Intermittent vs Continuous Pulse Oximetry in Hospitalized Infants With Stabilized Bronchiolitis: A Randomized Clinical Trial

Key Points

Question  What is the effect of intermittent vs continuous pulse oximetry in infants hospitalized with stabilized bronchiolitis?

Findings  In this multicenter randomized clinical trial of 229 infants hospitalized with stabilized bronchiolitis with and without supplemental oxygen and with care managed using an oxygen saturation target of 90% or higher, length of hospital stay, medical interventions, safety, and parent-reported outcomes were similar. Nursing satisfaction was greater with intermittent monitoring.

Meaning  Given that other important considerations for clinical practice favor less intense monitoring, these findings support the standard use of intermittent pulse oximetry in hospitalized infants with stabilized bronchiolitis.

Importance  There is low level of evidence and substantial practice variation regarding the use of intermittent or continuous monitoring in infants hospitalized with bronchiolitis.

Objective  To compare the effect of intermittent vs continuous pulse oximetry on clinical outcomes.

Design, Setting, and Participants  This multicenter, pragmatic randomized clinical trial included infants 4 weeks to 24 months of age who were hospitalized with bronchiolitis from November 1, 2016, to May 31, 2019, with or without supplemental oxygen after stabilization at community and children’s hospitals in Ontario, Canada.

Interventions  Intermittent (every 4 hours, n = 114) or continuous (n = 115) pulse oximetry, using an oxygen saturation target of 90% or higher.

Main Outcomes and Measures  The primary outcome was length of hospital stay from randomization to discharge. Secondary outcomes included length of stay from inpatient unit admission to discharge and outcomes measured from randomization: medical interventions, safety (intensive care unit transfer and revisits), parent anxiety and workdays missed, and nursing satisfaction.

Results  Among 229 infants enrolled (median [IQR] age, 4.0 [2.2-8.5] months; 136 [59.4%] male; 101 [44.1%] from community hospital sites), the median length of hospital stay from randomization to discharge was 27.6 hours (interquartile range [IQR], 18.8-49.6 hours) in the intermittent group and 25.4 hours (IQR, 18.3-47.6 hours) in the continuous group (difference of medians, 2.2 hours; 95% CI, −1.9 to 6.3 hours; P = .17). No significant differences were observed between the intermittent and continuous groups in the median length of stay from inpatient unit admission to discharge: 49.1 (IQR, 37.2-87.0) hours vs 46.0 (IQR, 32.5-73.8) hours (P = .13) or in frequencies or durations of hospital interventions, such as oxygen supplementation initiation: 4 of 114 (3.5%) vs. 9 of 115 (7.8%) (P = .16) and median duration of oxygen supplementation: 20.6 (IQR, 7.6-46.1) hours vs. 21.4 (11.6-52.9) hours (P = .66). Similarly, there were no significant differences in frequencies of intensive care unit transfer: 1 of 114 (0.9%) vs 2 of 115 (2.7%) (P = .76); readmission to hospital: 3 of 114 (2.6%) in the intermittent group vs 4 of 115 (3.5%) in the continuous group (P > .99); parent anxiety: mean (SD) parent anxiety score, 2.9 (0.9) in the intermittent group vs 2.8 (0.9) in the continuous group (P = .40); or parent workdays missed: median workdays missed, 1.5 (IQR, 0.5-3.0) vs 1.5 (IQR, 0.5-2.5) (P = .36). Mean (SD) nursing satisfaction with monitoring was significantly greater in the intermittent group: 8.6 (1.7) vs 7.1 (2.8) of 10 workdays; the mean difference was 1.5 (95% CI, 0.9-2.2; P < .001).

Conclusions and Relevance  In this randomized clinical trial, among infants hospitalized with stabilized bronchiolitis with and without hypoxia and managed using an oxygen saturation target of 90% or higher, clinical outcomes, including length of hospital stay and safety, were similar with intermittent vs continuous pulse oximetry. Nursing satisfaction was greater with intermittent monitoring. Given that other important clinical practice considerations favor less intense monitoring, these findings support the standard use of intermittent pulse oximetry in stable infants hospitalized with bronchiolitis.

