Overview Of Ich New E9


Published on

An overview of the ICH E9 guidance. Easy to follow, and I can provide a live presentation of this to your team! Great for those who are not familiar with statistics.

Overview Of Ich New E9

  1. 1. Overview of ICH E9: Statistical Principles for Clinical Trials Presented by Jeff Davidson
  2. 2. ICH E9 cover page <ul><ul><li>INTERNATIONAL CONFERENCE ON HARMONISATION OF TECHNICAL REQUIREMENTS FOR REGISTRATION OF PHARMACEUTICALS FOR HUMAN USE </li></ul></ul><ul><li>ICH HARMONISED TRIPARTITE GUIDELINE </li></ul><ul><li>STATISTICAL PRINCIPLES FOR CLINICAL TRIALS </li></ul><ul><li>E9 </li></ul><ul><li>Current Step 4 version dated 5 February 1998 </li></ul><ul><li>This Guideline has been developed by the appropriate ICH Expert Working Group and has been subject to consultation by the regulatory parties, in accordance with the ICH Process. At Step 4 of the Process the final draft is recommended for adoption to the regulatory bodies of the European Union, Japan and USA. </li></ul>
  3. 3. <ul><li>The great tragedy of science - </li></ul><ul><li>the slaying of a beautiful hypothesis </li></ul><ul><li>by an ugly fact.  </li></ul><ul><li>- Aldous Huxley </li></ul><ul><li>The most important science in the whole world: </li></ul><ul><li>for upon it depends the practical application of every other science </li></ul><ul><li>and of every art: the one science essential to all political and social </li></ul><ul><li>administration, all education, all organization based on </li></ul><ul><li>experience, for it only gives results of our experience.  </li></ul><ul><li>- Florence Nightingale </li></ul>
  4. 4. <ul><li>Focus on statistical principles </li></ul><ul><ul><li>Gives direction to researchers in design, conduct, analysis, and evaluation of trials </li></ul></ul><ul><ul><li>Does not address use of specific statistical tests </li></ul></ul><ul><ul><li>Emphasis on later phase, confirmatory trials </li></ul></ul><ul><li>Target audience: individuals from a broad range of scientific disciplines </li></ul><ul><ul><li>Statisticians, clinicians, pharmacologists, epidemiologists </li></ul></ul>Scope and Direction
  5. 5. <ul><li>Trial Statistician: </li></ul><ul><ul><li>Responsible for all the statistical work associated with the trial </li></ul></ul><ul><ul><li>Ensures statistical principles are appropriately applied </li></ul></ul><ul><ul><li>Has the proper training and experience to implement the principles in this guidance </li></ul></ul>Unjustified statistics are like smiling cats - not to be trusted. Scope and Direction
  6. 6. <ul><li>Minimizing bias </li></ul><ul><ul><li>Systematic tendency of any factors associated with design, conduct, analysis and interpretation </li></ul></ul><ul><ul><li>To lead to an estimate of treatment effect </li></ul></ul><ul><ul><li>Different from the true value </li></ul></ul><ul><li>Maximizing precision </li></ul><ul><ul><li>Obtaining small standard errors and narrow confidence intervals </li></ul></ul><ul><li>Evaluating robustness </li></ul><ul><ul><li>Sensitivity of overall conclusions to various limitations of the data, assumptions, analysis procedures used </li></ul></ul>Scope and Direction
  7. 7. <ul><li>Controlling the type I error </li></ul><ul><ul><li>Ensuring that the chance of declaring a treatment efficacious when it in fact does not work is low (e.g., α ≤ 0.05) </li></ul></ul><ul><li>“ Multiplicity” refers to having more than one opportunity to detect a difference between drugs (e.g., interim analyses, multiple endpoints of interest) </li></ul>Scope and Direction
  8. 8. <ul><li>Principal features of statistical analysis should be clearly specified in the protocol </li></ul><ul><li>Protocol (and amendments) should be approved by a trained statistician </li></ul><ul><li>A detailed Analysis Plan should be written before data analysis begins </li></ul>Protocol and Analysis Plan
  9. 9. Types of Trials <ul><li>Exploratory Trial </li></ul>Confirmatory Trial
