Download Statistical Analysis Plan for Dvectis Pads in Chronic Lumbar Pain Trial and more Lecture notes Statistics in PDF only on Docsity!
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 1 / 34
STATISTICAL REPORT
Clinical trial evaluating the effectiveness and safety of Dvectis
Single and Dvectis Double pads in comparison to “no pad use” in
patients with chronic lumbar spine pain
Protocol code: DVE- 17
Tested devices Dvectis Single
Dvectis Double
Indication Chronic lumbar spine pain
Design Randomized controlled clinical trial
Study start 9 Apr 2018
Study end 20 Aug 2018
Principal investigator MUDr. Martin Holinka
Department of Orthopaedics
Karvinská hornická nemocnice a.s.
Zakladatelská 975/
735 06 Karviná - Nové Město
Contracting entity DD - Hippero s.r.o.
třída Kpt. Jaroše 1922/
602 00 Brno
Document author PharmTest s.r.o.
Hradec Králové
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 2 / 34 DOCUMENT VERSION
Document version: Final
Version date: 07 - 12 - 2018
DOCUMENT VERSION HISTORY
Version Version date Author Commentary
Final 07 - 12 - 2018 Dalibor Svoboda Formal finalization only
CONTENTS
- Version: Final / 07 - 12 -
- Protocol: DVE- 17 Page 4 /
- 1 List of Abbreviations
- 2 Introduction
- 2.1 Document Purpose and Objective
- 2.2 Description of the Trial, Objectives, Hypotheses
- 2.2.1 Trial Description
- 2.2.2 Primary Objective, Quantity and Hypothesis
- protocol Detected changes in comparison to statistical analysis planned in study
- 2.4 Detected changes in comparison to statistical plan
- 3 Analysed Populations and Number of Subjects
- 3.1 ARS Population
- 3.2 ITT Population
- 3.3 PPS Population
- 3.4 Safety Population
- 3.5 Method of Defining Analysed Populations for Final Analysis
- 3.6 Overview of Analysed Populations and Patient History
- 3.7 Selection Size Calculation
- 4 Missing Data
- 5 Sequential and Final Analysis
- 6 Characteristics of Study Subjects and Treatment Regimen Compliance
- treatment Demographic quantities and other quantities measured before the start of
- 6.1.1 Analysis
- 6.1.2 Presentation of Results
- 6.2 Treatment regimen compliance
- 7 Efficacy
- 7.1 Primary quantity
- 7.1.1 Basic Information
- 7.1.2 Ordering
- 7.1.3 Final Analysis Related to Primary Hypothesis (Dvectis Single Pad)
- 7.1.4 Final Analysis Related to Dvectis Double Pad Efficacy
- 7.1.5 Primary Quantity Exploratory Analysis
- Version: Final / 07 - 12 -
- Protocol: DVE- 17 Page 5 / - 7.1.6 Final Analysis of Other Clinical Trial Objectives - 7.1.7 Presentation of Results - Quantities 7.1.7.1 Aggregate Statistics for Primary Quantity and Related Derived - 7.1.7.2 Final Efficacy Analyses of Both Pads (PID6) - 7.1.7.3 PID6 Exploratory Analysis
- 7.2 Secondary Quantities
- 7.3 Other Quantities
- 7.3.1 Analysis
- 7.3.2 Method of Result Presentation
- 8 Safety Analysis
- 8.1 Adverse Events and Effects
- 8.1.1 Analysis
- 8.1.2 Method of Result Presentation
- 9 Interpretation and Summary
- 10 References
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 7 / 34 2 Introduction 2.1 Document Purpose and Objective Presentation of results of the clinical trial data statistical analysis. 2.2 Description of the Trial, Objectives, Hypotheses
2.2.1 Trial Description
The trial was a monocentric, open, randomized, 3 - arm clinical trial that took place at the Department of Orthopaedics of the Karviná Miners' Hospital (Karvinská hornická nemocnice a.s.). The involvement of a maximum total number of 198 patients with chronic lumbar spine pain was planned; the patients were randomly and evenly assigned to one of the 3 treatment groups:
- Use of the Dvectis Single pad;
- Use of the Dvectis Double pad;
- Use of no pad. One sequential interim analysis was planned in the middle of the clinical trial and after this interim analysis, the trial was stopped early due to proven efficacy. The expected period of participation of each patient in the clinical trial was 6 weeks (± 5 days). For an overview of the individual study visits and procedures, study population and other detailed information, see the study protocol. According to the plan, the assessment subject was asked during Visit 1 and Visit 2 about the most intensive pain felt in the last 48 hours (PI). The pain was recorded by the subject in the visual analogue scale (VAS) in CRF under supervision of the investigator. Then the assessment subject recorded the intensity of their pain independently in the Patient Journal daily.
