This double-blind, parallel, two-group, RCT was conducted from April 24, 2013 to January 23, 2015 at a PICU with 20 beds within a tertiary university children's hospital. Patients between 2 months and 18 years of age who were placed on mechanical ventilators at the PICU were included as candidates. Patients with hypotension at the time of registration were excluded. In this study, hypotension was defined using the following criteria: systolic blood pressure (SBP) < 70 mmHg in patients between the ages of 2–12 months, SBP < 70 + (age in year × 2) mmHg in ages between 1–10 years, and SBP < 90 mmHg in ages greater than 10 years.14 Patients with the following underlying diseases were also excluded: any disorders of the central nervous system that may affect the consciousness level, and renal impairment that may influence both serum concentration and effectiveness of the drug. We regarded patients to have renal impairment when the calculated estimated glomerular filtration rate by the Schwartz equation was less than 50 mL/min/1.73m².15 Patients administered with neuromuscular blocking agents were also excluded, because evaluation of the level of sedation is impossible in such cases.

Eligible patients were randomly allocated into two sedation groups. The control group, which did not receive fentanyl, also needed sedation, so midazolam, a commonly used sedative in mechanical ventilation, was administered to all subjects.781016 Thus, the experimental group received fentanyl in combination with midazolam, and the control group received midazolam alone.

Randomization was performed through block randomization method without stratified factor. The randomization code of the patient was issued through a web-based random number generator operated by the medical research collaborating center of Seoul National University Hospital. After randomization, the randomization code was transferred to the assigned research pharmacist to prepare the ‘experimental drug’; normal saline in the control group, versus fentanyl in the combination group. In order to avoid being distinguished by appearance, the experimental drugs were packaged in 50 mL syringes, diluted into volumes of 25 mL. For patients in the control group, only 25 mL of normal saline was packaged as the experimental drug. For patients in the combination group, fentanyl citrate was diluted with normal saline into a final 20 μg/mL concentration to make a 25-mL solution. In order to maintain a double-blind status, the research pharmacist was not involved in any other processes of this study except in the preparation of the experimental drug.

In cases where intubation was performed in the PICU, midazolam 0.1 mg/kg, fentanyl 1 μg/kg, and vecuronium 0.1 mg/kg were administered as preparation drugs. For all study participants, the patients' assigned nurse prepared midazolam, which was to be continuously infused during the study. When both the experimental drug and midazolam were ready for administration, the study was initiated with continuous infusions of both drugs. The initial infusion rates of midazolam and the experimental drug were 0.06 mg/kg/hr and 0.05 mL/kg/hr, respectively. The infusion rates were then adjusted according to the sedation level.

The sedation level was assessed using the Comfort Behavior Scale (CBS), which is a scale used worldwide for children undergoing sedation.171819 Prior to initiating the study, considering the medical condition of each patient and their expected duration of mechanical ventilation, the patients' assigned research physician determined their individual target CBS score, and re-evaluated its validity through reassessments every 8 hours during the study period. CBS scores were measured regularly at one-hour intervals from study initiation, and additional CBS scores were measured whenever the patient was suspected of awakening from sedation.

Under-sedation was defined as the measured CBS score ≥ 4 points above the target CBS score. Over-sedation was defined as a measured CBS score ≥ 4 points below the target CBS score. In cases with under-sedation, the midazolam infusion rate was increased by 0.03 mg/kg/hr, and the experimental drug infusion rate by 0.025 mL/kg/hr, if the infusion rate had not been altered within one hour. Likewise, in cases with over-sedation, the infusion rate was decreased by the same amount for each drug. In order to prevent an overdose of medications, the infusion rates of midazolam and experimental drug were limited to a range of 0.03–0.36 mg/kg/hr and 0.025–0.3 mL/kg/hr, respectively. If the CBS difference (defined as the target score subtracted by the measured score) was ≥ 8 points, it was considered as serious under-sedation, and a bolus of midazolam (0.1 mg/kg) was injected intravenously for a rapid regulation of the under-sedation status.

The study lasted up to 48 hours if there were no interruptions throughout the study process, after 48 hours, the sedation therapy was performed according to the patient's condition regardless of sedation group or research sedation protocol. If mechanical ventilation weaning was carried out before 48 hours or if any adverse events occurred, the study was immediately terminated. The study was also discontinued for the following reasons: administration of other sedatives or analgesics making it difficult to evaluate the sedation level, application of continuous renal replacement therapy which affects the serum level of drugs, sedation failure even with maximal infusion rates of medications, and co-administration of medications that mask adverse reactions of fentanyl such as antihypertensive drugs.

The primary outcome of this study was to investigate the time ratio of under-sedation (defined as cumulative hours with a CBS difference of ≥ 4 points divided by hours of drug administration) in the two sedation groups. The secondary outcomes were to analyze the CBS differences, time ratio of serious under-sedation, and total amount of midazolam used during study period in each of the sedation groups.

For safety assessments, drug adverse reactions and mortality during the study period were analyzed. Among the known side effects of fentanyl, the following major adverse reactions were evaluated: coma, hypotension, and ileus.202122

In order to calculate a suitable sample size for this study, we need to know the mean difference and standard deviation (SD) of the CBSs of the two groups to be compared. Since there is no previous research on the pediatric population, the sample size was estimated based on a similar adult study instead.13 Because it was an adult study, the sedation level was assessed using Modified Ramsay Sedation Scale and therefore, the scale and score ranges of the CBS used in this study are different. However, as mentioned earlier, previous studies in the pediatric population do not exist, thus we inevitably used the results of the adult study to estimate the approximate number of samples. In the above study, the mean difference between the two groups was 4.9 and the SD was 4.9. For α error of 5% (two-tailed) and a β error of 20%, 17 samples were required for each group.23 Assuming a 20% drop-out rate, we planned to recruit a total of 44 participants, 22 patients per group.

Shapiro-Wilk normality test was performed for continuous data to determine whether the data were normally distributed or not. Continuous data were presented as mean ± SD if it was determined to be normally distributed. If the data was not normally distributed, data were shown as median (interquartile range [IQR]). Two-sample t-test was used in analyzing normally distributed continuous data, and Wilcox rank sum test for nonparametric data. For categorical data, χ² test or Fisher's exact test was adopted in the analyses. In order to analyze the effects of study time on the sedation level and CBS between the sedation groups, we used mixed effect logistic regression model with study time as random effect and sedation group as fixed effect. P values less than 0.05 were considered statistically significant. All statistical analyses were performed using R version 3.4.1 (R Foundation for statistical computing, Vienna, Austria).

If the inclusion and exclusion criteria of this study were satisfied, after explanations pertaining to the details of this study, a written informed consent was obtained from a legally authorized representative of the patient. Afterwards, the patient was regarded as an eligible study participant and research was then advanced on the patient. This trial was approved by the Institutional Review Board of Seoul National University Hospital (H-1210-127-437), and the procedures were in accordance with the Helsinki Declaration.