Article Text
PDF
Abstract
Background Serious bacterial infections in young infants with bronchiolitis are rare. Febrile infants <1 month old with bronchiolitis often receive a lumbar puncture (LP), despite limited data for this practice and lack of clinical practice guidelines for this population. The primary objective was to investigate practice patterns in performance of LPs in the ED management of febrile infants aged ≤30 days with bronchiolitis.
Methods A cross-sectional survey of two national paediatric emergency research networks (PediatricEmergency Research Canada (PERC) and the PediatricEmergency Research UK/Ireland (PERUKI)) was conducted January to November 2017 using a modified Dillman technique. The survey was preceded by a clinical vignette describing a well appearing, 21-day-old infant with low-grade fever, respiratory findings typical of bronchiolitis and no perinatal serious bacterial infection (SBI) risk features.
Results The response rate from PERC was 169/250 (68%) and 172/201 (86%) from PERUKI. Nine physicians in training were excluded, leaving 332 eligible participants. Although most physicians believe that neonates with bronchiolitis rarely have meningitis (PERC 141/161 (87.6%); PERUKI 154/171 (90%)) and feel comfortable diagnosing bronchiolitis in this group (PERC 136/161 (84.5%); PERUKI 143/171 (83.6%)), there was significant variation in the proportion who would be likely/very likely to perform an LP (PERC 100/161 (62.1%); PERUKI 15/171 (8.8%)) (p<0.0001). Practice in Canada, <10 years in practice and lack of comfort with diagnosing bronchiolitis represent multivariable predictors of LP; OR 23.7 (95% CI 11.7 to 47.9), 2.3 (95% CI 1.2 to 4.2) and 2.5 (95% CI 1.1 to 5.0), respectively. Rapid knowledge of respiratory syncytial virus positivity would decrease LP probability from 35.4% to 20.2%.
Conclusion Estimated probability of performing LPs and other interventions in otherwise healthy febrile neonates with bronchiolitis is highly variable between emergency physicians in Canada and the UK/Ireland. Network, <10 years in ED practice and comfort level with diagnosing bronchiolitis in newborns constitute independent predictors of the likelihood of LP performance.
- clincial management
- infectious diseases, viral
- paediatric emergency med
- research, clinical
- respiratory, pneumonia/infections
Statistics from Altmetric.com
- clincial management
- infectious diseases, viral
- paediatric emergency med
- research, clinical
- respiratory, pneumonia/infections
Key messages
What is already known on this subject
Infants <1 month of age with bronchiolitis and fever often receive full sepsis workup with lumbar puncture, despite lack of evidence for this practice and lack of clinical practice guidelines for this population.
As a recent international study found large variation among EDs, clinicians looking after febrile newborns with bronchiolitis in different geographic global regions may also adopt different diagnostic approaches.
What this study adds
In this survey of paediatric emergency physicians in the national emergency networks in Canada and in the UK/Ireland, we found substantial practice variation in diagnostic strategies in newborns with bronchiolitis, with those in the Canadian network more likely to perform a lumbar puncture.
Less practice experience and less comfort with diagnosing bronchiolitis was also associated with greater tendency to perform a lumbar puncture.
Future multinational outcomes research is needed to establish the best practice guidelines and decrease the costs of care in this population.