2.2.2 Primary Objective, Quantity and Hypothesis
The primary objective of the clinical trial was to assess the efficacy of using the Dvectis Single pad in comparison to “use of no pad” in patients suffering from chronic lumbar spine pain. PI was the primary quantity. The primary hypothesis was the superiority of Dvectis Single based on a check, assessed based on the difference in PI ( PID ) between week 2 and week 6 ( PID6 ). 2.3 Detected Changes in Comparison to Statistical Analysis Planned in Study Protocol Contrary to the study plan, the clinical trial took place in a single study centre. Contrary to the plans included in the study protocol, the factor of the study centre was not accounted for in the analyses in any way
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 8 / 34 and the presence of interaction between treatment and the study centre was not analysed. For the same reason, the plan to possibly combine small centres was not used either. There were no other changes detected in the plan of the first sequential interim analysis. 2.4 Detected Changes in Comparison to Statistical Plan The statistical plan was prepared before the statistical analysis performance and formally approved. It included technical details for the statistical analysis performance and the method of result presentation. The only deviation from the statistical plan is related to the summaries for the treatment regimen compliance assessment (see Table 3 Treatment regimen compliance). The original plan was based on the evaluation of cumulative sitting time. However, the average times were analysed in reality. This is due to the fact that the cumulative (or individual) sitting times were not included in the data sheet, therefore they were not included in the analytical datasets either. From this point of view, it is an error in the statistical analytical plan without a practical effect.
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 10 / 34
- Serious treatment regimen violation or another serious deviation from the clinical trial plan. The monitoring reports and discrepancy report implied:
- No study subject that has not met a participation criterion has been registered;
- No study subject with incorrectly allocated treatment has been identified;
- No use of unauthorized concomitant treatment has been documented;
- There was no PI0 value missing;
- Patients 45, 53 and 94 ended the study early; the number of PI values missing is over 25% (83%).
- The following patients violated the treatment regimen in terms of non-observance of the minimum time of sitting on the pad (21, 23, 35, 42, 55, 57, 60, 72, 75, 76, 84, 86, 88). Thus there was a serious deviation from the protocol in 16 study subject, i.e. 15.7%. The PPS population represented the secondary analyzed population for the assessment of efficacy. Its task was to demonstrate the robustness of the main clinical trial results. For this reason, comparisons of the tested group efficacy to the control group were used (in addition to ITT) in final analyses using PI in order to achieve confirmation of the main clinical trial conclusions. 3.4 Safety Population According to the plan, the safety population should have constituted of subjects from ARS, except for those with whom all contact was lost immediately after Visit 1 and those who were placed (in reality) in one of the tested groups but were not given a pad. A pad was given to all patients in the tested groups. All contact was lost with patients 53 and 94 and they were eliminated from the safety population. Patients in the safety population were analyzed based on the treatment groups they were really placed in, regardless of what group they had been randomized to. 3.5 Method of Defining Analyzed Populations for Final Analysis The required analyzed populations were defined before the analysis performance according to the plan.
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 11 / 34 3.6 Overview of Analysed Populations and Patient History Table 1 Numbers of subjects in treatment groups and analysed populations Dvectis Dvectis No treatment Total Single Double ARS population 35 34 33 102 ITT population 35 34 33 102 PPS population 29 25 32 86 Safety population^35 33 32 3.7 Selection Size Calculation The selection size calculation was based on the estimated difference in efficacy expressed by the primary quantity between the Dvectis Single group and the control group, which was 15 mm. Due to the degree of uncertainty in expectations and considering the fact that even differences of approximately 10 mm are of obvious practical and clinical significance, the interim analysis was planned to allow the detection of this minor difference strongly enough, even if the original expectations had been too optimistic. The following assumptions were applied to the determination of the number of patients:
- The primary hypothesis is based on a comparison of to arms, namely the Dvectis Single pad and the control group. For this purpose, the Dvectis Double arm plays no direct role in the selection size.
- The null hypothesis is based on zero difference between the compared groups.
- The alternative hypothesis corresponds to the difference of 10 mm in favour of the Dvectis Single pad.
- The spread of results is the same in both compared groups, is not efficacy- dependent and corresponds to the standard deviation of 15 mm. This value was estimated based on previous studies.
- The null hypothesis test is unilateral, at the significance level of α = 0.025.
