Factors in Pediatric Isotonic Fluid Resuscitation Efficiency
The study was a single-blind, non-clinical, parallel group randomized controlled trial with four study arms. The trial was conducted at McMaster Children's Hospital, a tertiary pediatric academic center in Hamilton, Canada. Approval for study conduct was obtained from the Faculty of Health Sciences/Hamilton Health Sciences Research Ethics Board. Written informed consent was obtained from all participants prior to participation. Although a non-clinical trial, we elected to register this study at www.ClinicalTrials.gov (NCT01494116). Conduct of this trial was supported by funds obtained from the Department of Pediatrics.
Eligible participants included staff physicians, postgraduate trainees, and nurses who were recruited by an e-mail and poster campaign. We excluded non-English speaking individuals and those incapable of performing manual fluid administration with a syringe. Gift certificates ($25 coffee card) were offered to each subject as a participation incentive. To further motivate peak performance among subjects, further prizes were awarded for those with the fastest fluid administration times. Participants were only allowed to participate on one occasion.
Participants were assigned to one of four study arms, in 1:1:1:1 ratio, using a third party randomization technique. The independent third party created the randomization schedule using http://www.randomization.com and kept this secret from and inaccessible to the investigators. Allocation was therefore concealed. The randomization schedule utilized permuted blocks of randomly varying size.
Participants were provided with details regarding the trial sufficient to achieve informed consent however they were not advised of the hypotheses of the investigators. It was not possible to blind the research assistants, as they needed to be familiar with the study protocol and administer the intervention. Primary outcome assessment was blinded in that the two individuals responsible for extracting outcome data from the trial video recordings were not familiar with trial details or its purpose. The outcome assessors were not advised of subject group assignment and only had access to the study ID number associated with a given video recording. They were also unaware of other data collected as part of the trial. Collected data was input by GH, with verification by MP. The research assistants conducting the trial were conceivably aware of the purpose of the trial (though had no conflicts of interest with respect to the study outcomes). All investigators had access to the raw data. We therefore considered our trial as single-blind.
The model in this study is best described as a low fidelity simulator rendered "physiologic" in that it incorporated a 1.00-inch, 22-gauge IV catheter. Catheters of this size are used clinically in newborn to adolescent age patients, making this a rational choice for our purposes. Please see Additional file 1 for figures of the model and a detailed description.
Following written consent, participants were randomized to one of the four syringe size groups (10, 20, 30, or 60 mL). Upon receiving the participant group assignment, a research assistant measured 900 mL of 0.9% normal saline using a graduated cylinder and then this was divided into three 300 mL aliquots (each aliquot 20 mL/kg based on a 15 kg simulated patient). Each aliquot was then drawn up into colour-coded syringes of the assigned size, with each of the three aliquots having a unique assigned colour. Syringes of the same colour were then placed in each of three kidney bowls.
To ensure adequate familiarity with study procedures and the model prior to formal testing, all subjects underwent a standardization procedure. This consisted of a brief video orienting participants to the study setting and demonstrating the 'disconnect-reconnect' technique. Subjects were then provided with 3 fluid-filled demonstration syringes and given a brief opportunity to practice the technique. Following this, subjects were verbally presented with a clinical vignette of a febrile 15 kg toddler in decompensated septic shock in need of immediate rapid fluid resuscitation. They were advised to administer the fluid using the provided syringes as rapidly as possible, finishing each 20 mL/kg colour set in sequence. Trials were commenced on verbal prompt by the research assistant and proceeded without interruption.
All subject testing was video recorded in a manner which captured the manual performance of fluid administration for outcome ascertainment purposes, but which did not capture participant identifiers. In addition to video recording all testing, the research assistant timed with a stopwatch the initial two participant trials. Due to clear inaccuracies with use of the stopwatch timing method, we reverted to use of the trial video recordings for outcome ascertainment as per our a priori plan. A data collection form was completed by the research assistant at the time of testing to record other secondary outcomes of interest. Each participant also completed a post-trial questionnaire.
Fluid administration times (in seconds) were determined from video review by two independent outcome assessors blinded to the purpose of the trial. For each video, the assessors were asked to determine four separate fluid administration times based upon a clear, a priori defined protocol to ensure consistency. A common software program (Apple Quicktimeâ„¢) was used to review the trial videos and the time bar function was used to identify times in the video frame sequence. Time outcomes extracted included time to administer the full 900 mL (60 mL/kg) of NS (primary outcome measure) and times to administer each of the three 300 mL (20 mL/kg) aliquots of NS, (secondary outcome measure). For the purposes of final data analysis, the times of the two independent assessors were averaged for each outcome of interest.