Trial Registration  ClinicalTrials.gov Identifier: NCT02947204

Bronchiolitis is the most common acute lower respiratory tract infection in infants. In the US, it accounts for more than 100 000 hospitalizations and more than $1.7 billion in hospitalization costs annually.¹ The focus of bronchiolitis inpatient management is supportive care because drug therapies are ineffective.^2,3 Clinical and physiologic monitoring guides fluid and oxygen supplementation, respiratory support, and discharge decisions.

In hospital practice, pulse oximetry for identifying hypoxia is considered the fifth vital sign.^4,5 Practitioners choose a pulse oximetry oxygen saturation target to assist with admission and discharge decisions and supplemental oxygen therapy initiation and discontinuation. The American Academy of Pediatrics (AAP) guidelines recommend using an oxygen saturation target of 90% or higher in hospitalized infants with bronchiolitis. This practice is supported by a multicenter trial⁶ that found that infants randomized to a target of 90% or higher, compared with a target of 94% or higher, were discharged 10 hours earlier and experienced a similarly low rate of adverse events and revisits after discharge.

Practitioners also choose to use pulse oximetry measured continuously or intermittently, such as performing a spot measurement every 4 hours. Retrospective studies^7,8 suggest that stable infants with bronchiolitis stay in the hospital longer than needed because of practitioner reliance on the pulse oximeter to make management decisions. Experts and practitioners postulate that continuous measurement of oxygen saturation results in greater detection of transient and false-positive desaturations, leading to unnecessary additional supplemental oxygen use, more interventions, prolonged hospital stays, and higher costs.^2,9,10 Increase in desaturation-related alarms may also increase nursing workload and alarm fatigue.^2,10 In contrast, some practitioners and parents fear that intermittent monitoring will delay the detection of hypoxia and compromise patient safety.²

According to the AAP guidelines, practitioners may choose not to use continuous pulse oximetry based on lower level of evidence.² The Choosing Wisely campaign recommends not using continuous monitoring in infants with bronchiolitis but without hypoxia.¹¹ A cross-sectional study¹² across 56 North American hospitals found that continuous pulse oximetry is used in 46% of infants without hypoxia hospitalized with bronchiolitis, and hospital use of continuous pulse oximetry varies substantially from 2% to 92%. Experts conclude that the evidence supporting the use of continuous vs intermittent pulse oximetry monitoring is “not equivocal, but rather non-existent.”^{13(p 1449)}

We conducted a multicenter, pragmatic randomized clinical trial in hospitalized infants with stabilized bronchiolitis with and without supplemental oxygen to compare the effect of intermittent vs continuous pulse oximetry. We hypothesized that intermittent pulse oximetry would reduce length of hospital stay compared with continuous pulse oximetry. We also compared the effect of the monitoring strategies on secondary outcomes, including the need for medical interventions, safety, parent anxiety and missed workdays, and nursing satisfaction.

This 6-center, pragmatic, parallel group, superiority randomized clinical trial compared intermittent (every 4 hours) and continuous pulse oximetry monitoring in infants with stabilized bronchiolitis hospitalized from November 1, 2016, to May 31, 2019. The study was conducted by the Canadian Paediatric Inpatient Research Network and included general pediatric inpatient units at 3 children’s and 3 community hospitals. Parents or guardians provided written informed consent to participate. Data were deidentified, and unique study identification numbers were used. (A master code list, stored separately from the deidentified data, linked the study identification numbers to participants identifying information.) Research ethics boards at all hospital sites approved the trial protocol (Supplement 1), which has been published previously.¹⁴