- Test strength of 90% is required to detect the difference corresponding to the alternative hypothesis (regardless of the stage when the clinical trial will be completed).
- At most 2 sequential analyses will be performed, equidistantly. The first will be performed in the middle of the expected maximum number of study subjects.
- After the first sequential analysis (interim analysis), the study may be ended due to efficacy (success) based on the clearly defined completion criteria stated above.
- The test statistics limits have been determined according to Pocock (i.e. they are the same for both sequential analyses and guarantee the maintenance of cumulative probability of type I error occurrence at the previously defined level of significance specified in (5).
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 13 / 34 4 Missing Data In the ARS population analyses, the missing data were be treated as missing according to the plan, i.e. stated in summaries as total numbers of missing data; however, they were excluded from the calculations of other statistics. In reality, only the time of sitting on the pad is of significance. In all analyses of the PI quantities, the method of imputation of the last observed value ( Last Observation Carried Forward – LOCF) was used, in accordance with the plan. The plans for a situation when the PI1 would be missing were not used since the situation did not occur. The use of LOCF generally also leads to a bias but with regard to the observed decrease in the PI values during the patient's participation in the study in the tested groups, compared to the reference group, such an approach is obviously conservative. For evaluation of the Oswestry questionnaire, the use of LOCF was planned primarily at the level of individual questions as well as the use of the scoring manual procedure accounting for missing values, if any. This procedure was beyond the data administration and statistical analysis activities since the questionnaire was evaluated by the investigator, entering only the total score in the patient record sheets. If the total score was missing, it was processed similarly to the processing of missing values in the ARS population analysis (i.e. solely elimination from the analysis).
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 15 / 34 6 Characteristics of Study Subjects and Treatment Regimen Compliance For the description of the study population within the sequential interim analysis, summaries of demographic quantities were calculated and presented, as established before the start of the treatment in the scope determined by the statistical analysis plan. The summaries were presented for the respective treatment groups (for all 3 treatment groups). The significance of the differences in these characteristics was not tested. The treatment regimen compliance was evaluated by a calculation of aggregate statistics for the average time of the device use in the course of participation in the study. 6.1 Demographic Quantities and Other Quantities Measured before the Start of Treatment
6.1.1 Analysis
For the analyzed quantities, summary statistics were calculated in the respective treatment groups and as a total. The analyzed ARS population was used.
6.1.2 Presentation of Results
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 16 / 34
Table 2 Demographics and other quantities before the start of
treatment
Dvectis Single Dvectis Double No treatment Total N = 35 N = 34 N = 33 N = 102 Age (year) Minimum 29 26 37 26 Average (SD 1) ) (^) 50.1 (9.1) 45.9 (9.9) 51.1 (7.5) 49.0 (9.1) Median 52.0 47.0 52.0 50. Maximum 65 60 63 65 Total 2)^ 35 34 33 102 Sex, N (%) Male 3 (8.6%) 6 (17.6%) 7 (21.2%) 16 (15.7%) Female 32 (91.4%) 28 (82.4%) 26 (78.8%) 86 (84.3%) Total 2)^ 35 34 33 102 Body weight (kg) Minimum 43 52 53 43 Average (SD 1) ) (^) 78.7 (18.3) 78.4 (13.7) 78.3 (16.8) 78.5 (16.3) Median 75.0 78.5 78.0 77. Maximum 120 103 123 123 Total 2)^ 35 34 33 102 Body height (cm) Minimum 153 148 158 148 Average (SD 1) ) (^) 168.1 (8.3) 168.4 (8.0) 168.7 (7.3) 168.4 (7.8) Median 168.0^ 168.0^ 168.0^ 168. Maximum 189 185 187 189 Total 2)^ 35 34 33 102 Pain intensity before treatment (mm VAS) Minimum 45 45 43 43 Average (SD 1) ) (^) 62.6 (14.2) 63.0 (13.7) 59.5 (12.8) 61.7 (13.5) Median 60.0 60.0 59.0 60. Maximum^98 90 96 Total 2)^ 35 34 33 102 Analyzed population: ARS Dataset: Baseline Data image: 2018 - 12 - 05 Comments:
- SD = Standard deviation.
- Number of analyzed data (without missing values).
