1. Steven Snapinn JSM July 30, 2018
Symptom Trials Vs Morbidity/Mortality Trials: Are Different Estimands Required?
Steven Snapinn
JSM
July 30, 2018
One imperfect but convenient way of classifying clinical trials is on the basis of the type of primary endpoint being evaluated. Two broad classes of endpoints include symptoms, such as pain relief or glucose control, and morbidity/mortality, such as a major adverse cardiovascular event or overall survival. In the case of symptom trials the occurrence of intercurrent events, such as the use of rescue medication, often calls into question the strict application of the intention-to-treat (ITT) estimand. In the case of morbidity/mortality trials, on the other hand, even though such intercurrent events can occur, tradition often calls for strict ITT. In this presentation I will discuss the rationale for this distinction, and argue that the justification for an ITT estimand is the same in both situations.

2. Outline ICH E9 and the Intention-to-Treat (ITT) Principle
ICH E9 (R1) and Estimands
Estimands for Symptom Trials and Morbidity/Mortality Trials
ICH E9 (R1) Appears to Allow Flexibility in Defining Estimands Other Than ITT/Treatment Policy – However:
ITT Is the Only Acceptable Estimand for Morbidity/Mortality Trials
There Is Little Scientific Rationale for Using Different Estimands for Symptom and Morbidity Mortality Trials

3. ICH E9: Statistical Principles for Clinical Trials
If all subjects randomised into a clinical trial satisfied all entry criteria, followed all trial procedures perfectly with no losses to follow-up, and provided complete data records, then the set of subjects to be included in the analysis would be self-evident. The design and conduct of a trial should aim to approach this ideal as closely as possible….
The protocol should … specify procedures aimed at minimising any anticipated irregularities in study conduct….

4. ICH E9: Statistical Principles for Clinical Trials [2]
… the term 'full analysis set' is used to describe the analysis set which is as complete as possible and as close as possible to the intention-to-treat ideal of including all randomised subjects.
Preservation of the initial randomisation in analysis is important in preventing bias and in providing a secure foundation for statistical tests.
Under many circumstances [use of the full analysis set] may … provide estimates of treatment effects which are more likely to mirror those observed in subsequent practice.

5. ICH E9: Statistical Principles for Clinical Trials [3]
Special problems arise in connection with subjects withdrawn from treatment … who provide no data after this point…. Measurements of primary variables …subsequently collected in accordance with the intended schedule of assessments in the protocol, are valuable in this context; subsequent collection is especially important in studies where the primary variable is mortality or serious morbidity.

6. What Is ITT? ITT Involves Both Trial Conduct and Analysis
Conduct: All Randomized Subjects Are Followed and Have Endpoints Measured, Regardless of Compliance With Protocol
Analysis: All Data From All Subjects Are Analyzed as Part of the Assigned Treatment Group, Regardless of the Treatment Actually Received

7. What Question Does ITT Address?
ITT Focuses on a Treatment Policy, Or Treatment Decision, Not Specifically on a Treatment Effect
Question: Will the Decision to Begin an Intervention Ultimately Benefit the Patient?
While Not the Only Potential Question of Interest, This Question Is Always Relevant And Often of Primary Importance

8. Benefits and Limitations of ITT
Preserves Randomization
Maintains Balance in Prognostic Factors
Allows Valid Inference From Statistical Tests
Provides Unbiased Answer to Question of Clinical Effectiveness
Limitations
Conservative Estimate of Treatment Effect
May Not Be Appropriate for Non-Inferiority Trials or Safety Analyses
Refs: Lachin, CCT 2000; Heritier et al, MJA 2003; Sainani, Am Acad of Phys. Med & Rehab 2010

9. Some Opinions in the Literature on Handling Missing Data
“… by far the best course is to avoid the problem to the extent possible.”
O’Neill & Temple, Clinical Pharmacology & Therapeutics, 2012
“… the best method to handle non-ignorable data is to prevent it.”
Hardy et al, J Am Geriatr Soc., 2009
“The best way to deal with missing data is to avoid it.”
Sainani, American Academy of Physical Medicine and Rehabilitation, 2010
“… the preferred and often only satisfactory approach to addressing missing data is to prevent it.”
Fleming, Annals of Internal Medicine, 2011
“Of course, the best way to handle missing data is to avoid it …”
Siddique et al, Psychiatr Ann, 2008
The only disagreement is between whether we want to avoid missing data or prevent it.

10. National Research Council Report on Missing Data
Recommendation 3: Trial sponsors should continue to collect information on key outcomes on participants who discontinue their protocol-specified intervention in the course of the study, except in those cases for which a compelling cost-benefit analysis argues otherwise, and this information should be recorded and used in the analysis.

11. ICH E9 (R1) and ITT Treatment policy strategy
The occurrence of the intercurrent event is irrelevant: the value for the variable of interest is used regardless of whether or not the intercurrent event occurs.
For example, when specifying how to account for rescue medication as an intercurrent event, occurrence of the intercurrent event is ignored and the observations on the variable of interest are used. If applied across all types of intercurrent events, this reflects the comparison described in the ICH E9 Glossary (under Intention to Treat Principle) as the effect of a treatment policy.
“However, the question remains whether understanding the effect of a treatment policy always targets the treatment effect of greatest relevance to regulatory and clinical decision making.”