After admission to the general pediatric inpatient unit, infants 4 weeks to 24 months of age were eligible for inclusion if they had a clinical diagnosis of first-episode bronchiolitis, were generally healthy, and had a stable clinical status (full eligibility criteria are provided in the trial protocol in Supplement 1). Bronchiolitis was defined as signs and symptoms of respiratory distress associated with a viral respiratory tract infection in accordance with the AAP guidelines.² Only infants with a stable clinical status were included because during the stabilization period infants would not be considered for discharge because of being too unwell (ie, significant hypoxia) and/or having an uncertain illness trajectory. Stable clinical status criteria were defined by the trial investigators based on clinical consensus, institutional practices, and a pilot trial. Specifically, hospitalized infants who required supplemental oxygen had to be stable for at least 6 hours defined by the following: stable or decreasing requirement for supplemental oxygen, stable or decreasing respiratory rate by at least 2 measurements, respiratory rate less than 70 breaths/min, heart rate less than 180 beats/per min, or oxygen supplementation less than 40% fractional inspired oxygen or 2 L/min by nasal prongs. In infants who did not require supplemental oxygen, the clinical status needed to be stable (as defined above) for at least 6 hours from presentation to the emergency department. Infants who required intensive care unit (ICU) admission for mechanical or noninvasive ventilatory support were excluded. Infants treated with heated high-flow oxygen only became eligible after use of the heated high-flow oxygen was discontinued. Infants whose parents could not complete questionnaires because of limited English-language proficiency were excluded.

We randomized (1:1) infants to intermittent or continuous oxygen saturation monitoring. A computer-generated randomization sequence, stratified by center, with random permuted block sizes of 4 or 6 was used. A central, internet-based randomization system used the Research Electronic Data Capture software (REDCap) application.¹⁵ Research assistants at the sites enrolled infants and then used REDCap to obtain the participant group allocation. Masking of the assigned treatment was not possible, given the obvious differences between the 2 interventions.

Bronchiolitis management followed institutional practices and the trial protocol (Supplement 1 and eTable 1 in Supplement 2).¹⁴ Hospitals used 1 of 2 oxygen saturation targets for clinical management (ie, oxygen supplementation and discharge decision-making) in keeping with their usual practice: (1) 90% or greater in room air while the infant was awake or asleep (2 children’s hospitals and 1 community hospital) and (2) 90% or greater in room air while awake and 88% or greater in room air while asleep (1 children’s hospital and 2 community hospitals). Infants received continuous monitoring in the emergency department and inpatient unit until stabilization as per institutional practices.

For the intermittent monitoring group, nurses measured oxygen saturation levels every 4 hours until discharge. If the infant had a clinical deterioration as assessed by the clinical team, the infant could be switched to continuous monitoring and intermittent monitoring restarted when the status stabilized. In the continuous monitoring group, nurses measured oxygen saturation continuously until discharge. For both groups, vital signs were measured every 4 hours, and weaning of supplemental oxygen was managed by the clinical team (Supplement 1).

The primary outcome was length of hospital stay from randomization on the inpatient unit to discharge from the hospital, measured in hours. Prespecified secondary outcomes included length of hospital stay from randomization to meeting discharge criteria, defined as absence of fever (temperature <38 °C), no supplemental oxygen, normal respiratory rate (using World Health Organization age-specific criteria¹⁶), and parent-reported adequate feeding (defined as a score of ≥7 on a 10-cm visual analog feeding scale); length of hospital stay from inpatient unit admission to discharge; the need for medical interventions after randomization (blood testing, blood culture testing, chest radiography, oxygen supplementation [duration and initiation], bronchodilator treatment, systemic corticosteroid therapy, nasopharyngeal virus testing, nasopharyngeal suctioning, and intravenous fluids nasogastric feeds); parent anxiety level on a 4-point Likert scale after randomization (1 indicating not at all at ease; 2, somewhat at ease; 3, moderately at ease; and 4, very much at ease)¹⁷; number of parent workdays missed from randomization to 15 days after discharge; nursing satisfaction with the monitoring strategy assigned to their patient, measured on a 10-mm visual analog scale (0 indicating not at all satisfied and 10 indicating completely satisfied); unscheduled bronchiolitis-related ambulatory physician visits within 15 days of discharge; ICU admission and consultation after randomization; revisits to the emergency department or admission to the inpatient unit at the participating hospital within 15 days of discharge; and mortality. Research staff also assessed participants for crossover from the assigned to the alternate monitoring group by direct observation twice daily and review of electronic records.

