Quality assurance:
Data is entered in to a spreadsheet as it is extracted exactly from the electronic medical record. During review of the findings, the data is again checked against the medical record to ensure accuracy. Botulinum toxin types are also verified with scanned copies of the botulinum toxin injection sheet which is filled out and scanned in to the medical record at the time of treatment.
Data dictionary:
Not necessary as limited new terminology is used and all understood among the research group. Additional terminology is clearly described for the trial participants and described in the manuscript reporting the results of the study.
Recruitment:
Patients with facial synkinesis who were appropriate candidates for botulinum toxin chemodenervation therapy were offered voluntary participation in this trial. Patients who had returned to their baseline SAQ score after a minimum of 12 weeks since their previous treatment were permitted to re-enroll in the study for subsequent treatments.
Data collection:
At 1, 2, and 4 weeks post treatment, patients were again administered the SAQ. SAQ scores were calculated for each follow up time point. The average SAQ score in each treatment group (onabotulinumtoxinA, abobotulinumtoxinA, and incobotulinumtoxinA) for each week was calculated, as well as the average improvement in SAQ scores.
Data management:
Data were initially recorded and stored in the secure, password protected, electronic medical record. Data were then extracted from the electronic medical record to a secure Excel spreadsheet stored on a password protected computer. Only the principal investigator and those researchers involved in data analysis and interpretation were provided access to this data. Names and identifiable information were excluded from this spreadsheet so that the information could not be linked back to specific patients.
Data analysis:
Data were analyzed by first organizing the data in Excel, within the original spreadsheet used to record data, then exporting this data in to Prism 6 for Windows (GraphPad Software). Using Prism 6, characteristics of the data were assessed and statistical calculations were performed to determine significance of comparisons between groups. Details of these statistical tests are described below.
Reporting for adverse events:
Adverse events were recorded in the electronic medical record at each follow-up visit. These were then noted in the data spreadsheet as well to keep track of any events that had occurred over the study period. Minor adverse effects reported during the study included redness, bruising, swelling, and fullness at the injection sites which were not felt to be significant or noted to be qualitatively different between products. No major adverse treatment effects were reported during the course of the study.
Change management:
There were no significant changes to study protocol over the course of the study.
Sample size assessment to specify the number of participants or participant years necessary to demonstrate an effect:
https://www.dssresearch.com/knowledgecenter/toolkitcalculators/statisticalpowercalculators.aspx At an alpha level of 0.05, an estimated standard deviation of 10, an estimated average of SAQ 44 and comparison/test value of SAQ 55, a sample size was calculated of approximately 400. We were unable to obtain this sample size after nearly 3 years of enrolment from July 2012 to March 2015. However, despite not reaching this estimated sample size, we did identify a statistically significant finding at alpha level of 0.05 and with a standard deviation of approximately 10-15 per group.
Plan for missing data to address situations where variables are reported as missing, unavailable, "non-reported," uninterpretable, or considered missing because of data inconsistency or out-of-range results:
Patients without follow-up at the post-treatment time points of 1, 2, and 4 weeks were excluded and therefore no missing data was expected for SAQ values. The plan for any missing data in the study was to leave the data point as missing without attempts to imputate the value or otherwise correct for missingness.
Statistical analysis plan describing the analytical principles and statistical techniques to be employed in order to address the primary and secondary objectives, as specified in the study protocol or plan:
The average SAQ score in each treatment group (onabotulinumtoxinA, abobotulinumtoxinA, and incobotulinumtoxinA) for each week was calculated, as well as the average improvement in SAQ scores. Lower SAQ scores represented less severe patient reported disease. Percent improvement in SAQ score was calculated for each time point as the average post-treatment score divided by the average pre-treatment score for the treatment group, subtracted from 1. The SAQ scores for each time point and treatment groups were found to be normally distributed (passing both the D'Agostino \& Pearson omnibus normality test as well as Shapiro-Wilk normality test) allowing for an assumption of Gaussian distribution in our following tests of statistical significance. Comparisons between treatments, and between time points within treatment groups were made with a one-way ANOVA followed by appropriate follow-up test for multiple comparisons. Comparisons within treatment groups between each follow-up time point, and the control pre-treatment time point were made using the Dunnett test (compares every mean to a control mean), whereas comparisons of different treatments at a given time point were made with the Tukey test (compares every mean to every other mean). A threshold of p ≤ 0.05 was used for statistical significance.
Patients who had returned to their baseline SAQ score after a minimum of 12 weeks since their previous treatment were permitted to re-enroll in the study for subsequent treatments. Twenty-seven patients were studied, and 6 of these 27 were enrolled multiple times accounting for 11 of the 38 treatments studied. To ensure the baseline SAQ scores of these patients receiving repeat treatments did not differ from patients receiving a single treatment, the average pre-treatment SAQ score for patients undergoing repeat treatments was compared to the average pre-treatment SAQ score of single treatment patients for each type of botulinum toxin neuromodulator. These groups were evaluated for normality of distribution by the same method as described above, with single treatment patients passing tests of normality; normality for the repeat patients could not be determined due to the small number of treatments for each of these patients. Statistical comparisons were then made between the pre-treatment SAQ scores for repeat treatment patients and the respective (same type of botulinum toxin) single treatment patients. In order to preference type I error and identify any possible difference between these groups, we made an assumption of normality and used a one-way ANOVA followed by a Holm-Sidak multiple comparison test. The Holm-Sidak method was chosen as it has more power to identify significant differences than other possible methods such as that of Bonferroni.
