Rhinovirus in Febrile Infants and Risk of Bacterial Infection
BACKGROUND: Febrile infants with viral respiratory infections have a reduced risk of bacterial infection compared with virus-negative infants. The risk of concomitant bacterial infection in febrile infants positive for human rhinovirus (HRV) by polymerase chain reaction (PCR) is unknown.
METHODS: Infants 1–90 days old managed using the care process model for well-appearing febrile infants and with respiratory viral testing by PCR (RVPCR) in the emergency department or inpatient setting of 22 hospitals in the Intermountain Healthcare system from 2007-2016 were identified. Relative risk (RR) of bacterial infection was calculated for infants with HRV, non-HRV viruses, or no virus detected.
RESULTS: Of 10 964 febrile infants identified, 4037 (37%) had RVPCR. Of these, 2212 (55%) were positive for a respiratory virus; 1392 (35%) for HRV alone. Bacterial infection was identified in 9.5%. Febrile infants with HRV detected were more likely to have bacterial infection than those with non-HRV viruses (7.8% vs 3.7%; P < .001; RR 2.12 [95% CI 1.43–3.15]). Risk of urinary tract infection was not significantly different for HRV-positive infants at any age, nor was risk of invasive bacterial infection (IBI; bacteremia and/or meningitis) meaningfully different for infants 1–28 day olds. Infants 29–90 days old with HRV had a decreased likelihood of IBI (RR 0.52 [95% CI 0.34–0.80]).
CONCLUSIONS: HRV is common in febrile infants. Detection did not alter risk of concomitant urinary tract infection at any age or risk of IBI in infants 1–28 days old. HRV detection may be relevant in considering risk of IBI for infants 29–90 days of age.
- BSI —
- bloodstream infection
- CI —
- confidence interval
- CSF —
- cerebrospinal fluid
- DFA —
- direct fluorescent antibody
- ED —
- emergency department
- EV —
- HRV —
- human rhinovirus
- IBI —
- invasive bacterial infection
- PCR —
- polymerase chain reaction
- RR —
- relative risk
- RSV —
- respiratory syncytial virus
- RVPCR —
- respiratory viral testing by multiplex PCR
- UTI —
- urinary tract infection
What’s Known on This Subject:
Rhinoviruses are ubiquitous and can be detected in symptomatic and asymptomatic individuals by polymerase chain reaction. Febrile infants are at risk for bacterial infection. It is unknown whether human rhinovirus (HRV) detection may indicate a lower risk of bacterial infection in febrile infants.
What This Study Adds:
HRV detection was common among well-appearing febrile infants. HRV detection did not impact risk of concurrent urinary tract infection at any age or invasive infection in infants 1 to 28 days old. These data have implications for febrile infant management.
The management of a well-appearing febrile infant 1 to 90 days old often includes rapid medical evaluation, sterile site cultures, and empirical antibiotic therapy.1 This strategy is based on the need for early recognition and treatment in a population in which fever may be the only sign of a life-threatening infection.2
Whereas ∼10% of febrile infants have a bacterial infection, most have a viral illness.2,3 Previous studies have demonstrated that infants with laboratory-confirmed viral infections, particularly respiratory infections, are less likely to have a bacterial infection than infants in whom no viral infection is detected.3–6 In most instances, researchers of these reports used rapid antigen tests, direct fluorescent antibody (DFA) testing, or viral cultures for the identification of viral infections. The introduction of respiratory testing with the multiplex polymerase chain reaction (PCR) allows for the detection of more viral pathogens and may increase sensitivity.7 Pathogens that have not been detectable previously, including human rhinovirus (HRV), can be detected by using PCR.
The clinical interpretation of HRV detection by PCR is challenging. Although HRV is a known cause of respiratory tract infection, it is also frequently detected in asymptomatic children or codetected with other pathogens.8–11 Prolonged shedding of HRV has been documented.9,11,12 Although data from studies of influenza and respiratory syncytial virus (RSV) have been used to suggest that a more limited evaluation and/or early discharge could be considered for febrile infants who are positive for these infections,1,2,13,14 no data exist to describe the risk of bacterial infection in febrile infants who are HRV-positive.