We anticipated a median length of hospital stay from randomization to discharge of 36 hours based on our pilot study and published studies.^18-20 Using a significance level of P < .05 (2-sided), we calculated that a sample size of 210 (105 per group) would provide 90% power to detect a minimum median difference of 12 hours in length of stay between groups. We choose a 12-hour minimal clinically important difference based on discussions with practitioners, administrators, and parents. It is also aligned with the frequency at which discharge decisions are made on the inpatient unit.

All analyses were specified a priori. The primary and secondary outcomes were analyzed according to the intention-to-treat principle. To control for multiple testing, the level for declaring statistical significance for the secondary outcomes was set at .002 using the Bonferroni correction. Analyses were performed using R software, version 3.6.2.²¹ All statistical tests were 2-sided.

For the primary outcome of length of hospital stay from randomization to discharge, we estimated the difference in median time between groups and the ratio of the 2 medians along with a 95% CI.²² Because no censoring was necessary, the groups were compared using a Mann-Whitney–type test stratified for site, using the method of Kawaguchi and Koch.²³ A priori, we planned a secondary analysis of the primary outcome adjusted for any clinically important imbalances in baseline variables. Because there were between-group sex differences, additional adjustment of the primary outcome for sex was performed using the method of Kawaguchi and Koch.²³ In addition, a Cox proportional hazards regression analysis stratified for site and adjusted for sex was performed. For each of the secondary outcomes, we report the ratio of medians (interquartile ranges [IQRs]), odds ratios (ORs) (95% CIs), or mean differences (95% CIs). Statistical testing for continuous, dichotomous, and count secondary outcomes was performed using linear, logistic, and negative binomial models, respectively, with site as a covariate.

A total of 229 infants (median [IQR] age, 4.0 [2.2-8.5] months; 136 [59.4%] male) were enrolled in the study, with 101 (44.1%) from community hospital sites (Figure 1 and eMethods in Supplement 2). A total of 100 infants (43.7%) received supplemental oxygen before randomization on the inpatient unit, and 34 (14.8%) were receiving supplemental oxygen at randomization. The median time from inpatient unit admission to randomization was 16.4 hours (IQR, 11.1-24.4 hours). Except for sex, the baseline characteristics of the infants were similar between groups, including the proportion receiving supplemental oxygen, feeding adequacy, and time from inpatient unit admission to randomization (Table 1). For 17 participants (7.4%), there was crossover from the assigned monitoring group: 11 of 114 (9.6%) in the intermittent group and 6 of 115 (5.2%) in the continuous group (eTable 2 in the Supplement).

The median length of hospital stay from randomization to discharge was 27.6 hours (IQR, 18.8-49.6 hours) in the intermittent group and 25.4 hours (IQR, 18.3-47.6 hours) in the continuous group (difference of medians, 2.2 hours; 95% CI, −1.9 to 6.3 hours); ratio of medians, 1.09 (95% CI, 0.93-1.27; P = .17) (Table 2).²³ A Kaplan-Meier plot shows that the length of stay between groups is similar (Figure 2). Sex-adjusted analysis that included hospital site as a stratification variable using Cox proportional hazards models did not show a statistically significant difference in the hazard of discharge between groups (adjusted hazard ratio, 0.93; 95% CI, 0.71-1.23; P = .62).

No significant difference between groups was found for median length of stay from randomization to meeting discharge criteria (18.8 hours [IQR, 2.7-32.6 hours] in the intermittent group and 19.4 hours [IQR, 2.4-30.1 hours] in the continuous group; P = .97) or in the median length of stay from inpatient unit admission to discharge (49.1 hours [IQR, 37.2-87.0 hours] in the intermittent group and 46.0 hours [IQR, 32.5-73.8 hours] in the continuous group; P = .13) (Table 2).²³

