Study design and settings
This single-center, triple-blinded, prospective interventional study was conducted at Aichi Children’s Health and Medical Center, a 200-bed tertiary care children’s hospital in Japan. The study was conducted from June 2021 to March 2022 and adhered to the Consolidated Standards of Reporting Trials (CONSORT) statement.[7] The study was approved by the institutional review board of Aichi Children’s Health and Medical Center (2020057, August 25, 2020) and was conducted in accordance with the tenets of the Declaration of Helsinki. Informed consent was obtained in all cases from the patient’s parents or guardians. The registration ID issued by the University Hospital Medical Information Network (UMIN) was UMIN000041330, date 06/08/2020 (Registry URL: https://center6.umin.ac.jp/cgi-open-bin/icdr_e/ctr_view.cgi?recptno=R000047182).
Inclusion and exclusion criteria
We recruited patients who underwent elective surgery and had a body weight between 10 and 20 kg, as well as an American Society of Anesthesiologists Physical Status (ASA-PS) score of 1 or 2. Exclusion criteria comprised the following: history of congenital heart disease, diagnosis of chromosomal abnormalities, PIV access in the lower limbs, and duplicated cases during the study period. In addition, we excluded cases in which the guardians’ primary language was not Japanese. Patients undergoing ambulatory surgery were also excluded due to the difficulty in obtaining consent before randomization in a timely manner.
Randomization and blinding
The study participants were randomly allocated to receive one of the following three NS doses: 0.1 mL/kg, 0.3 mL/kg, and 0.5 mL/kg (simple randomization). Following the sample size estimation, the research investigator (TK) created a random allocation table in a 1:1:1 ratio via STATA 17.0® (StataCorp, College Station, TX, USA). Two data managers (YS, TY) consecutively allocated the participants into the three NS dose groups using the random allocation table. The data managers (YS, TY) wrote the allocated NS dose information on a small piece of paper, which was then wrapped with aluminum foil and concealed in an envelope.
Preparation of NS
The case-assigned anesthesiologist opened the envelope and prepared the allocated amount of NS before the study participants entered the operating suites. The smallest possible syringe size was chosen to draw NS. The anesthesiologists were able to select either a slip-tip 2.5 mL syringe, slip-tip 5 mL syringe, or luer-lock 10 mL syringe (Terumo®, Tokyo, Japan).
Anesthesia induction
Inhalational anesthesia was induced using 5–8% sevoflurane, and 40% N2O was administered for patients without preoperative PIV access. A PIV access was placed on the patient’s upper limb. Immediately after PIV access placement, several measures were used to confirm correct PIV placement and function, as well as to detect any signs of PIV infiltration or dysfunction. First, an operating nurse visually confirmed that infusion via gravity drip proceeded smoothly via the PIV access. Second, the case-assigned anesthesiologist confirmed the absence of resistance by administering a small amount of NS, followed by observation of the insertion site. Another PIV access was placed when PIV infiltration or dysfunction was detected or suspected. After the PIV access on the upper limb was secured, intravenous anesthetic medications (2 mg/kg propofol and 2 µg/kg fentanyl, with or without 0.6–1.2 mg/kg rocuronium) were injected. For cases in which a PIV had already been secured preoperatively for any reason, intravenous anesthesia was induced using the aforementioned medications; the same measures used to confirm correct PIV access and its function were performed before intravenous anesthetic injection. Once deep sedation had been achieved, either endotracheal intubation or supraglottic airway device insertion was performed, and mechanical ventilation was initiated.
Data collection
Data collection (i.e., recording the sounds and blood flow velocity captured using the precordial Doppler machine) was performed by the investigators (AO, TK) in the operating suites after the patient’s airway was secured with an endotracheal tube or a supraglottic airway device.
All data were collected using a precordial Doppler ultrasound machine (ES-100V3®, Hadeco®, Kanagawa, Japan). Doppler sounds were recorded using a microphone (Sanwa Supply USB microphone®, Sanwa Supply®, Okayama, Japan) and recording software (Audacity 2.0®, The Audacity Team®, California, USA). The blood flow velocity was captured using a precordial Doppler machine and recorded using software (Wavetest®, Hadeco®, Kanagawa, Japan).
A precordial Doppler probe (BF8M1558A®, Hadeco®, Kanagawa, Japan) with the following properties was used: maximum intensity of < 310 W/cm2, intensity spatial peak temporal average of < 94 mW/cm2, intensity spatial peak pulse average of < 190 W/cm2, frequency of 2.25 MHz, and beam area of 15.7 cm2. The Doppler probe was fixed on the anterior chest wall using adhesive tape, with patients in the supine position. The probe location was chosen to maximize the baseline Doppler heart sound on the right or left side of the parasternal border, at the level between the 3rd and 6th intercostal space. The Doppler sounds and blood flow velocity were recorded simultaneously.
