were instructed to enter data on eligible infants monthly with an option to limit to 20 infants per month, though it was an infrequent occurrence to have exactly 20 infants entered into the data collection system. Additionally, some sites did not enter infants into the data collection system for all months of the study period. Our prevalence estimates may therefore not reflect true prevalence among the overall population of full-term, previously healthy, well-appearing febrile infants, but rather represent the prevalence in a convenience sample of eligible infants presenting to the ED. Third, we were not able to conclude whether the downward temporal trends in prevalence were secondary to changes in the prevalence of circulating viruses or to changes in the incidence of UTI and IBI among infants in the community. We also do not know the denominator of total ED visits during the study period, and we were not able to assess whether visit rates for febrile infants changed during the study period. Fourth, although our definition of IBI was the presence of an a priori pathogen that was treated as an IBI, a definition used previously,⁹  it is possible that some organisms were misclassified as contaminants, resulting in an underestimate of prevalence. Fifth, we calculated monthly COVID-19 prevalence using the denominator of all febrile infants entered into the data collection system that month, which reflects, but does not equate to, community COVID-19 prevalence. Sixth, 90% of infants in our study were evaluated at children’s hospitals, which may not be representative of all febrile infants. However, a strength of our study is the inclusion of infants at geographically diverse children’s hospitals and some general hospitals, including sites in Canada.

In this multicenter, cross-sectional study, we found a peaking prevalence of UTI and IBI in full-term, previously healthy, well-appearing febrile infants 8 to 60 days old in the early pandemic with decreasing prevalence through March 2022, with some association between COVID-19 waves and the odds of an infant having UTI and bacteremia without meningitis. With the potential for future pandemics, understanding the impact of mitigation measures upon bacterial infection prevalence in febrile infants is a key component to informed decision-making between clinicians and parents.