Fewer infants in the intermittent group had oxygen supplementation initiated after randomization than in the continuous group (4 of 114 [3.5%] vs 9 of 115 [7.8%]; OR, 0.43; 95% CI, 0.13-1.43; P = .16]), although this finding was not statistically significant. No other significant differences were found between groups in medical interventions, including median duration of oxygen supplementation: 20.6 (IQR, 7.6-46.1) hours vs 21.4 (IQR, 11.6-52.9) hours (ratio of medians, 0.96; 95% CI, 0.37-2.52; P = .66); chest radiography: 8 of 114 (7.0%) vs 11 of 115 (9.6%) (OR, 0.71; 95% CI, 0.28-1.85; P = .49); blood tests: 14 of 114 (12.3%) vs 18 of 115 (15.7%) (OR, 0.75; 95% CI, 0.36-1.60; P = .49); nasopharyngeal virus: 14 of 114 (12.3%) vs 19 of 115 (16.5%) (OR, 0.71; 95% CI, 0.34-1.49; P = .33); blood culture testing: 5 of 114 (4.4%) vs 5 of 115 (4.3%) (OR, 1.01; 95% CI, 0.28-3.58; P = .96); bronchodilator treatment: 21 of 114 (18.4%) vs 20 of 115 (17.4%) (OR, 1.07; 95% CI, 0.55-2.11; P = .88); systemic corticosteroid treatment: 7 of 114 (6.1%) vs 9 of 115 (7.8%) (OR, 0.77; 95% CI, 0.28-2.14; P = .59); or nasal suctioning: 63 of 114 (55.3%) vs 63 of 114 (54.8%) (OR, 1.02; 95% CI, 0.61-1.72; P = .85). Likewise, no statistically significant differences were found in frequency of initiation of intravenous fluid therapy: 9 of 114 (7.9%) vs 8 of 115 (7.0%) (OR, 1.15; 95% CI, 0.43-3.08; P = .76); median duration of intravenous fluid therapy: 26.9 (IQR, 17.0-48.3) hours vs 24.0 (IQR, 17.7-49.4) hours (ratio of medians, 1.12; 95% CI, 0.71-1.76; P = .86); frequency of initiation of nasogastric fluid therapy: 1 of 114 (0.9%) vs 3 of 115 (2.6%) (OR, 0.33; 95% CI, 0.03-3.22; P = .33) or median duration of nasogastric fluid therapy: 34.1 (IQR, 18.1-50.2) hours vs 18.4 (IQR, 10.5-47.2) hours (ratio of medians, 1.86; 95% CI, 0.01-471.44; P = .37) (Table 3).

The frequency of the need for an ICU transfer was similar between groups: 1 of 114 (0.9%) in the intermittent group and 2 of 115 (1.7%) in the continuous group (OR, 0.50; 95% CI, 0.04-5.59; P = .76). Similarly, the frequency of an emergency department revisit within 15 days of discharge was not significantly different between groups: 7 of 114 (6.1%) in the intermittent group and 10 of 115 (8.7%) in the continuous group (OR, 0.69; 95% CI, 0.25-1.87; P = .71), and no statistically significant difference was found in the frequency of readmissions to hospital within 15 days of discharge: 3 of 114 (2.6%) in the intermittent group and 4 of 115 (3.5%) in the continuous group (OR, 0.75; 95% CI, 0.25-3.43; P > .99). No deaths occurred in either group (Table 3).

The mean (SD) nursing satisfaction score with the assigned oxygen monitoring group was greater in the intermittent (8.6 [1.7] of 10) compared with the continuous group (7.1 [2.8] of 10) (mean difference, 1.5; 95% CI, 0.9-2.2; P < .001). Parent anxiety score during the hospitalization, frequency of unscheduled visits to a physician within 15 days of discharge, and median workdays missed by parents did not differ significantly between groups (Table 3).

This pragmatic randomized clinical trial of infants with and without hypoxia hospitalized for bronchiolitis at children’s and community hospitals found that clinical outcomes were similar with intermittent or continuous pulse oximetry when using an oxygen saturation target of 90% or higher. No important differences were found between groups in the primary outcome of median length of hospital stay from randomization. Furthermore, the confidence limits around the difference exclude our prespecified minimal clinically important difference of 12 hours. Although no significant differences were found between groups in medical interventions and safety outcomes as well as parent anxiety or parent workdays missed, nursing satisfaction was greater with intermittent pulse oximetry.

