Anna Taddio, Charlene Lee, Amelia Yip, Boriana Parvez, Patrick J. McNamara, Vibhuti Shah. Intravenous Morphine and Topical Tetracaine for Treatment of Pain in Preterm Neonates Undergoing Central Line Placement. JAMA. 2006;295(7):793–800. doi:10.1001/jama.295.7.793
Author Affiliations: Departments of Pharmacy (Dr Taddio and Ms Yip), Population Health Sciences (Dr Taddio and Ms Yip), and Paediatrics (Dr McNamara), The Hospital for Sick Children, Faculty of Pharmaceutical Sciences, University of Toronto (Dr Taddio and Ms Yip), and Departments of Pharmacy (Ms Lee) and Paediatrics (Dr Shah), Mount Sinai Hospital, Toronto, Ontario; and Department of Pediatrics, New York Medical College, Maria Fareri Children's Hospital, Valhalla (Dr Parvez).
Context There is limited evidence of the analgesic effectiveness of opioid analgesia or topical anesthesia during central line placement in neonates, and there are no previous studies of their relative effectiveness.
Objective To determine the effectiveness and safety of topical tetracaine, intravenous morphine, or tetracaine plus morphine for alleviating pain in ventilated neonates during central line placement.
Design, Setting, and Participants Randomized, double-blind, controlled trial enrolling 132 ventilated neonates (mean gestational age, 30.6 [SD, 4.6] weeks at study entry) and conducted between October 2000 and July 2005 in 2 neonatal intensive care units in Toronto, Ontario.
Interventions Prior to central line insertion, neonates were randomly assigned to receive tetracaine (n = 42), morphine (n = 38), or both (n = 31); a separate nonrandomized group of 21 neonates receiving neither tetracaine nor morphine was used as a control group.
Main Outcome Measures The primary outcome measure was a pain score for the proportion of time neonates displayed facial grimacing (brow bulge) during different phases of the procedure (skin preparation, needle puncture, and recovery). In randomized neonates, safety assessments included blood pressure, ventilatory support, and local skin reactions.
Results Compared with no treatment, pain scores were lower in the morphine and tetracaine-morphine groups during skin preparation (mean difference, −0.22; 95% confidence interval [CI], −0.4 to −0.04; P = .02 and −0.29; 95% CI, −0.49 to −0.09; P = .01, respectively), and needle puncture (mean difference, –0.35; 95% CI, −0.57 to −0.13; P = .003 and −0.47; 95% CI, −0.71 to −0.24; P<.001, respectively), but pain scores did not differ statistically for tetracaine alone vs no treatment. Pain scores were lower for morphine and tetracaine-morphine vs tetracaine during the skin preparation phase and for tetracaine-morphine vs tetracaine during needle puncture. Compared with neonates without morphine, morphine-treated neonates required larger increases in ventilation rate in the first 12 hours (mean difference, 3.9/min; 95% CI, 1.3-6.5/min; P = .003). Local skin reactions occurred in 30% of neonates given tetracaine vs 0% for morphine (risk difference, 0.30; 95% CI, 0.19-0.41; P<.001).
Conclusion In this study of ventilated neonates undergoing central line placement, morphine and tetracaine plus morphine provided superior analgesia to tetracaine; however, morphine caused respiratory depression and tetracaine caused erythema.
Clinical Trials Registration ClinicalTrials.gov Identifier: NCT00213200
About 10% to 15% of neonates require prolonged hospitalization for conditions such as preterm birth, congenital anomalies, and sepsis.1 As part of their medical care, hospitalized neonates are exposed to multiple invasive procedures.2 There is accumulating evidence that untreated procedural pain in neonates leads to long-term changes in pain sensitivity.3 Pharmacologic interventions aimed at decreasing pain and its effects are recognized as a priority in neonatal pharmacology research.4
Many hospitalized neonates require insertion of central venous catheters in order to maintain long-term vascular access for administration of medications and parenteral nutrition.2 Consensus statements recommend local or systemic analgesia to minimize pain during this procedure.5 However, these recommendations are based on limited data regarding either the effectiveness or safety of these agents.6- 8 Moreover, there are no data on the relative effectiveness of local anesthesia vs opioid analgesia. Based on the limited published data on the clinical pharmacology of available agents, it has been difficult to implement analgesic protocols in clinical practice. The objectives of this study were to determine the relative efficacy and safety of topical local anesthesia and intravenous opioid analgesia, used alone or in combination, for management of pain in neonates undergoing percutaneous central venous catheter (PCVC) placement in the neonatal intensive care unit.