Introduction
Bronchiolitis represents the most common cause for infant hospitalisation and a common reason for ED visits.1 2 Fever constitutes a common clinical sign of bronchiolitis; yet, in very young infants, the fever poses a dilemma for clinicians: is the fever a consequence of the bronchiolitis or a superimposed serious bacterial infection (SBI)?1–3 These infants are not vaccinated and are susceptible to perinatally acquired pathogens.3
Clinical practice guidelines recommend routine comprehensive screening for SBI in febrile infants under 1 month of age without a clear source of infection.4 5 Since the incidence of concurrent SBI in febrile infants with bronchiolitis is less than 2%, leading bronchiolitis clinical practice guidelines) do not endorse routine laboratory investigations.1–3 However, there is currently no guidance on the management of febrile infants with bronchiolitis in the first month of life. Many such children routinely undergo a full sepsis workup, including a lumbar puncture (LP), despite published evidence that meningitis occurs in less than 1% of these infants.3 6 7 However, given the rarity of bacterial meningitis in young infants with fever and bronchiolitis, a major limitation of the available evidence is lack of optimal power for accurate risk assessment of this outcome. Therefore, the aforementioned practice needs to be evaluated in the context of potential utility and possible adverse effects of this procedure in bronchiolitis.8 9
A recent international study found large variation in practice patterns of bronchiolitis management between EDs.10 It is therefore possible that clinicians looking after febrile newborns with bronchiolitis in different geographic global regions may adopt different diagnostic approaches. This information constitutes an important initial step for creating further evidence about best practice management and optimising resource utilisation in this subpopulation.
The primary objective of this multinational survey of two paediatric emergency research networks in Canada and the UK/Ireland which are part of a large international collaborative paediatric emergency research network was to compare the proportions of emergency physicians who would be very likely/likely to perform an LP in a febrile, well-appearing full-term 21-day-old infant with a typical presentation of mild bronchiolitis and without perinatal sepsis-associated risk factors. We hypothesised there would be a significant difference in this outcome between the networks.
Methods
Study design and population
In accordance with the modified Dillman’s Total Design Method for contact and follow-up,11 we used purposive cluster sampling and sent this international electronic survey to all staff emergency physicians who were also members of the Pediatric Emergency Research Canada (PERC) and the Pediatric Emergency Research in the UK and Ireland (PERUKI) collaborative research networks, using comprehensive membership and email lists of both organisations. We excluded physicians in training, those no longer practising emergency medicine and physicians who did not treat children. The survey period took place from January to March 2017 in Canada and from September to November 2017 in the UK and Ireland. Physicians received a letter informing them that their participation was voluntary and responses confidential. Non-responders were sent three subsequent reminders. We tracked the characteristics of the respondents and non-respondents to assess participation bias. Prior to the administration of the survey, the research ethics boards of the lead hospital and of both network associations approved the survey, under the relevant research governance principles of each country.
Survey content and development
The survey was preceded by a clinical vignette describing a hypothetical case of an alert and well-hydrated 21-day-old full-term infant with low-grade rectal fever, nasal congestion, mild to moderate respiratory distress typical of bronchiolitis and no high-risk perinatal features for SBI.5 12 (figure 1) Due to the absence of a validated questionnaire instrument for management of neonatal bronchiolitis, we have employed the survey technique described by Streiner and Norman.13 We utilised the evidence from the literature regarding the suggested management of febrile newborns,5 12 clinical criteria for the diagnosis and recommended use of laboratory testing in bronchiolitis1 2 and published level of risk for meningitis in young febrile infants with bronchiolitis.3 6 7 Listed authors with content and survey methodology expertise generated, pretested the survey questions for ambiguity, redundancy and proper question sequence, revised the questions and also reduced the number of questions. We have also calculated the Cronbach’s alpha to test for internal consistency of survey questions related to the use of lumbar punctures in bronchiolitis.
Vignette with a hypothetical index case.
The following domains constituted the survey infrastructure: (1) participants’ demographics (six questions), (2) participants’ knowledge about (a) recommended management of febrile newborns in general, (b) typical clinical presentation of bronchiolitis, (c) published risk of bacterial meningitis in febrile neonates with bronchiolitis and (d) the physicians’ level of comfort with diagnosing bronchiolitis in a febrile neonate (five questions), (3) estimated probability of performing the LP, other laboratory tests, chest radiography, use of intravenous antibiotics and hospitalisation of the index case as well as the reasons behind this practice (six questions), (4) estimated level of probability of utilising the aforementioned testing and management strategies in the event of rapid availability of a positive respiratory syncytial virus (RSV) test (five questions) and (5) estimated maximum level of tolerable evidence-based risk of bacterial meningitis required to forego routine LP performance in well-appearing febrile full-term newborns with mild bronchiolitis (one question). The primary outcome was the LP performance as this test represents the most controversial investigation in this population.