Several possible explanations exist for why a reduction in length of hospital stay was not found with intermittent monitoring. In contrast to the previous nonrandomized and retrospective studies^8,9 that suggested that continuous monitoring prolongs hospital stay, this study was a randomized clinical trial. This trial used the current recommended oxygen target saturation of 90% or higher rather than a target of 93% or higher or 94% used in the earlier retrospective studies.^8,9 Oxygen saturation target level may be a more important determinant of oxygen supplementation and length of hospital stay than monitoring frequency. For safety reasons, this trial compared oxygen monitoring strategies when patients were clinically improving and met stability criteria. Although initiation of intermittent monitoring early in the stabilization period when a greater proportion of infants had hypoxia might have been effective in reducing length of hospital stay, that strategy may not be acceptable to many practitioners.

One previous bronchiolitis trial (n = 161) that compared intermittent with continuous pulse oximetry found no significant differences in mean length of stay or ICU transfer.²⁴ However, that study²⁴ did not include infants with hypoxia; the time of intermittent monitoring initiation was not reported, limiting the comparison on the time saved to discharge²⁵; and parent, nurse, and postdischarge outcomes were not measured.

This trial and several quality improvement reports found similar percentages of ICU transfers and hospital revisits with intermittent pulse oximetry, supporting the safety of this strategy in infants with stabilized bronchiolitis.^26-28 Of importance, this study also found that parental anxiety in the hospital was similar between groups, suggesting that less monitoring does not erode parental confidence, as has been postulated.^13,29

Current AAP guideline recommendations to limit continuous pulse oximetry in hospitalized infants with bronchiolitis carry “the worst evidence-quality grade (D) and weakest recommendation strength.”^{2,13(p 1449)} Almost half of infants without hypoxia hospitalized with bronchiolitis in North American hospitals are monitored continuously.¹² This practice continues because of the weak evidence base; perceived value that nurses, physicians, and parents may place on technology when dealing with uncertainty^30,31; and barriers to deimplementing the established practice of continuous monitoring.¹³

This trial found that clinical outcomes were similar between intermittent and continuous pulse oximetry. Young et al³² suggest application of broader considerations to make a practice recommendation when a randomized trial of 2 established treatments finds nonstatistically and nonclinically significant differences in the primary outcome based on the prespecified criteria. Broader considerations for the use of intermittent vs continuous pulse oximetry argue for the use of less intense monitoring. These considerations include, first, lower nursing workload because of fewer oximeter alarms and enhanced patient safety because of reduced alarm fatigue.^2,10 Of importance, this study found that nursing satisfaction was greater with intermittent monitoring. Second, infants recovering from bronchiolitis commonly experience transient desaturations that are of little clinical importance.³³ Third, intermittent monitoring better aligns with the parent and child hospital-to-home transition preparation. Fourth, doing less simplifies care, which is a valued principle in hospital care.³² Given these considerations, the findings of this study support the standard use of intermittent pulse oximetry in infants hospitalized with stabilized bronchiolitis and provide strong evidence for quality improvement efforts in deimplementing continuous pulse oximetry.

This study has several limitations. First, it was not possible to mask the oxygen monitoring intervention. However, in this pragmatic trial, the interest was in the real-world effect of the intervention, with practitioner and parent knowledge of the monitoring strategy used. Second, this trial does not exclude smaller differences between groups in the primary outcome of length of stay from randomization to discharge, such as 4 hours. Third, data on the frequency of desaturations were not collected, which might have provided a mechanistic understanding between monitoring strategies and clinical outcomes. In addition, future qualitative studies will provide further understanding of nursing monitoring preferences. Fourth, revisits to hospital after discharge did not capture revisits to other hospitals. Fifth, neurocognitive outcomes were not assessed; although there is controversy, some have raised concern that transient hypoxia associated with the acute illness may have neurocognitive impacts.³⁴

In infants hospitalized with stabilized bronchiolitis with and without hypoxia and managed using an oxygen saturation target of 90% or higher, clinical outcomes of intermittent and continuous pulse oximetry monitoring were similar. Given that other important clinical practice considerations favor less intense monitoring, these findings support the standard use of intermittent pulse oximetry in infants hospitalized with stabilized bronchiolitis.

Accepted for Publication: October 6, 2020.