The study participants were medically stable neonates who required insertion of a PCVC and were receiving ventilatory support in the form of high-frequency oscillation, conventional ventilation, or continuous positive airway pressure. Neonates were excluded if they had clinical seizures, concomitant muscle relaxant or inotrope therapy, or skin disorders with visually apparent skin lesions or disruptions in skin integrity. There were no restrictions based on gestational age or concurrent administration of analgesia for other indications (eg, continuous opioid infusions). The protocol was approved by the institutional ethics review boards of The Hospital for Sick Children and Mount Sinai Hospital in Toronto, Ontario, and was conducted in neonates who were hospitalized in the neonatal intensive care units of both institutions. A member of the study team obtained written or oral consent from parents of the neonates.
Prior to PCVC placement, each neonate received 1 of 3 possible regimens by the bedside nurse in a double-blind fashion: 0.5 g of tetracaine 4% gel applied to the insertion site; 0.1 mg/kg of intravenous morphine; or both 0.5 g of tetracaine 4% gel and 0.1 mg/kg of intravenous morphine. Identical-appearing placebos were available for both tetracaine and morphine (ie, double dummy), so that all neonates received either tetracaine or placebo applied to the insertion site and all neonates received either morphine or placebo infusions prior to the procedure.
Concealment of treatment allocation was achieved by having the research pharmacist prepare the randomization assignment before enrollment of neonates using a computerized random-number generator, stratified by corrected gestational age (<30 weeks, 30-36 weeks, or >36 weeks) and in random block sizes of 6 or 9, with an equal probability of being allocated to each active treatment. The randomization assignment was stored in a secure location that could not be accessed by study personnel. For each participating neonate, the pharmacy prepared and dispensed 2-unit dose study syringes containing topical study medication (to account for the possibility of multiple procedure attempts) and 1 intravenous infusion syringe containing intravenous study medication in sequentially numbered containers with identical labels. Pharmacy personnel were not involved in any other aspect of the trial. Each neonate received 0.5 g of topical study medication at 2 potential puncture sites occluded with a dressing. Twenty minutes after administration of the topical medication, the study infusion was administered intravenously at a rate sufficient to deliver 0.1 mg/kg of morphine or placebo over 20 minutes. The dressing was removed and the cream was wiped away after 40 to 45 minutes.
In addition, a fourth, nonrandomized group of neonates who were undergoing the procedure without analgesia were added as a no-treatment control group after the study was under way (2003). Neonates undergoing PCVC placement whose parents refused anesthesia but provided consent for videotaping and monitoring were included. The no-treatment group was subject to the same inclusion and exclusion criteria as the randomized neonates and was otherwise treated in the same way as randomized neonates during the procedure.
Percutaneous central venous catheter insertion was performed by trained registered nurses, clinical nurse specialists–nurse practitioners, neonatology fellows, or staff neonatologists using polyurethane catheters with a 24-gauge introducer. The procedure was subdivided into 4 distinct observation periods: baseline, skin preparation (application of antiseptic and tourniquet), needle insertion (skin puncture and catheter threading), and recovery.
The neonates' faces were videotaped during the entire procedure using a digital camera. Blood pressure was obtained using an arterial transducer if there was an indwelling arterial line or with an external cuff attached to a noninvasive blood pressure module. Heart rate was monitored using an electrocardiogram lead that was attached to an electrocardiogram/respiratory module. Blood pressure and heart rate modules were attached to a bedside monitor from which readings were obtained. Oxygen saturation was monitored using an oxygen saturation probe attached to a monitor.