  10. 10. Exploratory Trial <ul><li>Clear and precise objectives, however, tests of hypothesis may be data dependent </li></ul><ul><li>Such trials cannot be the basis of the formal proof of efficacy </li></ul>
  11. 11. Confirmatory Trial <ul><li>An adequately controlled trial in which the hypotheses are stated in advance and evaluated </li></ul><ul><li>Key hypothesis of interest </li></ul><ul><ul><li>Follows directly from the trial’s primary objective </li></ul></ul><ul><ul><li>Is always pre-defined </li></ul></ul><ul><ul><li>Is the hypothesis that is subsequently tested when the trial is complete </li></ul></ul><ul><li>Adherence to protocols and SOPs is particularly important </li></ul>
  12. 12. Population <ul><li>Earlier phases may focus on a very narrow subgroup </li></ul><ul><li>Confirmatory trials should more closely mirror target population of the therapy under study </li></ul><ul><ul><li>Issues of Generalizability </li></ul></ul><ul><li>Clear Inclusion/Exclusion criteria </li></ul>
  13. 13. Outcome Variables <ul><li>Primary variable(s) </li></ul><ul><ul><li>Directly related to the primary objective </li></ul></ul><ul><ul><li>Preferable to specify only one reliable and validated variable </li></ul></ul><ul><ul><li>Used in earlier studies or in published literature </li></ul></ul><ul><ul><li>Used when estimating the sample size </li></ul></ul><ul><li>Secondary variables </li></ul><ul><ul><li>Either supportive measurements related to the primary objective, or measurements of effects related to the secondary objectives </li></ul></ul>
  14. 14. Avoiding Bias by Design: Blinding <ul><li>Blinding limits the occurrence of conscious and unconscious bias arising from the influence one’s knowledge of treatment may have on </li></ul><ul><ul><li>Recruitment and allocation of subjects </li></ul></ul><ul><ul><li>Their subsequent care </li></ul></ul><ul><ul><li>Attitudes of subjects to the treatments </li></ul></ul><ul><ul><li>Assessment of endpoints </li></ul></ul><ul><ul><li>Handling of withdrawals </li></ul></ul><ul><ul><li>Exclusion of data from analysis </li></ul></ul><ul><ul><li>Choice of analysis methods </li></ul></ul>
  15. 15. Avoiding Bias by Design: Randomization <ul><li>Introduces a deliberate element of chance into the assignment of treatments </li></ul><ul><li>Provides a sound statistical basis for the comparison of treatment groups </li></ul><ul><li>Tends to produce treatment groups with distributions of prognostic factors (measured and unmeasured) are similar </li></ul>
  16. 16. <ul><li>Parallel group: random assignment to A vs B </li></ul><ul><li>Randomization </li></ul>Design Configuration Population A B
  17. 17. <ul><li>Crossover: </li></ul><ul><li>random assignment to AB or BA </li></ul><ul><li>subject serves as own control </li></ul><ul><li>Randomization Washout period </li></ul>Design Configuration Population A B B A
  18. 18. <ul><li>Factorial Designs: </li></ul>Design Configuration Population A (A1, A2) B (B1,B2) A1, B1 A1, B2 A2, B1 A2, B2
  19. 19. Other Study Designs <ul><li>Cohort Studies </li></ul><ul><li>Case-Control Studies </li></ul><ul><li>Descriptive Studies (e.g., surveys) </li></ul><ul><li>Registries </li></ul><ul><li>Adaptive </li></ul><ul><li>Etc., etc… </li></ul>
  20. 20. Type of Comparison <ul><li>Trials to show “superiority” </li></ul><ul><ul><li>New treatment vs. placebo, or new treatment vs. active control </li></ul></ul><ul><ul><li>Test (and hopefully reject) the null hypothesis that there is no difference in outcomes between groups </li></ul></ul><ul><ul><li>vs. the alternative hypothesis that there is a difference between the groups </li></ul></ul><ul><ul><li>One-sided or two-sided </li></ul></ul>
