SAMPLE SIZE CALCULATIONS. Assuming a responder rate (microbiological cure) of 75% in the Solosec treatment group and a 40% placebo response rate and based on the use of a two-sided, two-sample comparison of proportions at the alpha=0.05 level of significance, a sample size of 100 patients (50 patients in each group) who meet the mITT criteria will provide approximately 95% power to demonstrate a statistically significant difference between Solosec and placebo. Assuming 70% of patients randomized in the study will meet the mITT criteria, approximately 144 patients will be enrolled into the study.
SAMPLE SIZE CALCULATIONS. Estimated rates of subsequent MRSA infection were based on the available literature. For patients without nasal colonization it was estimated that 2.5% of the population would develop a MRSA infection during follow-up while 10% of those with low nasal colonization (Ct > 24) and 30% of those with high (Ct ≤ 24) nasal colonization would subsequently develop an infection. Using a power of 80% and an alpha of 0.05, approximately 28 patients would be needed to show an effect when comparing high colonization to no colonization, 62 patients needed to compare high colonization to low colonization, and 150 patients comparing low colonization to no colonization. Approximately 50 patients per month are admitted to the AVAMC with MRSA nasal colonization. Of these patients, 30% have high nasal MRSA burden (≈ 15/month) and 70% have low nasal MRSA burden (≈ 35/month). Based on the above estimates and admission patterns at the AVAMC, it was estimated that four months worth of admission data would be needed to provide an adequate sample size. Descriptive statistics were used to compare the study population stratified by colonization status (negative, low burden, high burden). The purpose of this comparison was to identify potential factors that are associated with the exposure variable (colonization status). Differences in proportions of categorical variables (including demographic and clinical characteristics and co-morbidities) were tested using χ2. If expected cell counts were less than 5, Xxxxxx’x exact test was utilized. Continuous variables were analyzed with a one-way analysis of variance to compare means or two- sample T test. A p-value of ≤ 0.05 was considered significant. Unadjusted risk ratios were obtained for all covariates and the outcome in univariate analysis. Due to the limited published clinical data on nasal colonization burden, an exploratory analysis for potential interaction terms was performed. In this analysis, the study population was stratified on individual covariate levels and risk ratios for each level of the exposure variable (colonization status) were compared with the Xxxxxxx-Day test. A p-value ≤ 0.10 was considered significant in addition to biologically plausible interaction terms. A multivariate logistic regression model was used to analyze the relationship between MRSA nasal colonization and subsequent infection. All covariates significant in univariate analysis (P < 0.10) and those considered clinically or epidemiologically relevant we...
SAMPLE SIZE CALCULATIONS. The American Psychological Association (APA; 1999) recommends a minimum sample size of 100 for a norm-referenced test. Further, for the development of regression-based norms for a psychological test which is short in length, with a small number of item scores, a minimum sample size of 100 is recommended (Oosterhuis, van der Ark, & Sijtsma, 2016). The previous unpublished norming study on the CVWMT-I (Xxxx, 2016) had a sample size of N = 37. Based on these recommendations, the previous study and time constraints on the current study, the statistical tool, Quantifying the Uncertainty Attached to Normative Data (QUAND; Xxxxxxxx & Garthwaite, 2008) was used to compare the 95% confidence intervals for a sample where N = 37 and N = 100. At the 50th percentile, a substantially narrower confidence interval was obtained for N = 100 (CI: 15 points) compared with N = 37 (CI: 25 points). Hence, this larger normative reference group was thought to represent a substantial improvement relative to the previous study. Based on this, the current study aimed to collect data for a sample of 100 participants for the norming study. The statistical package ICCSampleSize in R package version 1.0 (Rathbone, Shaw & Xxxxxxxx, 2015) was used to identify the sample size needed for test-retest and inter- rater reliability using intra-class correlation (ICC). To detect an effect size of 0.8, a sample size of 8 was found to give 80% power to detect effects of d = 0.8 where p = 0.05. A power analysis was conducted using G*Power (Xxxx, Erdfelder, Xxxx, & Xxxxxxx, 2007) to identify the sample size needed for parallel-form reliability. To detect a moderate effect size (0.6) a sample size of 45 for each group was found to give 80% power to detect effects of d = 0.6 or greater where p = 0.05. A power analysis was conducted to identify the sample size required for a construct validity analysis (i.e. comparing test performance on the CVWMT-I between the clinical and non-clinical groups). To detect a moderate effect size (0.5), guided by a preliminary study comparing scores on the ACT between controls and participants with severe TBI (Merkley, Xxxxxx, Xxxxxx & Good, 2013), a sample size of 51 for each group was found to give 80% power to detect effects of d = 0.5 or greater, where p = 0.05. To detect a correlation of 0.6 for convergent validity, a sample size of 47 was found to give 80% power to detect effects of d = 0.6 or greater where p = 0.05, using G*Power (Xxxx et al., 2007). Table 1.1 belo...