Our objectives in this study were to evaluate the frequency of bacterial infections, including urinary tract infection (UTI), bloodstream infection (BSI), or meningitis (BSI and meningitis are collectively referred to as invasive bacterial infection [IBI]) in well-appearing febrile infants with HRV and non-HRV respiratory viruses and infants with no respiratory virus detected.
The institutional review boards of the University of Utah and Intermountain Healthcare (both in Salt Lake City, UT) approved this study and granted a waiver of informed consent.
Location and Setting
This retrospective observational study was performed within Intermountain Healthcare, a not-for-profit, integrated health care system providing inpatient care for >90% infants <1 year old in Utah. Infants who received care at 1 of 22 Intermountain hospitals (including Primary Children’s Hospital, a 289-bed, tertiary-care, pediatric referral center) as well as regional medical centers and community hospitals were included. Primary Children’s Hospital and the regional medical centers provide care for >80% of febrile infants in the state.1
Selection of Infants
Febrile infants were identified from the Intermountain Enterprise Data Warehouse. The Enterprise Data Warehouse is shared across all facilities and contains clinical, laboratory, radiographic, and administrative data for all encounters. Infants who were evaluated at an Intermountain Healthcare facility by using the evidence-based care process model for well-appearing febrile infants 1 to 90 days old were identified by using a previously validated algorithm.15
Infants were included if they had been evaluated in the emergency department (ED) or had an inpatient admission within the first 24 hours after presentation with fever in the outpatient setting. We excluded infants who were evaluated exclusively as outpatients because these infants infrequently underwent respiratory testing by PCR (<15%; data not shown). The main analyses in this study were focused only on infants with respiratory testing by PCR because this is the only way to detect HRV.
Infants who were evaluated between August 21, 2007 and August 20, 2016, after the institution of respiratory viral testing by multiplex PCR (RVPCR), were included. Data collected included demographics and results of the evaluation for infection, including respiratory viral testing and bacterial cultures.
RVPCR was performed at the Intermountain Central Laboratory by using the Luminex xTAG Respiratory Viral Panel (Luminex Corporation, Madison, WI) beginning in August 2007. In 2012, respiratory testing by using the FilmArray Respiratory Panel (BioFire Diagnostics LLC, Salt Lake City, UT) was introduced at Primary Children’s Hospital, the Central Laboratory, and regional centers, replacing the Luminex xTAG. Viruses detected with these multiplex panels included the following: adenovirus; coronaviruses HKU1, NL63, 229E, and OC43; human metapneumovirus; influenza A and B; parainfluenza 1 through 4; RSV; and HRV and/or enterovirus (EV). For the purposes of this analysis, positive HRV and/or EV results were classified as HRV.16,17
Identification of Bacterial Infection
We included infants in the study regardless of whether a full complement of bacterial cultures (blood, cerebrospinal fluid [CSF], and urine) was performed. The electronic medical records of all infants were managed for 72 hours after their initial evaluation to ensure that there were no bacterial infections identified during subsequent evaluations. This time frame has been used since 2008 for quality monitoring of the Intermountain care process model for well-appearing febrile infants and captures the vast majority of infant readmissions for missed bacterial infection.1 Overall, 80.3% of infants in the baseline cohort had at least 1 bacterial culture performed. Infants who had no bacterial cultures on initial evaluation and no subsequent positive culture results within 72 hours were considered to be negative for bacterial infection. Culture results from 24 hours before through 72 hours after the febrile infant evaluation were included. All cultures that were flagged as equivocal or significant were manually reviewed and adjudicated by a microbiologist or pediatric infectious diseases physician by using an algorithm and clinical judgment (E.K.K. and C.L.B.).
Logic for classifying UTI included the following: <10 000 colony-forming units per milliliter was considered contamination, organisms present at 10 000 to 49 999 colony-forming units per milliliter were considered equivocal, and organisms present at >50 000 were considered significant (UTI present), although skin flora was considered a contaminant.18
IBI (BSI and Meningitis)
Logic for classifying IBI included the following: skin flora (coagulase-negative Staphylococci, viridans streptococci, Bacillus spp., Corynebacterium spp., or Propionibacterium spp.) were considered contaminants unless 2 separate culture results were positive with the same organism within 48 hours. All other organisms detected in blood or CSF cultures were considered significant.