The response options included a five-point Likert scale ranging from either ‘strongly agree to ‘strongly disagree’ or from ‘very likely’ to ‘very unlikely’, as appropriate for a given question. The reported agreement results included those where participants designated either strong or moderate level of agreement with or level of likelihood of a given event.
Analyses
To detect a difference (agreed on by the investigators as clinically important) of 15% in the proportion of ED physicians in PERC versus the PERUKI intending to ‘likely’ or ‘very likely’ perform an LP in the index case, and assuming a 80% LP rate in PERC and 65% in the PERUKI with 80% power, a minimum of 138 participants per network were needed (276 participants total).14
Differences in physician characteristics between PERC and PERUKI participants were analysed using the Χ2 statistic. We also used the Χ 2 test to examine other relevant differences in proportions. The association between the LP performers versus non-performers (defined as ‘likely or very likely’ to perform this test) and other candidate predictors of this outcome were analysed using the binomial logistic regression analysis. The candidate predictor variables were the network, physician years in practice, formal paediatric emergency medicine (PEM) training (yes vs no), full-time versus part-time (<50% full-time equivalent) ED practice, ‘comfortable’ or ‘very comfortable’ with diagnosing bronchiolitis in a newborn (vs a lower comfort level) and knowledge about evidence-based probability of meningitis in young febrile infants with bronchiolitis (‘agree’ or ‘strongly agree’ with the statement that very young febrile infants with bronchiolitis rarely have bacterial meningitis or sepsis vs a lower level of agreement or lack of related knowledge). Predictor variables associated with the outcome in the bivariate analysis with a p value <0.2 were included in the multivariable logistic regression analysis. We used Wald and likelihood ratios to remove the unremarkable variables from the model and the Hosmer-Lemeshow test was employed to test the goodness-of-fit of the model.
Results
Respondents and non-respondents
The survey was distributed to a total of 451 members across the networks (PERC: 250, PERUKI: 201). The response rate in PERC was 169/250 (68%) and in PERUKI 172/201 (86%), yielding a total response rate of 341/451 (76%). Nine physicians in training were removed for a total of 332 eligible participants. Significantly more physicians in PERC have been PEM trained and practising in paediatric EDs for 10 years or more compared with PERUKI (table 1). The proportions of PERC participants versus non-respondents were comparable with respect to ED practice for more than 10 years (55% vs 58%), PEM training (65% vs 51%) and full-time ED work (89% vs 82%). The same was true for the PERUKI network-ED Practice >10 years: 33% versus 25%, PEM training 53% versus 38%.
Characteristics of PERC versus PERUKI network participants
Cronbach’s alpha for the survey was 0.88, indicating an excellent internal consistency.15
Knowledge about management of febrile newborns and risk of meningitis in bronchiolitis
Virtually, all participants were in agreement about the typical clinical presentation of bronchiolitis (PERC: 161/161 (100%); PERUKI: 169/171 (98.8%)). The majority agreed that febrile infants under 3 months of age with bronchiolitis rarely have bacterial meningitis (PERC: 141/161 (87.6%); PERUKI: 154/171 (90%)) and feel comfortable/very comfortable making a diagnosis of bronchiolitis in newborns (PERC: 136/161 (84.5%); PERUKI: 143/171 (83.6%)). However, significantly more participants in Canada interpret the current scientific evidence as suggesting that well-appearing febrile neonates with bronchiolitis should undergo a full sepsis workup compared with their UK/Ireland counterparts (PERC: 152/161 (94.4%); PERUKI: 120/171 (70.2%)), p<0.0001).
