STATISTICAL ANALYSIS PLAN forinsert the short title of the trial
Instructions are in red.
The text provided in this template is for guidance and can be modified according to the protocol. Additional details should be added, as necessary.
Delete all coloured text from the completed document
Detailed guidance on what should be considered when writing a statistical analysis plan is given in the ICH E9 guideline, section 5 (Data Analysis Considerations).
Sponsor nameSponsor address
EudraCT number / If not applicable write “not applicable”
Trial ID / Insert short title or protocol number
Insert the protocol title
SIGNATURE PAGE
SPONSOR REPRESENTATIVE:Name, title
Signature / Date (dd/mmm/yyyy)
PRINCIPAL/COORDINATING INVESTIGATOR:
Name, title
Affiliation
Signature / Date (dd/mmm/yyyy)
TRIAL STATISTICIAN:
Name, title
Affiliation
Signature / Date (dd/mmm/yyyy)
OTHER:
Name, title
Affiliation
Signature / Date (dd/mmm/yyyy)
ABBREVIATIONS
AE / Adverse EventATC / Anatomical/Therapeutic/Chemical
BMI / Body Mass Index
BPM / Beats Per Minute
CI / Confidence Interval
DMC / Data Monitoring Committee
ECG / Electrocardiogram
eCRF / electronic Case Report Form
EDC / Electronic Data Capture
MedDRA / Medical Dictionary for Regulatory Activities
SAE / Serious Adverse Event
SAS / Statistical Analysis System
SD / Standard Deviation
SOC / System Organ Class
WHO / World Health Organization
TABLE OF CONENTS
1STUDY OBJECTIVES
1.1Primary Objective
1.2Secondary Objectives
2OVERALL STUDY DESIGN
3GENERAL STATISTICAL CONSIDERATIONS
4HYPOTHESES AND DECISION RULES
4.1Statistical Hypotheses
4.2Statistical Decision Rule
5DEFINITIONS AND DERIVED VARIABLES
5.1Change from baseline
5.2Other calculations
5.3Safety definitions
5.4Previous and Concomitant medications
6EFFICACY AND SAFETY ENDPOINTS / VARIABLES
6.1Primary Endpoint
6.2Secondary Endpoints
6.3Safety Parameters
7DETERMINATION OF SAMPLE SIZE
8DATA SETS TO BE ANALYSED
8.1Treatment Misallocation
9BLIND REVIEW
10STATISTICAL METHODOLOGY
10.1Adjustment for covariates
10.2Centre effect
10.3Multiplicity adjustments
10.4Demographic and other baseline characteristics
10.5Primary efficacy analysis
10.6Secondary efficacy analyses
10.7Patient Characteristics
10.8Exploratory Analysis
11Safety Analysis
11.1Adverse events/adverse device effects
11.2Other safety assessments
12Interim analysis
13Data monitoring
14Table of Contents of Tables, Listings and Figures
14.1List of Tables
14.2List of Listings
14.3List of Figures
15References
16Appendices
AMENDMENTS TO THE SAP
List amendment to any previous approved versions of the SAP or write “Not applicable”.
1STUDY OBJECTIVES
The study objectives should usually be quoted directly from the study protocol/ clinical investigation plan.
1.1Primary Objective
1.2Secondary Objectives
2OVERALL STUDY DESIGN
Procide a summary of the study design from the protocol.
3GENERAL STATISTICAL CONSIDERATIONS
All analyses described in this plan are considered a priori analyses in that they have been defined in the protocol and/or this SAP. All post hoc analyses will be identified as such in the Clinical Study Report.
All categorical (binary and ordinal) data will be summarised using frequency counts and percentages of patient incidence. Percentages will be calculated using the study population; any exceptions to this will be highlighted in the table footnote. The continuous variables will be summarised using number of patients (N), mean, standard deviation (SD), median, 25/75 percentile and range (minimum/maximum). In general, minimum and maximum will be presented to the same degree of precision as data is recorded, with mean and median having 1 additional place after the decimal and standard deviation having 2 additional places after the decimal. Percentages less than 100 will be displayed to 1 place after the decimal, where space permits.
For skewed data the interquartile range will be displayed.
Insert item description data will be summarised by n, geometric mean and its 95% confidence interval (CI), as well as the minimum and maximum values.
P values will be displayed to three decimal places after rounding. P values below 0.001 will be displayed as “<0.001”.