The statistics for this article were generated by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). Figure 1 was created by using R 3.1.319 by using the forest-plot package.20 χ2 tests were used to compare categorical values. Statistical significance for the 9 respiratory virus comparisons were assessed at a Bonferroni adjusted significance level of 0.05/9 = 0.006. All other statistical tests were considered significant at 0.05.
To identify the risk of bacterial infection on the basis of the results of respiratory testing, we compared infants with (1) any non-HRV respiratory virus detected by multiplex PCR (adenovirus; coronaviruses HKU1, NL63, 229E, and OC43; human metapneumovirus; influenza A and B; parainfluenza viruses 1 to 4; or RSV) with or without HRV coinfection, (2) HRV detected alone by multiplex PCR, and (3) no respiratory virus detected by multiplex PCR. We calculated relative risks (RRs) with 95% confidence intervals (CIs) for all infants 1 to 90 days old as well as stratified by age (1–28 days old and 29–90 days old). For instances in which there were 0 cases of bacterial infection in 1 of the comparator groups, the RR estimate and CI were calculated by using exact methodology (specifically, the RR score statistic).21
During the study period, 10 964 well-appearing febrile infants were evaluated in the ED or admitted to the hospital within 24 hours of outpatient evaluation. Of these, 4037 (37%) had RVPCR. Demographics and clinical characteristics of infants with and without RVPCR performed are shown and compared in Table 1. The majority (72%) of infants were 29 to 90 days old.
Infants with RVPCR were more likely to be admitted to the hospital within 24 hours of presentation (58% vs 51%; odds ratio 1.4; 95% CI 1.24–1.45). The proportion of infants who were positive for a respiratory virus was higher among those tested by RVPCR compared with conventional methods (55% vs 38%; P < .001). The remainder of the study focused only on infants in whom RVPCR was performed.
Identification of Respiratory Viruses
RVPCR results are shown in Table 2. Overall, 2212 of 4037 (55%) infants were positive for a respiratory virus. Infants 29 to 90 days old had a significantly higher rate of respiratory virus detection (1788 of 2946; 61%) than did infants 1 to 28 days old (424 of 1091; 39%; P < .001).
HRV was the most common virus detected (1525 of 4037; 38%), and HRV was detected alone in 1392 (35%) infants. A higher proportion of infants 29 to 90 days old (1226 of 2946; 42%) were HRV-positive when compared with infants 1 to 28 days old (299 of 1091; 27%; P < .001). A non-HRV respiratory virus was detected in 820 (20%) infants. The frequency of detection of individual non-HRV viruses ranged from 0.7% for adenovirus to 7% for RSV. All viruses, with the exception of RSV and influenza B, were more commonly detected among infants 29 to 90 days old.
Bacterial infections were detected in 9.5% of infants (Fig 1, Table 3). Infants 1 to 28 days old were more likely than those 29 to 90 days old to have a bacterial infection (11.5% vs 8.8%; P = .01). When compared with infants with non-HRV viruses detected, the RR of bacterial infection was more than twice as high for infants with HRV detected alone (3.5% vs 7.5%; RR 2.12; 95% CI 1.43–3.15) and >3.5 times as high for infants who were negative by RVPCR (3.5% vs 13%; RR 3.66; 95% CI 2.53–5.31).
Types of Bacterial Infection and Respiratory Viral Status
Table 3 shows the overall frequency of infection and the frequency of specific bacterial infections (UTI, IBI, BSI, and meningitis alone) stratified by respiratory viral status and age. Among infants 1 to 28 days old, the detection of neither non-HRV nor HRV viruses was associated with a statistically lower risk of UTI when compared with virus-negative infants. There were no infants 1 to 28 days old with a non-HRV virus detected with IBI, compared with 6.1% of those with no virus detected and 2.1% of infants with HRV. Compared with infants with no virus, HRV detection in this age group was associated with a statistically decreased risk of IBI (RR 0.41; 95% CI 0.19–0.88). The risk of BSI alone was lower when HRV was detected, but it was not statistically different (RR 0.49; 95% CI 0.23–1.03). No infant 1 to 28 days old with any respiratory virus detected had bacterial meningitis.