The following collaborators in the AAP REVISE II QI Collaborative are acknowledged as group authors for this study: Chisom Agbim, MD, MSHS (Children’s Hospital Colorado, Aurora, CO), M. Alex Ahearn, MD, MPH (University of North Carolina Rex Hospital, Raleigh, NC), Harinder Kaur Bawa, MD (Hackensack Meridian School of Medicine, Hackensack, NJ), Heather J Becker, MD (Bridgeport Hospital, Bridgeport, CT), Amanda Benaderet, MD, MPH (Indiana University School of Medicine, Indianapolis, IA), Eric Biondi, MD (Johns Hopkins Children’s Center, Baltimore, MD), Sarah Bram, MD (Washington University School of Medicine, St. Louise, MO), Jaimie Brandley, DO (Valley Children’s Hospital, Madera, CA), Jacqueline G. Brunetto, MD (Unterberg Children’s Hospital at Monmouth Medical Center, Long Branch, NJ), Claudia Friederike Busse, MD (Lehigh Valley Reilly Children’s Hospital, Allentown, PA), Chien-Rong Chen, DO (Harbor-UCLA Medical Center, Torrance, CA), Melanie Cheng, MD (Kings County Hospital, Brooklyn, NY), Melissa S. Chladek, MD (Memorial Hermann the woodlands, Houston, TX), Nancy Chung, MD (University of California Riverside, Riverside, CA), Erin Clark, MD (Eastern Maine Medical Center, Bangor, ME), Katie D’Ardenne, MD (Bozeman Health, Bozeman, MT), Nickolas E. Dawlabani, MD (K. Hovnanian Children’s Hospital at Jersey Shore University, Neptune, NY), Guillermo A. De Angulo, MD, MSCR (Stanford University School of Medicine, Palo Alto, CA), Andrew Ellis, MD (Mercy children’s Hospital St. Louis, St. Louis, MO), Amy Ferguson, MD (Unity Point Health/Blank Children’s Hospital, Des Moines, IA), Jennifer Falcon, MD (Mary Washington Hospital, Fredericksburg, VA), Allison Gauthier, MD (St. Luke’s Hospital Boise, Boise, ID), David E. Gilliam, MD (University of Mississippi Medical Center, Jackson, MS), Katie Giordano, DO (Nemours Children’s Hospital, Wilmington, DE), Heidi G. Greening, DO (Advocate Children’s Hospital, Park Ridge, IL), Amy Gonzalez, MD (University of Texas Medical Branch, Galveston, TX), David Guernsey III, DO (Maimonides Medical Center, Brooklyn, NY), Nisha Gupta, MD (Inova Children’s Hospital, Falls Church, VA), Andrea Hadley, MD (Spectrum Health Helen DeVos Children’s Hospital, Grand Rapids, MI), Marissa A. Hendrickson, MD (University of Minnesota Medical School, Minneapolis, MN), Alana Ju, MD (Children’s Hospital Los Angeles, Los Angeles, CA), Patti Jo Jaiyeola, MD (St. Luke’s University Health Network, Bethlehem, PA), Sonal Kalburgi, DO, MSHS (Holy Cross Hospital, Children’s National, Silver Springs, MD), Aditi Kamat, MD (Maria Fareri Children’s Hospital, Valhalla, NY), Ann Kane, MD (Johns Hopkins University School of Medicine, Baltimore, MD), Stephanie A. Karnik, MD (Children’s Mercy Hospital, Kansas City, MO), Sanghee Kim, MD (Ann & Robert H. Lurie Children’s Hospital, Chicago, IL), Gift Kopsombut, MD (Nemours Children’s Hospital, Orlando, FL), Olena Kostyuk, MD (Children’s National Medical Center, Washington DC), Anika Kumar, MD (Cleveland Clinic Children’s, Cleveland, OH), Bhumi Kumar, DO (Advent Health Tampa, Tampa, FL), Alicia La, DO (Huntington Hospital, Pasadena, CA), Jamie Laughy, MD (Oklahoma Children’s Hospital, Oklahoma City, OK), Rianna C. Leazer, MD (Children’s Hospital of The King’s Daughters, Milwaukee, WI), Begem Lee, MD (University of California San Diego, San Diego, CA), Jennifer Lee, MD (Columbia University Irving Medical Center, New York, NY), Larissa Lester Truschel, MD, MPH (Duke University Hospital, Duke University School of Medicine, Durham, NC), Sydney Levy, BS (University of Florida College of Medicine, Gainesville, FL), Tian Liang, MD (New York University Langone Health, New York, NY), Michelle Long, MD (Children’s Hospital of Eastern Ontario, Ottawa, Ontario CA), Joseph D. Lynch, MD (West Virginia University Children’s Hospital, Morgantown, WV), Deborah Margulis, MD (Northwestern Lake Forest Hospital, Lake Forest, IL), Amit Mehta, MD (University of Texas Southwestern, Dallas, TX), Emily J. McCarty, MD, MA (Kootenai Health, Coeur d’Alene, ID), Lauren McGovern, MD (Tidal Health Peninsula Regional Health, Salisbury MD), Kristen Newcomer, DO (Studer Family Children’s Hospital – Sacred Heart, Pensacola, FL), Nolan Nielsen, MD (Richmond University Medical Center, Staten Island, NY), Eberechi I. Nwaobasi-Iwuh, MD (Goryeb Children’s Hospital, Morristown, NJ), Cheryl Okado, MD (Kapiolani Medical Center for Women and Children, Honolulu, HI), Andrew W. Osten, MD (SUNY Upstate Medical University, Syracuse, NY), Snehal T. Patel, MD, MHS (University of Maryland Children’s Hospital, Baltimore, ME), Alison Payson, MD (Nicklaus Children’s Hospital, Miami, FL), Kaitlyn Phuong Le, MD (Sutter Medical Center Sacramento, Sacramento, CA), Mary Jane Piroutek, MD (Children’s Hospital Orange County, Orange County, CA), Emily Powers, MD, MHS (Yale School of Medicine, New Haven, CT), Giusy Romano-Clarke, MD (Massachusetts General Hospital, Boston, MA), Brian Sanders, MD, MHCM (Salem Hospital, Salem, MA), Patrick Seitzinger, MD (University of Saskatchewan, Saskatoon, Saskatchewan Canada), Adam Sivitz, MD (Newark Beth Israel Medical Center, Newark, NJ), Hannah Sneller, MD (Children’s Hospital and Medical Center, Omaha, NE), Maria Aurora Soriano, MD, MBA (Sinai Hospital of Baltimore, Baltimore, ME), Ellen P. Soufleris, MD (Moses H. Cone Memorial Hospital, Greensboro, NC), Karthik Srinivasan, MD (Cook Children’s Medical Center, Fort Worth, TX), Mary Elizabeth Swift-Tayler, MD (Providence St. Vincent Medical Center, Portland, OR), Sara Szkola, MD (Stanford Healthcare Valleycare, Pleasanton, Ca), Daina Thomas, MD (University of Alberta, Edmonton, Alberta CA), Jennifer Thull-Freedman, MD, MSc (University of Calgary Cumming School of Medicine, Calgary, Alberta CA), Andrea Van Wyk, MD (Gundersen Health System, La Crosse, WI), Lynsey Vaughan, MD, MSHS (Dell Children’s Medical Center, Austin, TX), Yenimar Ventura-Polanco, MD (University of Florida College of Medicine – Jacksonville, Jacksonville, FL), Christopher Vlasses, MD (UCSF Benioff Children’s Hospital Oakland, Oakland, CA), Travis Walker, MD (Johns Hopkins All Children’s Hospital, St. Petersburg, FL), Rabbia Waris, MD (Stanford University, Palo Alto, CA), Lindsay Weiss, MD (Joe DiMaggio Children’s Hospital, Hollywood, FL), Wendi-Jo Wendt, MD (Medical College of Wisconsin, Milwaukee, WI), Lindsay Zremba, MD (University Hospitals Rainbow Babies, Cleveland, OH).

AAP

American Academy of Pediatrics

CI

confidence interval

CPG

clinical practice guideline

CSF

cerebrospinal fluid

ED

emergency department

HPF

high powered field

IBI

invasive bacterial infection

LP

lumbar puncture

OR

odds ratio

SDM

shared decision-making

UTI

urinary tract infection