Data will be presented, unless otherwise indicated, according to the two / three / nnn treatment groups.
In general, analyses presented in the tables, listings, and figures will be confined to those subjects included in the (safety analysis set, full analysis set and per protocol set (see below), or a specific subset thereof. Data listings will include all randomised study subjects unless otherwise noted. Subjects will be identified in the listings by a variable which uniquely combines site (investigator) number and subject number. Listings will be sorted by this unique subject number.
Study Days will be computed from the date of first administration, with Study Day 1 representing the first treatment date.
AE duration will be calculated as (Stop Date – Start Date) + 1.
4HYPOTHESES AND DECISION RULES
Include a precise statement of the primary statistical hypothesis to be tested, possibly with statement on how to address multiple testing if more than one hypothesis is being tested
4.1Statistical Hypotheses
This protocol is designed to establish the superiority of Insert study treatment description compared to Insert comparator descriptiontreatment with regard to Insert item description.
The null hypothesis is that there is no difference in the probability of Insert variable / endpoint description between the two treatment regimes
The alternative hypothesis is that there is a difference in the probability of Insert variable / endpoint description between the two treatment regimes
This protocol was designed to establish the non-inferiority of Insert study treatment descriptioncompared to Insert comparator descriptiontreatment with regard to Insert variable / endpoint description.
The null hypothesis is that Insert study treatment descriptionis inferior to Insert comparator descriptiontreatment with regards to the proportion of patients < > by a 15% inferiority margin.
The alternative hypothesis is that Insert study treatment descriptionis non-inferior with regards to the proportion of patients with Insert variable / endpoint descriptionby a 15% inferiority margin
If the primary null hypothesis is rejected, a superiority test on the 5% significance level will be performed to assess the superiority Insert study treatment descriptionover Insert comparator descriptiontreatment. The superiority null hypothesis will be that Insert study treatment descriptionis equal or superior to Insert comparator description with regards to the proportion of Insert variable / endpoint description, while the alternative hypothesis test is that Insert study treatment description is superior to Insert comparator description.
4.2Statistical Decision Rule
This protocol is designed to address a single primary endpoint. Non-inferiority is claimed if the null hypothesis is rejected on the significance level (alpha) of 0.025 (one-sided). That is, if the upper limit of the 95% confidence interval for the treatment difference is less than 15%.
Superiority of Insert study treatment descriptionover Insert comparator descriptionis claimed if the primary non-inferiority null hypothesis is rejected, and also the superiority null hypothesis is rejected on the one-sided 0.025 significance level.
This protocol is designed to address a single primary endpoint. Statistical significance is claimed if the null hypothesis is rejected on the significance level (alpha) of 0.05 (two-sided).
That is, if the p-value of the null hypothesis test is less than or equal to 0.05.
5DEFINITIONS AND DERIVED VARIABLES
For all planned measures, visits should occur within a window of the scheduled visit, see table below for definitions.
Provide visit definitions, the study epochs etc.
Visit Label / Target Day / Definition (Day window)Visit 1 (Screening) / -1 / Prior to Day 1
Visit 2 (Baseline) / Day 1 / (Randomisation) Day 1
Visit 3
Last trial visit*
*The last study visit is defined as the visit following the last visit with randomised treatment, and where there is a study end statement.
For analysis and tabulation purposes, we define study time points as:
Time Point Label / Target Day / Definition (Day window)TP1. Baseline / Day 1 (Randomisation) / Information up to randomisation
TP2. Week X / n / Days n to nn
TP3. Week x / n / Days nn to nn
If more than one visit falls into the same time point interval, information on all visits will be used in theanalyses.
State how any derived variables are computed e.g. SF-36, BMI, change from baseline, time to event-variables etc. for all efficacy and safety variables.
5.1Change from baseline
Change from baseline (Δ) = time-point value - baseline value.
% change from baseline (%Δ) = [(time-point value – baseline value) / baseline value] *100%
5.2Other calculations
Age (years) = [(date of baseline – date of birth)/365.25].
BMI = weight in kilograms / (height in metres) x (height in metres)
BMI will be categorized according to the WHO definitions for underweight, normal,overweight and obese.
Area under the curve (AUC) will be calculated as the integral under the measure curve usingtrapezoids.