Data collection was performed using the following procedure: 1) a syringe filled with an allocated amount of NS was connected to the three-way stopcock, which was positioned at the most proximal location to the insertion site of the PIV catheter; 2) the baseline precordial Doppler sound was confirmed; 3) the baseline Doppler sound and blood flow velocity were recorded for 10 seconds, followed by injection of the allocated amount of NS through the three-way stopcock at the highest speed possible; and 4) the recordings were continued until 10 seconds after the initiation of the NS bolus. The research investigator (TK) labelled each data recording with the participant number (based on the inclusion order) provided by the two data managers (YS, TY). The data managers (YS, TY) created a correspondence table that matched the participant number and random allocation results. This correspondence table was concealed from the raters (YO, MT), research investigators (AO, TK), and data analyst (AO) until the completion of the data analysis. The data managers (YS, TY) were not involved in data collection or analysis during the study.
Rater training
Two pediatricians (YO, MT) without knowledge of this study were recruited as raters to evaluate changes in precordial Doppler sounds. The raters were trained by research investigators (AO, TK) via previously recorded Doppler sounds with and without significant Doppler sound changes after NS injection.
Procedures for measurement of outcomes
Changes in the precordial Doppler sound test (S-test)
The two raters (YO, MT) listened to the recorded Doppler sounds in a quiet location outside the operating suites; this was performed independently on different days to avoid bias. After listening to the audio files, the raters documented whether they identified a change in Doppler sounds. Research investigators (AO, TK) played the audio files, and each rater listened to the sound. The raters were blinded from the computer display that showed graphical waveforms of Doppler sound to avoid providing clues of Doppler sound changes. Changes in Doppler sounds were considered to have occurred when both raters reported a noticeable increase in pitch and volume in the S-test.
Changes in the blood flow velocity test (V-test)
The mean blood flow velocity was calculated for each 5-second period before and after NS injection; each 5-second period included 500 separate data measurements. Subsequently, the absolute difference between the mean velocity values before and after injection was obtained. The difference between mean velocity values was classified as positive when the difference was ≥ 1 cm/s (V-test). The cut-off value of 1 cm/s was based on the validation results of a previous study.[5]
Outcomes
The primary outcomes were 1) the proportion of cases with a detected change in precordial Doppler sounds (S-test) and 2) the proportion of cases with a change in mean blood flow velocity of ≥ 1 cm/s in the 5-second period before and after NS injection via PIV access.
Statistical analysis
In describing summary statistics, categorical variables are reported as numbers and percentages, while continuous variables are reported as means and standard deviations (or medians and interquartile ranges, based on the normality of the data). The proportion of cases with a positive S-test result was compared among the three dose groups using the chi-square test for rater 1 (0.1 mL/kg versus 0.3 mL/kg, 0.1 mL/kg versus 0.5 mL/kg, 0.3 mL/kg versus 0.5 mL/kg). These comparisons were only made for the results of rater 1 (YO) and not rater 2 (MT), in order to minimize the risk of Type I error inflation. For the S-test, the interrater agreement for each dose was evaluated using Cohen’s kappa statistic. The proportions of cases with positive V-test results in the three different dose groups were compared using the chi-square test. The proportions of positive test results for the S- and V-tests in each dose group were compared using McNemar’s chi-square test. Logistic regression was used to calculate the crude odds ratios (ORs) of positive S- and V-test results in the different dose groups (i.e., 0.1 mg/kg and 0.3 mg/kg using 0.5 mg/kg as the reference). Cases with missing data were excluded from the analysis. Data were analyzed using Stata 17.1 (StataCorp, College Station, TX, USA), with a two-sided p-value of < 0.05 serving as the criterion for statistical significance. In order to correct for potential Type Ⅰ error inflation, the Bonferroni test was used for comparisons among the three NS dose groups (0.1 mL/kg versus 0.3 mL/kg, 0.1 mL/kg versus 0.5 mL/kg, 0.3 mL/kg versus 0.5 mL/kg); a two-sided p-value of < 0.016 served as the criterion for statistical significance.
Sample size calculation
The proportion of positive S-tests was set at 50% for the 0.1 mL/kg dose, 70% for the 0.3 mL/kg dose, and 90% for the 0.5 mL/kg dose, based on our previous investigation.[5] The total estimated sample size was determined to be 375 (125 per group), based on the results of a chi-square test that assumed a Type Ⅰ error of 0.016 and Type Ⅱ error of 0.2. After allowing for a 5% data loss due to unusable or missing data, we aimed to recruit a total of 394 participants.