Facial grimacing was chosen as the primary measure of neonatal pain response because it is considered the most sensitive and specific indicator of pain and forms the basis of most pain assessment techniques.9,10 Neonates consistently display facial grimacing responses to painful stimuli that can be objectively measured; however, the robustness of neonate responsiveness varies with gestational age.11 In this study, we chose to use brow bulge as the single indicator of pain because it is the most specific to pain and, unlike other facial actions, is consistently expressed in neonates of young gestational ages.12
The facial grimacing activity of the neonates, expressed as the proportion of time that the neonate had bulging of the brow, was assessed by a trained research assistant who was unaware of treatment assignment. The presence or absence of bulging of the brow was scored in 2-second intervals for the first 20 seconds of each phase of the procedure. The proportion of time that brow bulge was observed was then calculated for each phase of the procedure by summing the number of times the action was observed in each 2-second block and then dividing by the total number of observation blocks. Scores ranged from 0 to 1. Interrater reliability was assessed by comparing the observed agreement in brow bulge activity within each 2-second block of scoring between raters (intraclass correlation coefficient, 0.87; 95% confidence interval [CI], 0.85-0.89). Secondary outcome measures included heart rate and oxygen saturation changes. The heart rate and oxygen saturation values were obtained in 10-second intervals and the means were calculated for the overall procedure and, specifically, for the needle puncture phase.
Safety was assessed in neonates who participated in the randomized controlled trial portion of the study. Blood pressure and ventilatory support were recorded prior to the infusion of study medication then at 15, 30, and 60 minutes after the beginning of the infusion. The mean blood pressure for the first 60 minutes after the beginning of the infusion was calculated. The number of neonates with clinically significant hypotension, defined as requiring intervention (volume expanders and/or inotropes), was determined for the first 6 hours after the infusion. The ventilation rate, change in oxygen requirement (fraction of inspired oxygen), and number of neonates who were extubated, reintubated, or changed from one form of ventilatory support to another were recorded for the first 12 hours after the infusion. Local skin reactions, such as local erythema and blanching, were assessed by visual examination of the skin within 5 minutes of removal of topical study medication.
There were no previous studies that compared topical local anesthesia with intravenous opioid analgesia in neonates undergoing PCVC placement. The sample size calculation was based on the ability to detect an effect size (the difference between the most effective and least effective therapies divided by the standard deviation) equal to 0.8 with α = .05, β = .20, and assuming an equal scatter in distribution of scores among treatment groups.13 Using these criteria, 31 neonates were required per group (n = 124 altogether).
Facial grimacing scores during the procedure were compared among groups using a mixed-model analysis14 with unstructured covariance, in which 3 of the procedure phases (skin preparation, needle puncture, and recovery) were included and the baseline value was a covariate. An unstructured correlation matrix was used since we did not a priori hypothesize any structure to the data. Overall mean heart rate and oxygen saturation during the procedure were compared using a 1-way analysis of covariance, with the baseline value included as a covariate. For analyses of blood pressure and respiratory parameters, the 2 morphine groups (ie, morphine and tetracaine-morphine) were merged and compared with the tetracaine group. Localized cutaneous reactions were compared between all neonates who received tetracaine (ie, tetracaine and tetracaine-morphine) and those who received morphine only. Dichotomous adverse event data were analyzed using the Fisher exact test; 95% CIs were estimated using the binomial distribution. The characteristics of the neonates were compared by χ2 test or t test. Analyses were performed using a modified intention-to-treat analysis that excluded neonates in whom the study was discontinued for reasons unrelated to efficacy or safety and neonates who withdrew from the study.15 Neonates were not excluded for protocol violations. Statistical tests were conducted using SAS software, version 9.1 (SAS Institute Inc, Cary, NC).
The study was conducted between October 2000 and July 2005. Altogether, 132 neonates were recruited (Figure). The mean (SD) gestational age at the time of the study was 30.6 (4.6) weeks. The parents of 111 neonates consented to randomization and the remainder (n = 21) consented to videotaping and monitoring of their neonates during the procedure without any analgesia. In the randomized cohort, the procedure was cancelled in 4 cases (n = 1 in the tetracaine group and n = 3 in the tetracaine-morphine group) and postponed in 1 case in the tetracaine group after the neonate had received the study medication. Thus, 107 neonates (96%) were included in the safety analysis; 105 neonates (98%) were included in the physiologic response analyses and 98 (92%) were included in the facial grimacing analyses. Six neonates were receiving dopamine at the time of the study and 1 neonate was not receiving any form of respiratory support; all were retained in the analyses. For the no-treatment group, 20 neonates (95%) and 19 neonates (90%) were included in the facial grimacing and physiologic response analyses, respectively. The characteristics of participating neonates are shown in Table 1.
Among the active treatment groups, there was no evidence of a difference in mean time interval between removal of the cream and commencement of the procedure or between the end of the infusion and commencement of the procedure; the overall means were 19 (SD, 23) minutes and 17 (SD, 14) minutes, respectively. Similarly, there was an absence of evidence of a difference in the duration of the first procedure attempt (overall mean, 7.1 [SD, 3.7] minutes).