  21. 21. Type of Comparison <ul><li>Trials to show “ equivalence ” </li></ul><ul><ul><li>New treatment vs. active control </li></ul></ul><ul><ul><li>Test (and hopefully reject) the null hypothesis that the new treatment performs differently than the active control by at least some small amount </li></ul></ul><ul><ul><li>vs. the alternative hypothesis that the difference between the groups is no greater than this small amount </li></ul></ul><ul><ul><ul><li>An amount that is sufficiently small that the treatments are considered equivalent for all practical purposes if the difference between the treatments is smaller than this amount (equivalence margin) </li></ul></ul></ul>
  22. 22. Type of Comparison <ul><li>Trials to show “noninferiority” </li></ul><ul><ul><li>New treatment vs. active control </li></ul></ul><ul><ul><li>Test (and hopefully reject) the null hypothesis that the active control performs better than the new treatment by at least some small amount </li></ul></ul><ul><ul><li>vs. the alternative hypothesis that the new treatment does not perform worse than the active control by more than this small amount </li></ul></ul>
  23. 24. Type of comparison <ul><li>Trials to show “equivalence” </li></ul><ul><ul><li>e.g., test null hypothesis that % cured with active </li></ul></ul><ul><ul><li>Control is ≥ 5% more than the % cured with the new treatment and vice versa </li></ul></ul><ul><li>Trials to show “noninferiority” </li></ul><ul><ul><li>e.g., test null hypothesis that % cured with active </li></ul></ul><ul><ul><li>Control is ≥ 5% than the % cured with the new treatment </li></ul></ul>
  24. 25. Sample Size <ul><li>Determine based on: </li></ul><ul><li>A primary endpoint </li></ul><ul><li>The null hypothesis </li></ul><ul><li>The test statistic (e.g., t-test, chi-square test, logrank test) </li></ul><ul><li>The treatment difference to be detected (the “alternative hypothesis”) </li></ul><ul><li>Significant level (Type I error) </li></ul><ul><li>Desired power (Type II error) </li></ul><ul><li>Variability assumptions </li></ul><ul><li>The plan for handling treatment withdrawals and protocol violations </li></ul>
  25. 26. Data Monitoring <ul><li>Oversight of Trial Quality </li></ul><ul><li>Monitoring of Treatment Effects </li></ul>
  26. 27. Oversight of Trial Quality <ul><li>Checks performed in a blinded manner: </li></ul><ul><ul><li>Whether the protocol is being followed </li></ul></ul><ul><ul><li>The acceptability of data being accrued </li></ul></ul><ul><ul><li>The success of planned accrual targets </li></ul></ul><ul><ul><li>The appropriateness of the design assumptions </li></ul></ul><ul><ul><li>Success in keeping patients in the trials </li></ul></ul><ul><li>Has no impact on Type I error </li></ul>
  27. 28. Monitoring Treatment Effects: Interim Analysis <ul><li>Usually for serious outcomes </li></ul><ul><li>Requires unblinded access to treatment group summary data </li></ul><ul><li>Should only be done if included in the protocol </li></ul><ul><li>Goal, stop the trial early if: </li></ul><ul><ul><li>Superiority of the new treatment is clear </li></ul></ul><ul><ul><li>Future demonstration of a treatment effect is unlikely </li></ul></ul><ul><ul><li>Unacceptable adverse effects are apparent </li></ul></ul>
  28. 29. Monitoring Treatment Effects: Interim Analysis <ul><li>May require a DMC, which should approve interim plans </li></ul>Independent Statistical Team Study Team IRB Sponsor DMC
  29. 30. Monitoring Treatment Effects: Interim Analysis <ul><li>Repeated testing of outcome data increases the chance of a </li></ul><ul><li>Type I error </li></ul><ul><li>Test for difference in proportions failing in two groups without adjustment for multiple testing </li></ul><ul><li>Decision Rule : Reject null if |Z| ≥1.96 Overall Type I Error Rate </li></ul><ul><li>Single test at end of study -----> 0.05 </li></ul><ul><li>Two tests, equally spaced -----> 0.08 </li></ul><ul><li>Five tests, equally spaced -----> 0.14 </li></ul><ul><li>-(Friedman, Furberg and DeMets, 1996) </li></ul>