Time of withdrawal = date of withdrawal – date of randomization +1
5.3Safety definitions
5.3.1Treatment emerging adverse events
Treatment emerging adverse events (TEAEs) are defined as AEs with a start date on or after the first treatment.
5.3.2Past disease and concomitant disease
Past disease/condition
A disease/condition is considered as past disease/condition if it is not ongoing at randomisation.
Concomitant disease
A disease/condition is considered as concomitant disease/condition if it is ongoing at randomisation.
5.4Previous and Concomitant medications
The following definitions will be used:
- previous medication (stop date < date of randomisation);
- concomitant medication (stop date ≥ date of randomisation or ongoing at study end)
In case of missing or incomplete dates/times not directly allowing allocation to any of the two categories of medications, a worst-case allocation will be performed according to the available parts of the start and the end dates. The medication will be allocated to the first category allowed by the available data, according to the following order:
- concomitant medication
- previous medication
6EFFICACY AND SAFETY ENDPOINTS / VARIABLES
Describe all efficacy and safety variables for the primary and any secondary endpoints.
6.1Primary Endpoint
6.2Secondary Endpoints
6.2.1Efficacy
6.2.2Patient Reported Outcomes (PRO)
6.3Safety Parameters
7DETERMINATION OF SAMPLE SIZE
The sample size has been based on the primary endpoint of Insert variable / endpoint description.
The assumed rate of Insert variable / endpoint description is xx% in the Insert study treatment description group and xx% in the Insert comparator description group, giving a treatment difference of xx%.
On basis of the assumptions, a total sample size (study completers) of nnn (nn in each group) was deemed necessary to achieve 80% power to detect a difference of xx% between the groups in a two-sided test at 5% significance level.
To compensate for withdrawals/loss to follow-up, a total enrolment of nnn patients was targeted.
8DATA SETS TO BE ANALYSED
The Safety Analysis Set will include all subjects who were exposed to treatment, i.e. at least one dose.
The Full Analysis Set (FAS)will include all randomised subjects with a few exceptions,
The following exceptionswill be allowed:
The Per Protocol Set (PPS) isthesubset of the subjects in the FAS who are with no major protocol deviations .
To be included in the FAS subjects must meet the following criteria:
It should be stated for which analyses the different analysis sets will be used.
8.1Treatment Misallocation
If patients were:
- Randomized but not treated: patient will appear on the patient evaluation table as randomized but not treated; this is the extent of how much the patient will be reported.
- Treated but not randomized: then by definition the patient will be excluded from both the efficacy and safety analyses since randomized treatment is missing
- Randomized but did not follow protocol according to allocation: then they will be reported under their randomized treatment group for all efficacy and safety analyses as part of the FAS and safety analyses, but omitted from the PPS
Note that treated in this section is defined as having completed at least one post-baseline visit.
9BLIND REVIEW
Describe any blind statistical reviews of the data.
10STATISTICAL METHODOLOGY
The primary efficacy analyses will be based on the Per Protocol Analysis Set (PPS). Secondary efficacy analyses will be based both on the PPS and the Full Analysis Set (FAS). As there is only one identified primary analysis, there will be no adjustments for multiple testing in the secondary analyses.
When random numbers are warranted for inference (such as for bootstrapping and multiple imputation), the seed will be set to the date the statistical analysis plan is signed-off (in the format yyyymmdd).
All efficacy analyses will be presented by the size (point estimate) of the difference between the treatments and the associated 95% confidence interval. P-values are generally not presented as this is a non-inferiority trial, and presentation of the p-value could be misinterpreted.
All statistical analyses will be done in Stata v14 (StataCorp. 2015. Stata Statistical Software: Release <insert release number. College Station, TX, USA).
10.1Adjustment for covariates
Describe how adjustments will be made for any covariates.
10.2Centre effect
Describe how adjustments will be made for any centre effects.
10.3Multiplicity adjustments
The handling of multiplicity problems should always be discussed and the statistical analysis should reflect awareness of this.
10.4Demographic and other baseline characteristics
Demographic and baseline characteristics will be summarised descriptively by treatment group using the FAS.
The demographic characteristics include age, sex, race and ethnicity. A subject’s integer age in years will be calculated to the date of informed consent. Baseline characteristics include the randomisation stratum (insert definition), height, weight, and BMI.
A descriptive summary of demographic and baseline data will also be provided for the ECG, VAS and >.