Descriptive data regarding the characteristics of participants were ascertained from the post-trial questionnaire. The questionnaire also asked participants to recall and rate their level of fatigue following each 20 mL/kg bolus on a 7-point Likert scale. Catheter dislodgement events (defined as physical displacement/removal from the conduit tubing) were noted by the research assistant during testing and on the data collection form. The volume of normal saline actually received by the model was determined by the research assistant by measuring the amount of fluid collected in the graduated cylinder.
The analysis results of subject baseline characteristics and outcome variables (both primary and secondary) were summarized using descriptive summary measures: expressed as mean (standard deviation) or median (minimum-maximum) for continuous variables and number (percent) for categorical variables. Final statistical analyses were performed using SAS (SAS Institute Inc., Cary, NC, USA), although SPSS (IBM Corporation, Armonk, NY, USA) was used for some preliminary analyses and figure generation.
The primary outcome was analysed using a One-way ANOVA analysis, with post hoc comparison of syringe group total intervention time means using Tukey's HSD. Secondary outcome analyses consisted of Generalized Linear Model (GLM) with repeated measures in order to compare bolus administration times and fatigue scores for each of the three sequential aliquots. We planned to use Chi-square testing to compare the proportion of catheter dislodgement events by syringe size group. One-way ANOVA was used to compare the mean volume of normal saline received by the model according to syringe size group. A two-way random effects model was used to compare the agreement of our blinded outcome assessors (both observer and subject were treated as random effects).
The 48 subject sample size for the trial was determined a priori based upon the planned One-way ANOVA primary outcome analysis. Sample size calculations utilized an estimated effect size, determined based upon preliminary testing in the model. Using a significance level of 0.05 and power of 80%, the sample size needed for the trial was conservatively estimated at 12 subjects in each group, 48 total.
Methods
The study was a single-blind, non-clinical, parallel group randomized controlled trial with four study arms. The trial was conducted at McMaster Children's Hospital, a tertiary pediatric academic center in Hamilton, Canada. Approval for study conduct was obtained from the Faculty of Health Sciences/Hamilton Health Sciences Research Ethics Board. Written informed consent was obtained from all participants prior to participation. Although a non-clinical trial, we elected to register this study at www.ClinicalTrials.gov (NCT01494116). Conduct of this trial was supported by funds obtained from the Department of Pediatrics.
Study Participants
Eligible participants included staff physicians, postgraduate trainees, and nurses who were recruited by an e-mail and poster campaign. We excluded non-English speaking individuals and those incapable of performing manual fluid administration with a syringe. Gift certificates ($25 coffee card) were offered to each subject as a participation incentive. To further motivate peak performance among subjects, further prizes were awarded for those with the fastest fluid administration times. Participants were only allowed to participate on one occasion.
Randomization, Allocation and Blinding
Participants were assigned to one of four study arms, in 1:1:1:1 ratio, using a third party randomization technique. The independent third party created the randomization schedule using http://www.randomization.com and kept this secret from and inaccessible to the investigators. Allocation was therefore concealed. The randomization schedule utilized permuted blocks of randomly varying size.
Participants were provided with details regarding the trial sufficient to achieve informed consent however they were not advised of the hypotheses of the investigators. It was not possible to blind the research assistants, as they needed to be familiar with the study protocol and administer the intervention. Primary outcome assessment was blinded in that the two individuals responsible for extracting outcome data from the trial video recordings were not familiar with trial details or its purpose. The outcome assessors were not advised of subject group assignment and only had access to the study ID number associated with a given video recording. They were also unaware of other data collected as part of the trial. Collected data was input by GH, with verification by MP. The research assistants conducting the trial were conceivably aware of the purpose of the trial (though had no conflicts of interest with respect to the study outcomes). All investigators had access to the raw data. We therefore considered our trial as single-blind.
The Model
The model in this study is best described as a low fidelity simulator rendered "physiologic" in that it incorporated a 1.00-inch, 22-gauge IV catheter. Catheters of this size are used clinically in newborn to adolescent age patients, making this a rational choice for our purposes. Please see Additional file 1 for figures of the model and a detailed description.