For infants 29 to 90 days old, non-HRV detection was associated with a decreased risk of UTI compared with infants with no virus detected (2.5% vs 9.1%; RR 0.36; 95% CI 0.23–0.56); HRV detection had a smaller impact on the risk of UTI (6.1% vs 9.1%; RR 0.78; 95% CI 0.65–0.95). For infants 29 to 90 days old, both non-HRV and HRV detection were associated with a decreased risk of IBI (RR 0.35 [95% CI 0.17–0.70] and RR 0.52 [95% CI 0.34–0.80], respectively).
The majority of cases of meningitis (10 of 12; 83%) occurred in infants who were negative for respiratory viruses by RVPCR. One case occurred in an infant 79 days old who was HRV-positive, and 1 case occurred in a 54-day-old infant who was positive for coronavirus OC43.
Our data show that respiratory viruses, especially HRV, are commonly detected among well-appearing infants who are evaluated for fever. More than half of the infants who were tested with RVPCR were positive for 1 or more respiratory viruses. HRV was detected alone in 35% of infants, more than all other respiratory viruses combined. HRV was detected more frequently among infants 29 to 90 days old than in younger infants. The detection of respiratory viruses other than HRV by PCR was associated with an almost fourfold reduction in the frequency of bacterial infection compared with infants with no virus detected, confirming previous observations made with nonmolecular methods. In contrast, the frequency of bacterial infection was only modestly decreased among infants with HRV. Among infants 1 to 28 days of age, the likelihood of bacterial infection was not meaningfully reduced when HRV was detected. Among older infants, the detection of HRV was not associated with a decreased frequency of UTI but was associated with a modest decrease in the frequency of IBI. Combined with other clinical data, HRV detection in older infants could be useful in decision-making.
Although there are limited data on the frequency of respiratory virus detection in infants 1 to 90 days old, a number of studies have shown that respiratory virus detection is common in young children, particularly infants <1 year old.9,11,12,22 Longitudinal surveillance studies report the detection of at least 1 respiratory virus in >50% of weekly tests.11 In our study of febrile infants, ∼55% of infants were positive for 1 or more respiratory viruses, and 38% had HRV detected. RSV, the second most common virus detected, was found in only 7% of infants. These data are similar to surveillance studies in which non-HRV respiratory viruses are generally detected in <5% of weekly tests, whereas HRV is found in >30%.9,11
Most multiplex PCR–based testing that includes HRV, such as the Luminex and FilmArray panels used in our study, are qualitative and do not distinguish between HRV and EV.23,24 In 2 studies in which sequencing has been done in either adults or children to confirm HRV or EV, the majority of HRV and/or EV detections (80%–90%) are found to be HRV.16,17 Confirmation of EV infection was seen in only 2% of detections.16,17 Additional studies in which specific PCR for either HRV or EV are performed document that a much higher proportion of detections are HRV.25 In a large study of infants with nasal sampling performed monthly and during visits for upper respiratory tract infection, 244 detections of HRV and 20 detections of EV were identified, indicating that HRV accounted for 92% of these picornaviral infections.25
The decreased risk of bacterial infection in febrile infants with viral infections has been well documented.3–6 Previous studies have shown a decreased risk of bacterial infection with both RSV and influenza virus as well as systemic EV.3–6 These studies have been used to suggest that a more limited evaluation and/or early discharge could be considered for febrile infants who are positive for specific viral infections.1,2,13,14 Our findings are consistent with previous results; in our study, the detection of a non-HRV virus was associated with an RR of 0.27 (95% CI 0.19–0.39) for bacterial infection when compared with infants who were virus-negative.