In addition, demographic and baseline data will be displayed separately for each of the randomisation strata in the FAS.
Demographic and baseline characteristics data will be presented in a data listing.
10.5Primary efficacy analysis
The primary endpoint (insert definition) will be analysed using
, adjusted for
Robustness of the results regarding centre effect will be checked using
a multi-level mixed logistic model approach with centre as random variable, and by generalised estimating equations.
The primary analysis will be performed on the primary analysis set, (the per protocol set), consisting of all patients having followed the protocol to maximise the probability of finding a difference if indeed there is one.
10.5.1Missing data imputation and sensitivity analyses
There will be no missing data for the primary endpoint in the primary analysis set (the PPS).
Robustness analysis will be performed on the Full Analysis set, where patients who withdrew will be treated as insert definition of treatment failure (worst case imputation).
10.6Secondary efficacy analyses
10.7Patient Characteristics
10.7.1Patient Disposition
The disposition of all patients will be listed and summarised by treatment arm. The number and percentage of patients who are randomised, received any study treatment, prematurely discontinued from treatment and lost to follow-up will be summarised.
The number and percentage of patients will be categorized by the reason(s) for
1) Discontinued from treatment: This is when it is decided that a patient will not receive furthertreatment according to study protocol. Reasons can be: Violation of eligibility criteria,withdrawal of informed consent, adverse event, lack of efficacy or disease worsening, lowdrug concentrations or anti-drug antibody formation and other.
2) Lost to follow-up: This is when the patient withdraws or is withdrawn from the study.
Reasons can be: violation of eligibility criteria, patient withdrawal of informed consent,adverse event, lost to follow-up, death, investigator decision or other.
10.7.2Protocol Deviations
Protocol deviations resulting in exclusion from the PPS will be determined and summarised bytreatment group. See section 8 for protocol deviation categories.
10.7.3Background and Demographic Characteristics
Patient demographics and baseline characteristics will be summarised both for the PPS and FAS.
Patient demographics and baseline characteristics will be summarised by randomised treatment armand overall using descriptive statistics (N, mean, standard deviation, median, 25/75 percentiles,minimum, and maximum) for continuous variables, and number and percentages of patients forcategorical variables.
The patient demographics and baseline characteristics to be summarised includeage in years, gender, symptom duration, weight, height, BMI, education, smoking status, CRP, ESR,use of concomitant immunosuppressive medication and diagnosis specific disease activity measures.
Demographics and baseline characteristics will also be summarised by diagnosis.
Medical history will be coded using the MedDRA dictionary (v17.0E) and will be summarised.
Concomitant medication will be coded using the ATC coding system and summarised.
10.7.4Treatment Compliance
Data summarizing the proportions of patients complying with the treatment regimen according toprotocol will be presented by treatment arm
and secondary stratified for .
10.7.5Concomitant Medications and Other Therapies
Concomitant medication information will be coded by the Anatomical Therapeutic Chemical(ATC) classification system. Concomitant medications taken during the study will be summarised bygeneric name. The number and percentage of patients who took at least one drug within each specificpreferred term will be presented. Patients will only be counted once if they are taking more than onemedication (within the same code) or take the same generic medication more than once. If it cannotbe determined whether a medication is concomitant (based on stop date or, if the stop date is missing,start date), then the medication will be considered to be concomitant.
10.7.6Patient reported outcome measure data
Analyses of patient reported outcome measure (PROM) data will be done using the proceduresdescribed for change from baseline continuous endpoints.
10.7.7Concomitant treatment
Concomitant medications will be defined as those medications used during the course of the trial. Partial start dates of prior and concomitant medications will be assumed to be the earliest possible date consistent with the partial date. Partial stop dates of prior and concomitant medications will be assumed to be the latest possible date consistent with the partial date.
Data for all prior and concomitant medications, including indication, dosage, route and frequency of administration, and start and stop dates will be presented in a data listing for the FAS.
10.8Exploratory Analysis
The following exploratory analysis will be performed:
Primary and secondary endpoints will be analysed by treatment for each diagnosis separatelyaccording to the analyses methods described above.
Exploratory endpoints will not be limited to those mentioned above, and will includevariables/endpoints and statistical methods/modelling as necessary to explore the secondary objectivesof the study as described in the protocol.
11Safety Analysis
11.1Adverse events/adverse device effects