SAMPLE SIZE CALCULATIONS. The sample size (N = 36 [to allow at least 32 enrolled subjects to complete]) for this study is based on clinical and practical considerations and not on a formal statistical power calculation. The sample size is considered sufficient to effectively assess the PK and safety profiles of TPOXX.
SAMPLE SIZE CALCULATIONS. The sample size for this study is approximately 60 subjects. Each of the 3 cohorts will include 20 subjects (15 subjects receive AT-752 and 5 subjects receive placebo). XxXxx Low et al (Low 2014) , the observed standard deviations for VLR, defined as mean change from baseline viral load on days 2, 3, and 4, ranged from 0.75 (N=26) to 1.07 (N=24). Taking into account that the VLR endpoint in that study was based on an average of 3 measurements and assuming that there is some correlation between measurements on days 2 through 4, we consider power to detect a true effect at a single time point using a one-sided
SAMPLE SIZE CALCULATIONS. The study will employ a Group Sequential Design (see Section 6.2.2.2) and is powered accordingly. In order to estimate the required numbers of deaths and randomised subjects required to provide at least 90% power for a true (absolute) difference of 4.3% compared to a placebo mortality rate of 17% at 3 years, the computer program PEST4 was used [MPS Research Unit, 2000]. If the true rates of survival among subjects randomised to treatment are 87.3% with SERETIDE and 83% with placebo, then the hazard ratio for treatment with SERETIDE compared to placebo is HR = 0.728. A total of up to 440 deaths in these two groups would provide 90% power to detect such a difference. Note that the possibility of satisfying the stopping rule at either of the planned interim analyses implies that the actual number of deaths in this study may be somewhat lower than this upper limit. For comparative purposes, using a fixed sample approach, the number of deaths required to provide this level of power is estimated as 418. In order to provide 440 deaths, it is estimated that approximately 1510 subjects must be randomised to either group. If the true mortality rates for SERETIDE and FLIXOTIDE are similar, then a similar number of deaths (440) among subjects treated with either FLIXOTIDE or placebo will provide 90% power to detect a true difference between FLIXOTIDE and placebo equivalent to a hazard ratio of 0.728 or less. The effect of Salmeterol on mortality is expected to be somewhat lower than for either SERETIDE or FLIXOTIDE. The primary objective of the study with regards the inclusion of the Salmeterol arm is therefore to provide as precise an estimate of the magnitude of the mortality effect as is practically achievable. Therefore, it is planned that approximately 1510 subjects will be randomised to each of the four treatment groups in order to provide up to 880 deaths. However, if the actual mortality rates are either lower or higher than anticipated, the actual numbers randomised may be amended compared to the current plan.
SAMPLE SIZE CALCULATIONS. For Part 1, up to 48 subjects are planned: 8 subjects each in the mild (Cohort A), moderate (Cohort B), and severe (Cohort C) hepatic impairment cohorts and 8 to 24 healthy control subjects with normal hepatic function (Cohort D). For Part 2, at least 8 subjects with mild and moderate or moderate and severe hepatic impairment and up to 8 healthy control subjects with normal hepatic function are planned. The sample size was chosen for this study based on clinical and practical considerations and not on a formal statistical power calculation. The US Food and Drug Administration Guidance for Industry, Pharmacokinetics in Patients With Impaired Hepatic Function: Study Design, Data Analysis, and Impact on Dosing and Labeling (DHHS 2003) was used as a guide to select the sample size, which is considered sufficient to adequately assess the safety and PK profiles of aramchol. In the event of early withdrawals or discontinuations, subjects may be replaced at the discretion of the investigator, and after consultation with the medical monitor and sponsor if the number of completers will be less than 6 subjects in any cohort.