Studies of EV detection from sterile sites, including blood and CSF, have used PCR-based detection6 and demonstrated a lower risk of bacterial infection in infants in whom sterile site EV PCR results are positive.3,26 Given the prolonged shedding of EVs from nonsterile sites and the fact that the vast majority of HRV and/or EV detections from respiratory samples are actually HRV, we discourage practitioners from using the respiratory PCR to detect EV in the febrile infant population. An exception is in those infants who may be severely ill with respiratory symptoms or present with paralysis. In these instances, EVD68 may be present; however, full identification of this infection requires additional testing from sterile sites and sequencing to distinguish EVD68 from poliovirus or other EVs.27
Published studies of RSV and influenza and risk of bacterial infection have primarily involved viral detection by DFA or rapid antigen detection.3–5 PCR-based detection is generally more sensitive and does not distinguish among active infection, persistence of viral nucleic acid, or asymptomatic carriage. Our study demonstrates that the risk of bacterial infection remains markedly decreased with a PCR-based detection of respiratory viruses other than HRV. Surveillance data indicate that for the majority of non-HRV viruses, the duration of detection by PCR is ≤2 weeks.11 This may allow for greater confidence in associating a positive detection with clinical signs (including fever) under evaluation. Viral testing by PCR is rapid and has a role to play in the evidence-based evaluation and management of febrile infants.
Studies have shown that ∼50% of HRV detections are associated with symptoms.9,11 Although young children are more often symptomatic than older children, young infants (<6 months old) are frequently asymptomatic.11,25 The high rate of detection and frequency of asymptomatic infection make the interpretation of HRV detection difficult in the context of the febrile infant.
In our study, HRV detection was associated with an ∼40% decrease in the rate of bacterial infection overall (RR 0.58; 95% CI 0.47–0.72). The association of HRV detection with concomitant bacterial infection was dependent on both the age of the infant and the type of infection. Two-thirds of the bacterial infections were UTIs. Although there was a significantly lower rate of UTI among children with HRV, the frequency of UTI remained at >6% at any age. This finding is consistent with current literature suggesting that febrile infants with a respiratory infection should still be evaluated for UTI.2,28–31
Overall, infants with HRV detected had a statistically lower rate of IBI (1.6%) compared with those with no viral infection (4.8%). However, for those 1 to 28 days old, 2.1% of HRV-positive infants had an IBI, all of which were due to a BSI. The frequency of BSI was lower than that seen in infants with no virus detected (5.1%) but was not statistically different. This finding contrasted with the finding that no infant 1 to 28 days old with a non-HRV virus detected had a BSI or IBI. Although the smaller number of 1- to 28-day-old infants decreased our power to detect significant differences in this group, our data do not provide evidence to suggest that HRV detection should play a role in the risk stratification of very young infants.
Infants 1 to 28 days old have generally been considered high risk.32,33 Our data further validate the increased risk for bacterial infection in febrile infants 1 to 28 days old and demonstrate the inability of HRV detection to allow for adequate risk stratification in this age group. We continue to recommend a full evaluation, including cultures of blood, urine, and CSF, for all febrile infants 1 to 28 days old. Admission to the hospital remains prudent pending bacterial culture results. The majority of bacterial cultures from febrile infants will be positive for pathogens by 24 hours.34 For well-appearing infants 1 to 28 days old with either HRV or a non-HRV virus detected, the viral results should be considered in light of other laboratory data, such as inflammatory markers and CSF profile, in deciding on appropriate management, including the initiation and duration of antimicrobial therapy.
For infants 29 to 90 days old, the risk of an IBI was ∼50% lower when HRV was detected (frequency of 1.4% vs 4.0%). This was not as large a decrease as that seen in infants with non-HRV viruses detected (frequency of 1.0%; 65% decrease); however, HRV detection could be considered when managing febrile infants in this age range, including the decision to perform a lumbar puncture or to admit.