The proportion of neonates who had a central line placed successfully during the procedure was 40% for neonates treated with tetracaine compared with 37%, 36%, and 24% for those given morphine, tetracaine-morphine, and no treatment, respectively (P = .65).
The brow bulge, heart rate, and oxygen saturation responses observed during PCVC placement are displayed in Table 2. Brow bulge demonstrated a significant phase × group interaction; thus, responses were compared between groups within each procedure phase (Table 3). Relative to the no-treatment group, brow bulge scores during the skin preparation phase were lower for the morphine group (mean difference, −0.22; 95% CI, −0.4 to −0.04; P = .02) and tetracaine-morphine group (mean difference, −0.29; 95% CI, −0.49 to −0.09; P = .01). Similarly, brow bulge scores in the morphine and tetracaine-morphine groups were lower than in the no-treatment group during the needle puncture phase (mean difference, −0.35; 95% CI, −0.57 to −0.13; P = .003 and −0.47; 95% CI, −0.71 to −0.24; P<.001, respectively). For tetracaine vs no treatment, scores did not differ beyond chance during the skin preparation or needle puncture phases (mean difference, 0.02; 95% CI, −0.16 to 0.20; P = .86 and −0.20; 95% CI, −0.42 to 0.02; P = .09, respectively). Brow bulge was significantly lower for morphine vs tetracaine during the skin preparation phase (mean difference, −0.24; 95% CI, −0.4 to −0.08; P = .003) but not during the needle puncture phase (mean difference, −0.15; 95% CI, −0.35 to 0.05; P = .12). Tetracaine-morphine scores were lower than tetracaine for both phases of the procedure (mean difference, −0.30; 95% CI, −0.48 to −0.12; P = .001 and −0.27; 95% CI, −0.49 to −0.05; P = .01, respectively). There was little difference between tetracaine-morphine and morphine during either phase (mean difference, −0.07; 95% CI, −0.25 to 0.11; P = .45 and −0.11; 95% CI, −0.33 to 0.11; P = .28). There was no evidence of a group difference for the recovery phase (P = .12); hence, pairwise comparisons were not made for this procedure phase.
Compared with the no-treatment group, heart rate during the procedure (Table 4) was lower in all active treatment groups. For tetracaine, the mean difference was −5.8/min (95% CI, −10.5/min to −1.1/min; P = .02); for morphine, it was −10.0/min (95% CI, −14.9/min to −5.1/min; P<.001); and for tetracaine-morphine, it was −10.9/min (95% CI, −16.2/min to −7.2/min; P<.001). When heart rate was compared during the needle puncture phase, it was lower for all active treatment groups compared with the no-treatment group. For tetracaine, the mean difference was −7.1/min (95% CI, −13.8/min to −0.4/min; P = .04); for morphine, it was −12.0/min (95% CI, −19.1/min to −4.9/min; P = .001); and for tetracaine-morphine, it was −13.1/min (95% CI, −20.5/min to −5.7/min; P = .001). Overall procedure heart rate was lower for morphine and tetracaine-morphine compared with tetracaine (mean difference, −4.3/min; 95% CI, −8.2/min to −0.4/min; P = .04 and −5.1/min; 95% CI, −9.4/min to −0.8/min; P = .02) but not during the needle puncture (mean difference, −4.8/min; 95% CI, −10.5/min to 0.9/min; P = .10 and −6.0/min; 95% CI, −12.3 to 0.3; P = .07). Heart rate during the overall procedure and needle puncture phases did not differ statistically between tetracaine-morphine and morphine (mean difference, −0.8/min; 95% CI, −5.1/min to 3.5/min; P = .70 and −1.1/min; 95% CI, −7.4/min to 5.2/min; P = .72).
We found little difference in oxygen saturation during the procedure (Table 2) among the 4 groups; hence, pairwise comparisons were not made for this outcome.