12. ICH E9 (R1) and Other Estimands
Composite strategy
The occurrence of the intercurrent event is taken to be a component of the variable, i.e. the intercurrent event is integrated with one or more other measures of clinical outcome as the variable of interest.
For example, the requirement to use a rescue medication may provide meaningful information on the effect of a treatment and hence may be incorporated into a variable, with appropriate summary measure, that describes a meaningful treatment effect.
Hypothetical strategy
A scenario is envisaged in which the intercurrent event would not occur: the value to reflect that scientific question of interest is that which the variable would have taken in the hypothetical scenario defined.
For example, when rescue medication must be made available for ethical reasons, a treatment effect of interest might concern the outcomes if rescue medication had not been available.
Principal stratum strategy
The target population might be taken to be the principal stratum in which an intercurrent event would not occur. For example, the target population of interest might be taken to be the stratum of patients in which failure to adhere to treatment would not occur.
While on treatment strategy
Response to treatment prior to the occurrence of the intercurrent event is of interest. If a variable is measured repeatedly, its values up to the time of the intercurrent event may be considered to account for the intercurrent event, rather than the value at the same fixed timepoint for all subjects.
For example, subjects might discontinue treatment, and in some circumstances it will be of interest to assess the risk of an adverse drug reaction during the period of adherence.

13. ICH E9 (R1) Generic Example
A new investigational treatment (Drug X) is considered for subjects with a specific chronic, non-life-threatening disease.
Response to treatment is monitored monthly using a continuous measurement. The full effect of Drug X is expected to be seen at four to six months after treatment start.
The main scientific question concerns the comparison of Drug X to placebo at month 6, and is best addressed by a randomised clinical trial.
Use of placebo in the clinical trial is considered ethical but only if provision is made for subjects to discontinue their treatment and switch to rescue medication due to lack of efficacy. Switch to rescue medication is an intercurrent event, after which it is still possible to collect the variable measurements.

14. Support for ITT: Summary
ICH E9 > NRC Report > ICH E9 (R1)
Strong Support for ITT for Morbidity/Mortality Trials, Less Strong Support for Symptom Trials
In My Experience, Many Clinical Trialists Will Not Consider Anything Other Than Strict ITT for Morbidity/Mortality Trials
Question: What Is the Rationale for Using Different Estimands for Morbidity/Mortality Trials and Symptom Trials?

15. Symptom Trials vs Morbidity/Mortality Trials
“Recommendation 3 … is potentially more controversial.”
“Although such an ITT approach is reasonably standard and broadly accepted for outcome studies, it is not usual—and not clearly desirable—in studies of symptoms.”
This Raises Key Issue: Is There A Fundamental Difference Between Outcome Studies and Studies of Symptoms?
Ref: Temple & O’Neill; Clinical Pharmacology & Therapeutics 2012

16. Symptom Trials vs Morbidity/Mortality Trials [2]
“… consider an outcome study … in which the test drug exacerbated heart failure. In these circumstances, subjects with heart failure … would be more likely to leave the test-drug group [which] would lower the mortality risk in the test-drug group…. The ITT approach is intended to protect against this kind of ‘informative censoring’….”
“In [symptomatic benefit] trials, the missing-data problem is of a different kind. Early dropouts can leave treatment groups unbalanced with respect to important prognostic [factors]. The effect of these dropouts on outcome could go in either direction….”
Ref: Temple & O’Neill; Clinical Pharmacology & Therapeutics 2012

17. Symptom Trials vs Morbidity/Mortality Trials [3]
“In symptomatic settings, it is not the usual practice to continue to assess effectiveness in patients after they have stopped taking the assigned treatment (ITT approach)….”
“Although such an ITT approach is reasonably standard and broadly accepted for outcome studies, it is not usual—and not clearly desirable—in studies of symptoms. In this category of studies, it is strongly expected that there could be a prompt loss of a drug’s effect after it is discontinued; therefore, the inclusion of data from subjects who have discontinued the drug would introduce a bias against the drug when ITT analysis is carried out. Also, it is the common practice for dropout patients to switch therapies; this would further confound the results of the analysis.”
Ref: Temple & O’Neill; Clinical Pharmacology & Therapeutics 2012

18. Intention-to-Treat Summary
ITT Involves Both Trial Conduct and Analysis
Addresses a Clinically Relevant Question, Ensures Balance in Prognostic Factors and Allows Valid Statistical Inference
Endorsed By ICH
Appears to Be Endorsed by NRC Report
FDA and ICH Distinguish Between Outcome Trials and Symptom Trials
No Clear Endorsement for ITT in Symptom Trials
Rationale for This Distinction Is Suspect

19. Hypothetical Example: Treatment Strategy Vs Principal Stratum Estimand
Treatment X Being Developed to Treat Diabetes
Primary Endpoint: Reduction in Hba1C at 6 Months
Principal Stratum Defined as Those Patients Who Tolerate Both Treatment X and Control
Suppose There Is Benefit in the Principal Stratum, But Not in the Entire Population
ie, Adverse Outcome Outside the Principal Stratum Cancels the Benefit Inside the Principal Stratum

20. Symptom Trial vs Morbidity/Mortality Trial
Benefit With Principal Stratification Estimand
No Benefit With Treatment Strategy Estimand
Morbidity/Mortality Trial
Benefit wrt HBa1C Leads to Morbidity/Mortality Benefit With Principal Stratification Estimand
Lack of Benefit wrt HBa1C Lead to Lack of Morbidity/Mortality Benefit With Treatment Strategy Estimand

21. Summary
Traditional in Morbidity/Mortality Trials to Require Strict Adherence to ITT Principle
Guidance Documents, Particularly ICH E9 (R1), Allow Greater Flexibility for Symptom Trials
Rationale for This Distinction Is Unclear