Our study has several limitations. The multiplex PCR testing used to identify HRV in our study does not distinguish between HRV and EV.23,24 We did not do secondary testing to specifically confirm that positive detections were HRV, and it is possible that a small proportion (∼2%) of the infants included in this study were infected or colonized with respiratory EVs.16,17,25 None of the well-appearing study infants had clinical evidence of infection with EVD68. We included only infants who were seen in the ED or inpatient setting and had RVPCR performed. Conclusions from this study therefore may not be generalizable to infants who were evaluated only in the ambulatory setting. More infants in the RVPCR group were admitted to the hospital within 24 hours of presentation, suggesting they may have been more ill. We did not have data on respiratory symptoms and could not correlate symptomatology with the risk of bacterial infection. Rates of bacterial infection by specific virus type other than HRV were too low for statistical analysis. Additionally, testing for many of these viruses is performed by using conventional, non-PCR diagnostics, and rates have been reported elsewhere.3–5 Sample sizes were small for comparing rates of UTI, IBI and BSI with viral detection stratified by age group; in particular, the 1- to 28-day-old age group had low numbers for IBI (n = 47) and BSI (n = 40). The frequency of meningitis was low in our cohort (12 of 4037 [0.3%]) but was similar to other reports.35,36 Ten of 12 infants with meningitis were respiratory virus–negative by PCR.
HRV detection is common among well-appearing infants 1 to 90 days old undergoing evaluation for fever. The detection of HRV does not change the likelihood of UTI in febrile infants at any age or the risk of concomitant invasive infection in those 1 to 28 days old. Infants 1 to 28 days old with HRV should be managed as high risk, which is similar to all others in this age group. HRV detection may be relevant in considering the risk of IBI for infants 29 to 90 days of age, although the risk reduction with HRV is smaller than that seen with other respiratory viruses. It is possible that well-appearing infants 29 to 90 days old with HRV and without evidence of a UTI could be managed expectantly.
We acknowledge Dr Chris Stockmann, PhD, a valued member of the University of Utah’s Division of Pediatric Infectious Diseases, who died before the publication of this article. Dr Stockmann contributed to the design of this study as well as the initial statistical analyses. He died in a rock-climbing accident in 2016 at 28 years old but made significant contributions to the health of children through his research during his brief career.
- Accepted November 15, 2017.
- Address correspondence to Anne J. Blaschke, MD, PhD, Division of Pediatric Infectious Diseases, Department of Pediatrics, University of Utah, PO Box 581289, Salt Lake City, UT 84158. E-mail:
FINANCIAL DISCLOSURE: Drs Blaschke and Byington have intellectual property and receive royalties from BioFire Diagnostics LLC through the University of Utah; the other authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Dr Byington received support from the National Institutes of Health and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (K24HD047249), the National Institutes of Health and the National Center for Advancing Translational Sciences (1UL1TR001067), and the H. A. and Edna Benning Presidential Endowment. Funded by the National Institutes of Health (NIH).
POTENTIAL CONFLICT OF INTEREST: Drs Blaschke and Byington have intellectual property and receive royalties from BioFire Diagnostics LLC through the University of Utah; the other authors have indicated they have no potential conflicts of interest to disclose.
- Byington CL,
- Enriquez FR,
- Hoff C, et al
- Krief WI,
- Levine DA,
- Platt SL, et al; Multicenter RSV-SBI Study Group of the Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics
- Levine DA,
- Platt SL,
- Dayan PS, et al; Multicenter RSV-SBI Study Group of the Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics
- Jansen RR,
- Wieringa J,
- Koekkoek SM, et al
- Loeffelholz MJ,
- Trujillo R,
- Pyles RB, et al
- Gesteland P,
- Korgenski K,
- Raines B,
- Byington CL
- Hoberman A,
- Wald ER,
- Penchansky L,
- Reynolds EA,
- Young S
- ↵R: a language and environment for statistical computing [computer programming]. Version 3.1.3. Vienna, Austria: R Foundation for Statistical Computing. 2017. Available at: www.R-project.org/
- Gordon M,
- Lumley T
- Santner TJ,
- Pradhan V,
- Senchaudhuri P,
- Mehta CR,
- Tamhane A
- Byington CL,
- Taggart EW,
- Carroll KC,
- Hillyard DR
- Centers for Disease Control and Prevention
- Copyright © 2018 by the American Academy of Pediatrics