Management of the index case
The infant described in the vignette would be likely/very likely subjected to an LP by 100/161 (62.11%) emergency physicians in Canada, in contrast to 15/171 (8.77%) physicians in the UK/Ireland (difference 0.53, 95% CI 0.45 to 0.62, p<0.0001; OR 17.0, 95% CI 9.19 to 31.63). Among the 61 physicians in Canada and the 156 physicians in the UK/Ireland who chose to forgo LP, the stated reasons for not performing the test included a low perceived risk of bacterial meningitis (PERC: 48%, PERUKI: 63%), concern about augmenting respiratory distress/parental discomfort (PERC: 29%, PERUKI: 35%) and a higher perceived risk of the procedure compared with the risk of meningitis (PERC: 9%, PERUKI: 13%). Table 2 illustrates the contrast in the differences in the overall management of the index infant between networks, highlighting that investigations and interventions are more frequently performed by the PERC participants. Treatment in Canada, ED practice for less than 10 years and lack of physician comfort with the diagnosis of newborn bronchiolitis represent significant multivariable predictors of LP performance (table 3). Rapid knowledge of RSV positivity would significantly decrease the estimated probability of LP performance and of projected antibiotic treatments (table 4).
Probability of management in febrile newborns with bronchiolitis in study networks
Multivariable predictors of lumbar puncture in febrile neonatal bronchiolitis
Probability of interventions with and without positive RSV test in study networks.
How risk averse are the emergency physicians?
A total of 18/161 (11.1%) physicians in Canada require a 0% risk of meningitis to forego an LP, in contrast to 5/171 (2.9%) of PERUKI participants (difference 8.3%, 95% CI 0.02 to 0.14, p=0.003). Among the participants in the UK and Ireland, the claimed tolerance for the risk level of >1% for meningitis in this population is 42/171 (24.6%) without performing an LP. In contrast, 6/161 (3.7%) participants in Canada would accept a meningitis risk of >1% to forego an LP (difference 20.9%, 95% CI 0.14 to 0.28, p<0.001).
Discussion
This study demonstrates that the estimated probability of performing an LP and other investigations in otherwise healthy febrile neonates with bronchiolitis is highly variable among emergency physicians in Canada and in the UK/Ireland. The network, length of ED practice and the physician level of comfort with diagnosing bronchiolitis in febrile neonates constitute multivariable predictors of the likelihood of LP performance. To our knowledge, this is the first examination of management strategies in this population.
While the North American guidelines for diagnosis and management of bronchiolitis are outlined in both the American Academy of Pediatrics and the Canadian Pediatric Society position statements,1 2 the PERUKI network utilises the National Institute for Health and Care Excellence (NICE) guidelines.16 These guidelines are similar in their recommendations for minimising diagnostic workup for infants with typical bronchiolitis.1 2 16 However, while the NICE guideline notes that infants less than 3 months of age are at risk for deterioration,16 all three guidelines target infants aged 1–24 months, making the management considerations in infants less than 1 month of age challenging.1 2 16
As a result, there are no published guidelines on the diagnosis, diagnostic workup and management of febrile infants less than 1 month of age with bronchiolitis. Physicians practising in both North America and in the UK/Ireland have similar recommended standards of care for febrile newborns with no apparent fever source, which involve full sepsis workup including blood culture, urine culture and LP.4 17 Therefore, many physicians likely default to these guidelines when looking after febrile newborns in the context of a suspected viral source such as bronchiolitis. While this hypothesis is reflected in our practice findings in Canada, our colleagues in the UK/Ireland adopt a less aggressive approach.