For analyses of adverse events, morphine and tetracaine were assessed separately and compared with their respective placebos. For analyses of adverse events during morphine use, 66 neonates allocated to either morphine alone or tetracaine-morphine were compared with 41 neonates allocated to tetracaine. We could not demonstrate a difference beyond chance between neonates treated with morphine and those treated with tetracaine in mean blood pressure in the first hour after the infusion, incidence of hypotension requiring intervention (3 [4.5%] vs 0), or mean oxygen requirement in the first 12 hours afterward (Table 5). Within the subgroup of neonates receiving conventional ventilation, the mean change in ventilation rate in the first 12 hours after the infusion was higher for morphine compared with tetracaine (mean difference, 3.9/min; 95% CI, 1.3/min to 6.5/min; P = .003). There was little evidence that the frequency of extubation (4 vs 2), reintubation (1 vs 0), or switch from conventional ventilation to high-frequency oscillation (1 vs 1) within 12 hours of infusion differed between the morphine-exposed and tetracaine-exposed neonates (P>.99 for all analyses).
Of the 69 neonates treated with tetracaine alone or tetracaine-morphine, 21 (30%) experienced erythema at the site compared with 0 of 38 neonates treated with morphine (risk difference, 0.30; 95% CI, 0.19-0.41; P<.001). One neonate experienced a severe local skin reaction (redness followed by blistering and desquamation) and systemic reaction (bradycardia) within 10 minutes of tetracaine application.16 The bradycardia was determined to be due to an underlying pneumothorax and resolved after needle aspiration of free air and insertion of a chest tube. The skin healed after 1 week with no sequelae.
Hospitalized preterm and full-term neonates routinely undergo painful procedures, and treating their pain is recognized as important. Investigation of the clinical pharmacology of analgesics in neonates has been identified as a critical issue for future research by the US Food and Drug Administration and the National Institute of Child Health and Human Development.2 To our knowledge, this is the first study to evaluate the relative effectiveness of local anesthesia and systemic analgesia in neonates undergoing procedural pain. We found that the combination of opioid analgesia and local anesthesia was more effective than either local anesthesia alone or no analgesia in reducing the pain of PCVC placement but similar in effectiveness to opioid analgesia alone. Local anesthesia alone, however, was not consistently different from no analgesia, suggesting that it was a weak analgesic. We hypothesize that morphine was more effective than tetracaine because morphine reduced the sensory input derived from multiple phases of the procedure that differed in both the location and degree of invasiveness, whereas tetracaine decreased sensation from the needle puncture site.
Our results are in partial agreement with previous studies investigating opioid analgesia6 and local anesthesia7,8,17 for neonates undergoing central venous catheter placement. In an open-label, nonrandomized study by Moustogiannis et al,6 morphine attenuated the pain-induced increases in heart rate and skin blood flow experienced during the procedure compared with a no-treatment control group. Garcia et al7 demonstrated a significant effect of lidocaine-prilocaine (vs placebo) on heart rate and respiratory rate after needle puncture. More recently, Ballantyne et al8 and Lemyre et al17 reported a lack of evidence of analgesia from tetracaine (vs placebo) on neonatal pain responses in 2 separate randomized controlled trials. However, the short application time used for tetracaine (30 minutes) coupled with the inclusion of multiple phases of the procedure in the efficacy analyses may have contributed to the inability to detect a treatment effect.
This study found that morphine and tetracaine were both associated with risks to neonates. Hypotension occurred in 5% of neonates who received morphine, and morphine-treated neonates experienced mildly increased ventilatory requirements. We believe that these effects were due to morphine because similar adverse effects have been reported in neonates treated with morphine in other studies.5,18 Tetracaine caused skin erythema in 30% of neonates, with eventual blistering of the skin and desquamation in 1 neonate.16 The frequency of adverse skin reactions observed with tetracaine in this study was higher than in previous studies19 and may reflect differences in assessment techniques. To our knowledge, this is the first study to document a severe skin reaction following use of tetracaine in neonates.
The majority of procedures (65%) resulted in unsuccessful placement of a central line on the first attempt, requiring neonates to be subjected to multiple attempts and additional pain. This further supports the use of analgesia to reduce pain during this procedure. While there was a trend toward fewer successful cannulations in neonates who received no analgesia (24%) compared with those who received any form of analgesia (38%), the difference was not statistically significant. We hypothesize that central line placement success rate in neonates may be influenced by factors including neonate sedation and analgesia (by promoting immobility), as is the success rate of peripheral intravenous cannulation in children.15 Future studies, however, are needed to address the specific effects of analgesia on procedure success rates.