  30. 31. Data Analysis <ul><li>Include main features of analysis in protocol </li></ul><ul><li>For confirmatory trial, include statistical methods to be used for the primary variable(s) </li></ul><ul><li>For exploratory trials, include general principles and directions </li></ul><ul><li>Additional ‘statistical analysis plan’ </li></ul><ul><ul><li>Detailed procedures for primary and secondary variables </li></ul></ul><ul><ul><li>Do blind review of data, record date of breaking blind </li></ul></ul>
  31. 32. Analysis Sets <ul><li>Disposition of participants enrolled, summary of protocol violations </li></ul><ul><li>Degree of compliance and missing data lead to different Analysis Sets: </li></ul><ul><ul><li>Full Analysis Set </li></ul></ul><ul><ul><li>Per Protocol Set </li></ul></ul><ul><ul><li>Rationale: </li></ul></ul><ul><ul><ul><li>Minimize bias (Analysis Sets defined a priori) </li></ul></ul></ul><ul><ul><ul><li>Demonstrate lack of sensitivity </li></ul></ul></ul>
  32. 33. Full Analysis Set <ul><li>‘ Full analysis set’ = the analysis set which is as complete as possible and as close as possible to the intention-to-treat ideal of including all randomized subjects, it may exclude, for example: </li></ul><ul><ul><li>Participants who failed to meet a major entry criteria </li></ul></ul><ul><ul><li>Participants who lack any data post randomization </li></ul></ul>
  33. 34. Full Analysis Set <ul><li>If specified in the plan, subjects who fail to meet an entry criterion may be excluded without the possibility of introducing bias under the following circumstances: </li></ul><ul><ul><li>The entry criterion was measured prior to randomization </li></ul></ul><ul><ul><li>The detection of the relevant eligibility violations can be made completely objectively </li></ul></ul><ul><ul><li>All subjects receive equal scrutiny for eligibility violations </li></ul></ul><ul><ul><li>All detected violations of the particular entry criterion are excluded </li></ul></ul>
  34. 35. Per Protocol Set <ul><li>‘ Per protocol set’ = subset of the participants in the full analysis set who are more compliant with the protocol </li></ul><ul><ul><li>Complete a certain pre-specified minimal exposure to the treatment regimen </li></ul></ul><ul><ul><li>Have some minimum number of measurements of the primary variable(s) </li></ul></ul><ul><ul><li>Have no major protocol violations </li></ul></ul><ul><li>May give overly optimistic results in superiority trials </li></ul><ul><li>May be the more conservative analysis set for equivalence or non-inferiority trials </li></ul>
  35. 36. Estimation and Confidence Intervals <ul><li>Not just p-values, include confidence intervals for estimated treatment effects </li></ul><ul><li>Prespecify any covariates to be controlled for in primary or secondary analysis </li></ul><ul><ul><li>To improve precision </li></ul></ul><ul><ul><li>To adjust for potential imbalances </li></ul></ul><ul><ul><li>To account for stratified designs </li></ul></ul><ul><ul><li>Never adjust for post-randomization variables </li></ul></ul><ul><li>Prespecify interactions and subgroups of interest if treatment effect is likely to vary by baseline factors (e.g., gender) </li></ul>
  36. 37. Evaluation of Safety <ul><li>Choice of variables: Laboratory tests, vital signs, adverse events </li></ul><ul><li>Safety Analysis Set: Usually those who received at least one dose of the investigational drug </li></ul><ul><li>Statistical Analysis </li></ul>
  37. 38. Example of an unacceptable adverse event
  38. 39. Evaluation of Safety is critical to the success of a new drug
  39. 40. Reporting <ul><li>Document deviations from analysis plan, when and why they occurred </li></ul><ul><li>Account for all subjects who entered the study </li></ul><ul><li>Describe all reasons for exclusion from analysis dataset and all protocol violations </li></ul><ul><li>Summarize measurements of all important variables </li></ul><ul><li>Consider the effect of loss of subjects, violations and missing data on analysis results </li></ul><ul><li>Describe participants lost, withdrawn, etc. </li></ul>
  40. 41. <ul><li>Thank you! </li></ul><ul><li>Questions? </li></ul>