We can hypothesise that this difference may in part relate to higher risk aversion among our Canadian colleagues, as suggested by a higher proportion of the physicians in Canada professing to require zero meningitis risk to forego a LP in febrile newborns with bronchiolitis compared with those in the UK/Ireland. Risk minimisers have low tolerance for adverse outcomes and adopt conservative approaches to lower the risk of potential liability.18 In contrast, test minimsers rely on thorough clinical evaluation to avoid potentially unnecessary interventions when the risk of serious outcomes is low. This description may characterise some of our UK/Ireland participating colleagues. A systematic review of 34 studies of malpractice claims from both North America and the UK, found that meningitis was the most commonly missed paediatric diagnosis.19
Concern about potential litigation in Canada may motivate the physicians to proceed with an LP, despite their acknowledged low risk of meningitis in the index case. The context of healthcare delivery may also impact the differences in intervention rates. As the National Health System in the UK compels clinicians to consider the costs of resource utilisation, investigations may well differ as a result of these undocumented pressures. This is supported by an evidence of much lower rates of CT utilisation in head injured patients in the UK compared with Canada.20 A previous study showed a higher rate of brief infant hospitalizations in the UK compared with fewer but longer admissions in Ontario, Canada.21 The stated reasons for this disparity included differences in support in early infancy, the strictly enforced 4-hour ED disposition policy in the UK and a greater presence of paediatric emergency clinicians in the EDs in Canada.21 Therefore, underlying differences in the clinical risk assessment and in the healthcare system pressures may both contribute to the interventional difference between the UK and Canada found in our study.
Literature shows that the risk of SBI in young febrile infants with bronchiolitis is very low, and the main bacterial co-infection is a urinary tract infection (UTI).3 Ralston et al conducted a systematic review summarising the risk of occult SBI in infants aged less than 60–90 days with bronchiolitis. In the 11 included studies, 5/1749 infants had bacteraemia (0.3%), none had meningitis and UTIs were diagnosed in 3.3% of infants.3 While the systematic review did not find differences in rates of SBI in newborns versus older infants, an important limitation was a relative paucity of studies targeting young infants and a small number of newborns with bronchiolitis.3 7 As a result, lack of adequate power prevents us from making firm conclusions about accurate risk estimates for the risk of meningitis in this population. Therefore, future multisite research on this topic would be beneficial.
Interestingly, knowledge of rapid RSV positivity was found to considerably reduce the projected rate of diagnostic tests and antibiotic use in both networks. Although the majority of infants hospitalised with bronchiolitis have positive RSV tests,2 there is currently no data supporting the impact of RSV testing on clinical outcomes in typical bronchiolitis.6 22 While a recent meta-analysis of 789 neonates with bronchiolitis demonstrated no cases of bacterial meningitis, the authors showed a comparable incidence of SBI (usually UTI) in RSV positive babies (11.5%) and those negative for RSV (15.3%).7 23 Therefore, based on current evidence, routine RSV testing is not indicated to guide further management in bronchiolitis. This notion of resource stewardship is also in keeping with recent Choosing Wisely initiatives about the need for medical testing and treatments that could not only cause potential harm to patients but could also add avoidable costs to the healthcare system.24 Hopefully, future research will enlighten us on potential benefit of RSV testing in very young febrile infants with bronchiolitis.
Other discrepancies in the management of bronchiolitis have been recently reported. A study of 3725 infants with bronchiolitis showed large differences in the use of evidence-based supportive therapies and investigations between EDs and networks in North America, UK/Ireland, Spain/Portugal and Australia/New Zealand.10 The current study demonstrates a similar variation in the management of newborn bronchiolitis and suggests the need for enhancing current evidence and identification of international best practices for this population.
A strength of this international survey is the high participation rate (76%), with the response rate much higher compared with the 52% in most surveys.25 Nevertheless, we cannot rule out participation bias. However, our responders and non-responders had similar characteristics, and the resulting bias was therefore likely relatively minor. Importantly, survey results may not accurately reflect actual practice. Therefore, future directions should include large multi-centre examination of practice patterns of diagnostic testing and outcomes in young infants with bronchiolitis. While these results obtained from a diverse international population are likely generalizable to other academic EDs, their generalizability to community EDs may be limited.
In conclusion, we found a large practice variation between the national emergency networks in Canada and in the UK/Ireland in the diagnostic workup of febrile infants under 30 days of age with bronchiolitis. Future multinational research needs to provide accurate clinical outcome estimates in this subpopulation in order to establish the best practice and management guidelines, optimise resource utilisation and decrease the costs of care.
Acknowledgments
The PERUKI collaborator champions and to express our gratitude to the PERC executive or supporting this study.