There are several limitations that need to be acknowledged. First, we included a control group of neonates that was not randomized and may have introduced some bias.20 We initially chose not to randomize neonates to a placebo control group because there was some evidence of the effectiveness and safety of opioid analgesia6 and topical local anesthesia7 for this procedure and it was not considered ethical to deny neonates analgesia for the purposes of this study. However, some recent studies that did not demonstrate effectiveness of morphine or tetracaine, and the serious adverse event in our study led us to believe that the results would not be convincing without a no-treatment group. Therefore, neonates whose parents refused analgesia but agreed to let investigators record pain responses during the procedure were added. On examination of their birth characteristics, they appear to have been smaller and sicker than the randomized neonates; however, at the time of study entry, they appeared to be healthier because they were receiving less ventilatory support. Apart from not receiving analgesia, they were treated in the same manner as randomized neonates and their pain responses were assessed by research assistants who were unaware of group assignment. In addition, we did not find any evidence of a pattern in pain responses according to neonate characteristics and chose to use brow bulge as the primary outcome because it is relatively unaffected by gestational age or ventilation status.
Second, our power calculation led to the recruitment of a sample size that may have been insufficient to detect important differences between groups in effectiveness and safety.21 Moreover, of the 132 neonates originally enrolled, 14 (11%) were excluded from the primary analysis, which further reduced our ability to detect differences between groups and may have also contributed to bias in treatment comparisons. A small sample size may have contributed to our inability to demonstrate a statistically significant decrease in brow bulge during the needle puncture phase for the tetracaine vs no-treatment groups. In addition, the small sample size may have affected our ability to demonstrate the added benefit of tetracaine when combined with morphine. Likewise, the pattern of adverse effects for morphine-exposed neonates was not demonstrated to be significantly different from that for tetracaine-exposed neonates for cardiorespiratory parameters other than ventilation rate.
In conclusion, we found that morphine or the combination of tetracaine and morphine were more effective than no analgesia in reducing pain in ventilated neonates during central line placement, as assessed by facial grimacing and heart rate. Tetracaine demonstrated some effectiveness, as determined primarily by heart rate. Morphine and the combination of tetracaine and morphine were consistently superior to tetracaine when the entire procedure was taken into account but not when only the needle puncture phase was considered. These findings are consistent with the underlying mechanisms of action for both drugs. Both agents were associated with risks to neonates: morphine caused mild respiratory depression and tetracaine caused local skin reactions. These data can be used to support evidence-based protocols for the pharmacologic management of pain during PCVC placement in ventilated neonates.
Corresponding Author: Anna Taddio, PhD, Departments of Pharmacy and Population Health Sciences, The Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 1X8 (firstname.lastname@example.org).
Author Contributions: Dr Taddio had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Taddio, Shah.
Acquisition of data: Taddio, Lee, Yip, Parvez, McNamara, Shah.
Analysis and interpretation of data: Taddio, Lee, Yip, Parvez, McNamara, Shah.
Drafting of the manuscript: Taddio, Shah.
Critical revision of the manuscript for important intellectual content: Taddio, Lee, Yip, Parvez, McNamara, Shah.
Statistical analysis: Taddio, Lee, Yip, Shah.
Obtained funding: Taddio, Shah.
Administrative, technical, or material support: Taddio, Shah.
Study supervision: Taddio, Parvez, McNamara, Shah.
Financial Disclosures: None reported.
Funding/Support: Funding for this study was provided by the Canadian Society of Hospital Pharmacists. Dr Taddio is supported by the Canadian Institutes of Health Research New Investigator Award. Ms Lee was supported through a studentship by the Ontario Student Opportunity Trust Fund–Hospital for Sick Children Foundation Student Scholarship Program.
Role of the Sponsor: The funding agencies had no input in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript.
Acknowledgment: We are grateful to Derek Stephens, MSc, and Eshetu Atenafu, MSc, for their assistance with the statistical analyses and Lucia Taddio, BA, Erwin Darra, and Omar Parvez for their assistance with data collection (all from The Hospital for Sick Children); and to the staff of the neonatal intensive care units of The Hospital for Sick Children and Mount Sinai Hospital and parents of participating neonates for their support of the study. We also thank Gideon Koren, MD (The Hospital for Sick Children) and Arne Ohlsson, MD (Mount Sinai Hospital) for their guidance during the trial. Ms Taddio and Mr Darra received compensation from the study sponsor.