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 Page 18 / 34 7 Efficacy 7.1 Primary Quantity
7.1.1 Basic Information
The primary quantity was pain intensity ( PI ) measured on a 100 mm visual analogue scale (VAS). Measurements were taken once a week in weeks 1 - 6 ( PI1, PI2, …, PI6 ). The basic aggregate statistics for each treatment group were calculated and presented individually each week. The pain intensity difference ( PID ) was calculated from PIi in weeeks 2 – 6 based on the relationship: 𝑃 I𝐷𝑖 = 𝑃𝐼 1 − 𝑃𝐼𝑖, 𝑖 = 2,…, 6
The basic aggregate statistics for each treatment group were calculated and
presented for the respective PIDi.
PID 6 was used for testing the primary hypothesis, statistically formulated in the
clinical trial plan (protocol) as follows: If we mark 𝜇𝐷S the mean value PID 6 for Dvectis
Single and 𝜇𝑅ef the mean value PID 6 for the control (reference) group, then the primary
null hypothesis H
H0: 𝜇𝐷S − 𝜇𝑅ef ≤ 0 is tested against the alternative HA HA: 𝜇𝐷S − 𝜇𝑅ef > 0 unilaterally at the significance level of α = 0.025. The testing itself is not a part of the final analysis presented in this document since the test and the subsequent efficacy conclusion were made based on the sequential analysis results. The analysis of PID6 was performed in the ANOVA model. The model included the treatment factor and the PI1 covariate. The results are presented in the form of a standard ANOVA table, including the sums of type III squares (in the SAS terminology).
7.1.2 Ordering
Ordering is significant for the performance of the final analysis. Testing of hypotheses, calculations of p - values and the construction of intervals were based on the notion that the resulting statistics can be ordered, the ordering criterion being their extremeness while the null hypothesis is valid. E.g. the Z - statistics is the more extreme the higher its (absolute) value. The performance of sequential analyses is complicated due to the fact the result is not a number but a pair of numbers ( M , Z ), constituted by the number of the interim analysis (stage) when the clinical trial is stopped within the performed sequential analyses and the calculated Z - statistics. Therefore a method of ordering the results by extremeness had to be defined. The ordering definition allows decisions to be made on the probability of inequality
Version: Final / 07 - 12 - 2018 Protocol: DVE- 17 (^1 ,^ 1)^ >^ (^ 2,^ 2)^ , Page 19 / 34 for arbitrary M1, M2, Z1, Z2. However, the optimal method of ordering cannot be generally determined since the sequential statistics densities do not have a monotonic value relationship, therefore the selection criterion may not be applied with regard to the test with the highest power. In general, there are several intuitive options. According to the plan in the study protocol and the statistical analysis plan, the results related to the primary hypothesis were ordered based on stage-wise ordering (analysis time ordering / Fairbanks and Madsen ordering ) [1]. A number of authors consider this option intuitively attractive, for the following reasons, among others:
- The P - value obtained during the final analysis is lower than the level of significance α used in the sequential test in the interim analysis (after the first and second stage) only when the null hypothesis is rejected [2]. This feature is important in relation to the internal consistence of the results and is generally not demonstrated by other types of ordering.
- When the clinical trial is stopped early, the P- value is not based on the information quantity, or the number of patients that would be included if the trial was not stopped early [2].
7.1.3 Final Analysis Related to Primary Hypothesis (Dvectis Single Pad)
Clinical trial stopping based on the observation of an extreme result during sequential analyses diverts the point estimates made in a way common for a design with a fixed number of patients without sequential analyses. Detailed information is given in the study protocol and the statistics plan. In this particular case, the clinical trial was stopped early after the first sequential interim analysis. This fact, with the application of the ordering described above, led to estimates identical to the naive 95% reliability interval calculated for the fixed selection size.
7.1.4 Final Analysis Related to Dvectis Double Pad Efficacy
The test of the Dvectis Double pad difference from placebo expressed an important secondary objective of the clinical trial, which was to allow additional statements concerning efficacy. The difference significance test could have been performed solely if the null hypothesis expressing the primary objective was rejected in any stage of the clinical trial. This condition was met after the first sequential analysis. The secondary hypothesis test was performed just once. The condition of test performance solely if the primary null hypothesis is rejected is a type of hierarchy testing when it is easy to maintain the total cumulative probability of type I error occurrence ( familywise error rate ) at a previously specified level α simply by the fact both tests are performed at the same level α individually (hierarchically), which may be formally proven with the use of the closed testing principle [6]. However, this procedure cannot be applied to sequential interim analyses exactly, therefore in the case of this clinical trial, it does not allow strict control of error inflation probability α. Hung [7] proposes an intuitive strategy based on secondary hypothesis testing at significance level α, which corresponds to (1- α ) percentile of the standard Gaussian distribution, if the primary null hypothesis was rejected at the total significance level α (i.e.