Intervention Procedure
Following written consent, participants were randomized to one of the four syringe size groups (10, 20, 30, or 60 mL). Upon receiving the participant group assignment, a research assistant measured 900 mL of 0.9% normal saline using a graduated cylinder and then this was divided into three 300 mL aliquots (each aliquot 20 mL/kg based on a 15 kg simulated patient). Each aliquot was then drawn up into colour-coded syringes of the assigned size, with each of the three aliquots having a unique assigned colour. Syringes of the same colour were then placed in each of three kidney bowls.
To ensure adequate familiarity with study procedures and the model prior to formal testing, all subjects underwent a standardization procedure. This consisted of a brief video orienting participants to the study setting and demonstrating the 'disconnect-reconnect' technique. Subjects were then provided with 3 fluid-filled demonstration syringes and given a brief opportunity to practice the technique. Following this, subjects were verbally presented with a clinical vignette of a febrile 15 kg toddler in decompensated septic shock in need of immediate rapid fluid resuscitation. They were advised to administer the fluid using the provided syringes as rapidly as possible, finishing each 20 mL/kg colour set in sequence. Trials were commenced on verbal prompt by the research assistant and proceeded without interruption.
All subject testing was video recorded in a manner which captured the manual performance of fluid administration for outcome ascertainment purposes, but which did not capture participant identifiers. In addition to video recording all testing, the research assistant timed with a stopwatch the initial two participant trials. Due to clear inaccuracies with use of the stopwatch timing method, we reverted to use of the trial video recordings for outcome ascertainment as per our a priori plan. A data collection form was completed by the research assistant at the time of testing to record other secondary outcomes of interest. Each participant also completed a post-trial questionnaire.
Outcome Measures and Assessment
Fluid administration times (in seconds) were determined from video review by two independent outcome assessors blinded to the purpose of the trial. For each video, the assessors were asked to determine four separate fluid administration times based upon a clear, a priori defined protocol to ensure consistency. A common software program (Apple Quicktimeâ„¢) was used to review the trial videos and the time bar function was used to identify times in the video frame sequence. Time outcomes extracted included time to administer the full 900 mL (60 mL/kg) of NS (primary outcome measure) and times to administer each of the three 300 mL (20 mL/kg) aliquots of NS, (secondary outcome measure). For the purposes of final data analysis, the times of the two independent assessors were averaged for each outcome of interest.
Descriptive data regarding the characteristics of participants were ascertained from the post-trial questionnaire. The questionnaire also asked participants to recall and rate their level of fatigue following each 20 mL/kg bolus on a 7-point Likert scale. Catheter dislodgement events (defined as physical displacement/removal from the conduit tubing) were noted by the research assistant during testing and on the data collection form. The volume of normal saline actually received by the model was determined by the research assistant by measuring the amount of fluid collected in the graduated cylinder.
Statistical Analyses and Sample Size Considerations
The analysis results of subject baseline characteristics and outcome variables (both primary and secondary) were summarized using descriptive summary measures: expressed as mean (standard deviation) or median (minimum-maximum) for continuous variables and number (percent) for categorical variables. Final statistical analyses were performed using SAS (SAS Institute Inc., Cary, NC, USA), although SPSS (IBM Corporation, Armonk, NY, USA) was used for some preliminary analyses and figure generation.
The primary outcome was analysed using a One-way ANOVA analysis, with post hoc comparison of syringe group total intervention time means using Tukey's HSD. Secondary outcome analyses consisted of Generalized Linear Model (GLM) with repeated measures in order to compare bolus administration times and fatigue scores for each of the three sequential aliquots. We planned to use Chi-square testing to compare the proportion of catheter dislodgement events by syringe size group. One-way ANOVA was used to compare the mean volume of normal saline received by the model according to syringe size group. A two-way random effects model was used to compare the agreement of our blinded outcome assessors (both observer and subject were treated as random effects).
The 48 subject sample size for the trial was determined a priori based upon the planned One-way ANOVA primary outcome analysis. Sample size calculations utilized an estimated effect size, determined based upon preliminary testing in the model. Using a significance level of 0.05 and power of 80%, the sample size needed for the trial was conservatively estimated at 12 subjects in each group, 48 total.
SHARE