Published Online: March 1, 2021. doi:10.1001/jamapediatrics.2020.6141

Corresponding Author: Sanjay Mahant, MD, MSc, Department of Pediatrics, Hospital for Sick Children, 555 University Avenue, Toronto ON M5G 1X8, Canada (sanjay.mahant@sickkids.ca).

Author Contributions: Drs Mahant and Barrowman had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Mahant, Wahi, Giglia, Sakran, Breen-Reid, Willan, Schuh, Parkin.

Acquisition, analysis, or interpretation of data: Mahant, Wahi, Bayliss, Kanani, Pound, Sakran, Kozlowski, Breen-Reid, Arafeh, Moretti, Agarwal, Barrowman, Willan, Parkin.

Drafting of the manuscript: Mahant, Sakran, Agarwal, Willan, Schuh.

Critical revision of the manuscript for important intellectual content: Mahant, Wahi, Bayliss, Giglia, Kanani, Pound, Kozlowski, Breen-Reid, Arafeh, Moretti, Agarwal, Barrowman, Schuh, Parkin.

Statistical analysis: Moretti, Agarwal, Barrowman, Willan.

Obtained funding: Mahant, Wahi, Breen-Reid, Schuh, Parkin.

Administrative, technical, or material support: Mahant, Bayliss, Sakran, Kozlowski, Breen-Reid, Arafeh, Moretti.

Supervision: Mahant, Bayliss, Sakran, Kozlowski.

Conflict of Interest Disclosures: Dr Mahant reported receiving grants from the Canadian Institutes of Health Research (CIHR) during the conduct of the study and grants from the CIHR and personal fees from the Journal of Hospital Medicine outside the submitted work. Dr Wahi reported receiving grants from the CIHR during the conduct of the study and grants from the CIHR and the Hamilton Health Sciences Foundation outside the submitted work. Dr Bayliss reported receiving grants from the CIHR during the conduct of the study. Dr Kozlowski reported receiving grants from the CIHR during the conduct of the study. Dr Breen-Reid reported receiving grants from the CIHR during the conduct of the study. Dr Moretti reported receiving grants from the CIHR during the conduct of the study. Dr Schuh reported receiving grants from the CIHR during the conduct of the study. Dr Parkin reported receiving grants from the CIHR during the conduct of the study, grants from the CIHR and Hospital for Sick Children Foundation and nonfinancial support from Mead Johnson Nutrition outside the submitted work, and peer-reviewed grants for completed investigator-initiated studies from Danone Institute of Canada (2002-2004 and 2006-2009) and the Dairy Farmers of Ontario (2008-2010). No other disclosures were reported.

Funding/Support: This trial was funded by grant PJT-148635 from the CIHR, a grant from the Department of Pediatrics, Hospital for Sick Children, Toronto, Ontario, Canada, and funds from the Pediatric Outcomes Research Team, Hospital for Sick Children, which receives funds from the SickKids Foundation, Toronto, Ontario, Canada.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The Canadian Paediatric Inpatient Research Network (PIRN): Sanjay Mahant, MD; Gita Wahi, MD; Ann Bayliss, MD; Lucy Giglia, MD; Ronik Kanani, MD; Catherine M. Pound, MD; Mahmoud Sakran, MD; Natascha Kozlowski, MPH; Dana Arafeh, BSc; Patricia C. Parkin, MD.

Data Sharing Statement: See Supplement 3.

Additional Contributions: The site research staff included Nurshad Begum, MD, Cynthia Fleming, BSc, Liz Lee, BSc, Alexandra Lostun, BSc, Barbara Murchison, RN, Mobina Khurram, BSc, Kimberley Kraseich, BSc, Rizani Rivindran, BSc, Arezoo Ebnahmady, BSc, Shelly Ke, BSc, CCRP, Natalia Guira, BSc, Amna Ali, BSc, CCRP, and Martin Romano, BSc. Site practitioners included Laila Premji, MD, and Mollie Lavigne, MSN. The Pediatric Research in Inpatient Settings Network executive council provided input into the early concept of the trial. We thank the children and their families for participating in this trial and the trial investigators and staff across all sites for the dedication and diligent conduct of the trial. The site research staff were compensated for their work.