References
Footnotes
Contributors LS and SS conceived the study, wrote the study protocol, extracted data and wrote the manuscript. LS, SS and MDL supervised data extraction. MDL and DS drafted the manuscript. MDL and DR aided in study design. MDL, DR, DS and SS revised the manuscript for intellectual content. DS conducted the statistical analysis. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Ethics approval This study was approved by the Research Ethics Board at The Hospital for Sick Children in Toronto, Canada, REB file no. 100005229. It was also approved by both the PERC and PERUKI networks.
Provenance and peer review Not commissioned; externally peer reviewed.
Collaborators The following PERUKI collaborators have kindly provided local site championship with this study: PERUKI - Addenbrooke’s Hospital, Cambridge: R Clay; Alder Hey Children’s Hospital, Liverpool: S Messahel; Barking, Havering & Redbridge University Hospitals NHS Foundation Trust: S Gomes; Birmingham Children’s Hospital: S Hartshorn; Bristol Royal Hospital for Children: N Creasey; Chelsea and Westminster Hospital, London: J Ross; Children’s Hospital for Wales, Cardiff: S Mullen; Cork University Hospital: R O’Sullivan; County Durham & Darlington NHS Foundation Trust: J Furness; Evelina Children’s Hospital, London: D Hall; Great North Children’s Hospital, Newcastle: M Anderson; Hull Royal Infirmary: R Toko; James Cook University Hospital, Middlesbrough: A Murad; King’s College Hospital, London: F Cantle; Leeds General Infirmary: H Mollard; Leicester Royal Infirmary: D Roland; Morriston Hospital, Swansea: C Dieppe; North Middlesex Hospital: C Clements; Nottingham Children’s Hospital: C Gough; Ormskirk & District General Hospital: S Gardner; Our Lady’s Children’s Hospital, Crumlin: M Barrett; Queen Alexandra Hospital, Portsmouth: C Vorwerk; Royal Alexandra Children’s Hospital, Brighton: M Lazner; Royal Belfast Hospital for Sick Children: JA Maney; Royal Derby Hospital: G Robinson; Royal Devon and Exeter Hospital: A Appelboam; Royal Free Hospital, London: F Taylor; Royal Hospital for Children, Glasgow: S Foster; Royal Hospital for Sick Children, Edinburgh: J Browning; Royal Manchester Children’s Hospital: K Potier; Royal United Hospital, Bath: N Zurick; Royal Wolverhampton NHS Trust: L Kehler; Sheffield Children’s Hospital: D Burke; St George’s Hospital, London: H Jarman; St Mary’s Hospital, London: N Thompson; Sunderland Royal Hospital: N Mullen; Temple Street Children’s University Hospital, Dublin: P Fitzpatrick; The Royal London Hospital: A Parikh; University College Hospital London: Y Baki; University Hospital Lewisham: A Alcock; University Hospital Southampton: J Bayreuther; Watford General Hospital: M Jacobs. We also wish to thank the PERC executive committee for their support of this study: PERC: Amy C Plint MD FRCP(C) (Past PERC Chair, Children’s Hospital of Eastern Ontario, ON, Canada), Stephen B Freedman MD FRCP(C) (PERC Chair, Alberta Children’s Hospital, AB, Canada), Roger Zemek MD FRCP(C) (Children’s Hospital of Eastern Ontario, ON, Canada), Vik Sabhaney (BC Children’s Hospital, Vancouver, BC, Canada), Quynh Doan MD FRCP(C) (BC Children’s Hospital, Vancouver, BC, Canada), Naveen Poonai MD FRCP(C) (London Health Sciences Centre Children’s Hospital, London, ON, Canada), Samina Ali MD FRCP(C) (Stollery Children’s Hospital, Edmonton, AB, Canada), Marie-Christine Auclair, R.N. (CHU Ste-Justine Hospital, Montreal, Quebec, Canada), Esli Osmanlliu MD FRCP(C) (Montreal Children’s Hospital, Quebec, Canada), Rebecca Emerton (Alberta Children’s Hospital, AB, Canada).
Patient consent for publication Not required.