This study was conducted at Texas Children’s Hospital between November 3, 2017 and June 2, 2020. The study was approved by the ethical committee at Baylor College of Medicine on 12/09/2017 (IRB#: H#39,878; Principal investigator Adam C. Adler, MD) with written informed consent obtained from parents or legal guardians participating. The trial was registered at www.clinicaltrials.gov #NCT03285243, Principal investigator Adam C. Adler, MD, on 15/09/2017 prior to patient enrollment. All methods were carried out in accordance with relevant guidelines and regulations and with CONSORT recommendations.
Children 2–6 years old with American Society of Anesthesiologists (ASA) physical classification category of 1,2 (3 when due to sleep apnea related to the procedure with no other co-morbidities) were eligible for inclusion if they were undergoing adenotonsillectomy under general anesthesia.
Children receiving premedication with midazolam or dexmedetomidine were excluded owing to the reduction in emergence delirium with these agents. Other exclusion criteria were patients with known ocular disorders, migraines, seizures, psychiatric or behavioral health conditions, pre-operative anxiety, developmental delay, and patients taking stimulants for appetite or attention deficit hyperactivity disorders. Patients having received midazolam, dexmedetomidine, or ketamine either preoperative or intraoperatively were also excluded. Patients unable to be transferred to PACU immediately following extubation were also excluded.
Patients are brought to the operating room either walking, in a toy car and occasionally with child life accompanying. At our institution, parental presence is rarely performed and only on insistence from the family.
Anesthetic conduct
All patients underwent routine anesthetics with inhalation induction (sevoflurane 8% with nitrous oxide in oxygen 70%/30%) and maintenance was with sevoflurane. All patients were intubated and received opioids at the discretion of the anesthesiologist. Propofol was given prior to intubation if deemed necessary by the anesthesiologist. Midazolam, ketamine and dexmedetomidine (pre or intraoperatively) were not administered. Intravenous ondansetron and dexamethasone were administered to all patients intraoperatively. At the conclusion of the procedure, the head of the bed was turned 90 degrees to the standard position and the patient extubated. Patients were immediately transferred to a stretcher and brought directly to PACU with blow by oxygen using a modified Jackson Rees (King Systems, Noblesville, IN) and facemask. Any patient unable to be immediately transferred to PACU following extubation (e.g. PACU recovery hold delay) was excluded. In line with institutional practice for adenotonsillectomy, all patients were extubated under deep general anesthesia and immediately brought to the post-anesthesia recovery unit (PACU). This was done to ensure that all emergence started uniformly in all patients as well as the assessment of the baseline PAED score from the time sevoflurane was discontinued.
Randomization
Randomization was performed using individually sealed envelopes assigning patients to group A or group B designating the study group (MBL) and the control group (Sham), respectively. Envelopes were sorted to allow for random allocation. The envelope was opened immediately on arrival to the PACU and the lightbox was set to A or B accordingly. To avoid influencing intraoperative practice, randomization was performed immediately on entry to the PACU. To simulate normal practice conditions, the PACU nurses were instructed to report subjective occurrence of ED to the anesthesia attending and treat per their standard practice. In addition, the PACU nurses were also asked to complete the Pediatric Anesthesia Emergence Delirium (PAED) scale with the baseline performed on arrival to the PACU (Supplemental Table 1) [4]. We chose to have the PACU nurse perform the PAED scale vs. a dedicated research team member to simulate normal practice conditions. To reduce bias, all PACU nurses were kept blinded to the patient grouping.
Study conduct and patient assessment
The lightbox Draeger Phototherapy 4000 (Draeger Medical, Lübeck, Germany) was placed behind the head to the bed directly over the patient at 90 degrees and 12 to 18 inches above the face. All patients were positioned supine for maximal exposure. Patients were immediately randomized to study group or placebo and the light was started within 1 min of entry to the PACU. To allow for concealed allocation, control patients were exposed to sham blue light that was not monochromatic but contained all wavelengths within the visible spectrum (3000 K) (i.e. it appeared blue due to an outer coating on the bulb) (Osram Dulux, Augsberg, Germany). To confirm that this coating simply tinted the light blue without altering the light mechanics, a GaP photodiode radiometer (Solarmeter, Glendale PA) was used prior to study initiation. The study group was exposed to monochromatic light with wavelength peak 460 nm (range 400–500) and 7100 K (Draeger Medical, Lübeck, Germany). To ensure irradiance was maintained throughout the study, irradiance of the bulbs was checked yearly. To maintain blinding, both the sham and experimental bulbs were within the same lightbox.
All PAED scores were taken on entry to the PACU (baseline) and at 10, 20, and 30 min for each patient by dedicated pediatric PACU nursing staff. PAED scores were not recorded if the patient was awake and appropriate or experienced emergence delirium during this time as the scores no longer had clinical relevance since the study outcome is now definitively known. The exposure to the light was also for 30 min or until the patient was awake and appropriate or had emergence delirium. In addition to recording the PAED scores, the nurses were also asked to report if the patient experienced ED. If the PACU nurse observed the patient to have ED, they notified the anesthesiologist or covering anesthesiologist to provide pharmacologic treatment at their discretion.
Statistical analysis
To describe the cohort, we calculated means and standard deviations (SD) or counts and frequencies as appropriate for the variables collected. We compared the MBL and Sham groups using the Student’s t-test for continuous variables (e.g., age, PAED score) and the chi-square test for categorical variables (e.g. % female, ASA category), except when a cell count was < 5, in which case we used the Fisher’s exact test. For baseline comparisons, we calculated absolute standardized differences instead of p-values. Missing data for the PAED scores over time indicated a patient woke up or experienced emergence delirium and so no imputation was performed. There was no other missing data in the variables. To adjust for any remaining confounding among the variables collected, we calculated a logistic regression model with the outcome being emergence delirium and a second logistic regression model with the outcome being either ED or the patient had a PAED score of 12 or more at time periods 10, 20, and 30 min. Both regression models used as independent predictors the following variables: treatment group, age, weight, gender, and ASA category. Robust standard errors were derived to account for any within-cluster correlation at the individual nurse level given the possible serial correlation of outcomes within each nurse caring for his or her patients [5].
An a priori calculation of sample size was done using an incidence of 40% of emergence delirium in similar patient population [6]. We assumed an absolute reduction of 25% (ie, 40% to 15%) in incidence of emergence delirium to be clinically significant. To detect this difference at 80% power with a two-sided alpha of 0.05, required a sample size of 98 participants. All analyses were done using Stata/MP 15.1 for Windows (StataCorp LLC, College Station, TX). All p-values < 0.05 were considered statistically significant.
