Clinical Trials

Medical Epidemiology Clinical Trials

A clinical trial is
* A cohort study
* A prospective study
* An interventional study
* An experiment
* A controlled study

The Structure of a Clinical Trial
Various Aspects Are Standardized and Protocol-based
* Subject selection (who are these people?!)
* Subject assignment
* H & P data
* Therapeutic intervention
* Lab calibration
* Outcome evaluation

Subject Selection
* Adequate number of subjects
* Adequate number of expected endpoints
* Easy to follow-up
* Willing to participate (give consent)
* Eligibility (criteria)
* Efficacy Versus Effectiveness
* Internal Validity (validity) versus External Validity (generalizability)

Phases
* Phase I: find toxic dose
* Phase II: no controls
* Phase III: RCT
* Phase IV: Post marketing?

Types of Control Groups
* Historical
* Contemporaneous
* Concurrent
* Randomized

Allocating Treatment
* Complete (Simple) randomization
* Restricted randomization

Complete Randomization
* Patients assigned by Identical chance process (but not necessarily in equal numbers)
* Mechanics
* Insures process fairness
* Does not insure balance, especially in small studies.Therefore, may still need statistical adjustment

Randomization
* The only way to deal with unknown confounders.

Philosophy of Randomization
* Why are randomized trials not “epidemiologic” studies?
* Why randomization is so special?
* Has nothing to do with sampling bias.
* Randomization (random allocation) versus random sample.
* Does NOT deal with “chance” as a possible explanation of the difference. To the contrary.
* Can be used to create groups of unequal size.
* Baseline characteristics (table 1).

Allocation Concealment
* Define.
* Why do we need it?
* How is it done?
* Buzz words
* Versus blinding

Buzz Words
* Central (phone) randomization
* Sequentially numbered, opaque, sealed envelopes
* Sealed envelopes from a closed bag
* Numbered or coded bottles or containers

Restricted Randomization
* Stratification
* Blocking (Permuted Block Design)
* Stratified Blocking

Stratified Randomization
* Why

Scheme of stratified randomization
Blocking
* Why?
* Ensures close balance of the numbers in each group at all times during trial.
* How is it done?
* More importantly when stratified.
* Problem If block size is discovered.
* Remedy: more blinding, varying block size, larger blocks.
* Basic, Randomized (random-sized), Stratified

Problems With Concurrent Controls

Use your imagination
Examples
Problems With Contemporaneous Controls
* Regional population differences.
* Regional practice differences.
* Diagnostic variations.
* Referral pattern biases.
* Variations in data collections.

Problems With Historical Controls
* A lot more

Why Do Controls in a Randomized Trial Do So Well ?!
* Volunteerism
* Eligibility
* Placebo effect
* Hawthorne effect
* Regression towards the mean

Placebo Effect
* Placebo can do just about anything (prolong life, cure cancer).
* Improve athletic performance
* Lower T4 count
* Placebo can do just about anything (prolong life, cure cancer).
* Placebo can also cause side effects (nocebo, Wile E Coyote effect).
* Placebo effect is very useful in medicine but in epidemiology it causes problems, so we try to equalize it between the 2 groups.
* We use placebo for other benefits.

Hawthorne Effect
* Hawthorne works of the Western Electric Co. Chicago, IL

Regression Towards the Mean
Course Evaluation Question
Explains difficult material:
* Strongly agree
* Agree
* Neutral
* Disagree
* Strongly disagree
* What difficult material ?

Regression Towards the Mean
Explains difficult material
ATTENTION
CAUTION
DIFFICULT MATERIAL AHEAD!

Regression Towards the Mean

* Example
* Individuals with initially abnormal results tend on average to have more normal (closer to the mean) results later.
* Lab tests, BP etc.
* Recheck before randomization. Run-in period.
* Sophomore slump, medical school, Airforce landing feedback, most dangerous intersections.

Quicken Loans
* Rates this low seldom stay around long and tend to go up quickly and without warning

Why Does Prognosis Improve Over Time ?
1. Initial reports come from referral centers..
2. Publicity
3. Physicians’ awareness
4. Development of a Diagnostic Test
* Allows diagnosis of atypical cases.
* Is an incentive for physicians. It’s more challenging to diagnose difficult cases.
* Physicians with zero diagnostic skill can now diagnose this disease.
* Allows diagnosis of non-cases.
* Allows population based studies.
5. Publicity that a disease is very common relieves clinician from worrying that they may be overdiagnosing it.
6. Placebo effect increases over time. Why?

7. Safer treatment (laparoscopic cholecystectomy) lowers the threshold for doing surgery. So patients having surgery are not as sick as before.

Stage Migration Bias
Will Rogers Effect
8.Improved staging tests cause an apparent improvement of prognosis in every stage.
Stage Migration Bias Will Rogers Effect

“BEFORE-AFTER” STUDY
Mortality by severity level
Severity distribution
Mortality by severity level
Will Rogers Effect
* When subjects with the most severe disease in each stratum are moved to the next (more severe) stratum - for whatever reason - this will cause an apparent improvement of prognosis in every stage.
* Common.
Exclusions
Exclusion Criteria
* Excluded patients are “ineligible”. So, Why the separate category?
* More informative
* Usually very large number.
* Usually underestimates. Real number even bigger. Why?

Exclusion Criteria
What to watch for
* Patient preference
* Clinician preference
* No reason given

Drop ins and Drop outs
* Define.
* Typical case
* Other

Subjects who drop out of study or change treatment. But available for outcome assessment.
* Intention to treat analysis
* Once randomized always analyzed
* Why ?

1. Change in therapy may be related to outcome or eligibility
2. To get the full benefit of randomization
3. Effectiveness versus efficacy
Five-Year Mortality in Coronary Drug Project
A COHORT STUDY OF RECURRENT MI BY PARTICIPATION IN
A GRADUATED EXERCISE PROGRAM FOLLOWING INTITIAL MI

RECURRENT MI
YES NO TOTAL
PARTICIPATION IN GRADUATED EXERCISE PROGRAM

A RANDOMIZED CLINICAL TRIAL OF ENDURANCE TRAINING FOR PREVENTION OF RECURRENT MI
More on “Intention to Treat”
* Always analyze the results of the subjects according to the group they were randomized to. No exclusions.
* Even if they received no intervention.
* Even if they didn’t have the disease (example).
* The philosophy of “Intention to Treat” analysis
* Addresses the ultimate (and only) question for the clinician: Does prescribing treatment make a difference?
* LDL targets?!

Alternatives to “Intention to Treat” Analysis. (PROBLEMATIC)
* “Per Protocol” analysis.
How is it done?
Problems.
* “As Treated” analysis.
How?
Problems.

“Per Protocol” analysis.
* Censoring data after subjects become non-adherent
* Preserves randomization
* Stops counting events (when? “Carry-over” effect)
* Reduced power

“As Treated” analysis
* Change the treatment arm of the subject as he/she changes exposure
* The follow-up time and the events will be assigned to current exposure
* Retain all events.
* Randomization violated.
* Have to assign “lag-time” (latency) and Carry-over time.

Loss to follow up
Differential vs. Random
+ Compare their baseline variables with the rest of the subjects.
+ Chase a subgroup.
+ Worst case scenario

Objectives of Subgroup Analysis
* Support the main finding
* Check the consistency of main finding
* Address specific concerns re efficacy or safety in specific subgroup

It may also generate hypotheses for future studies. But that is not a reason to do it.

Inappropriate Uses of Subgroup Analysis
* Rescue a negative trial
* Rescue a harmful trial
* Data dredging: find interesting results without a prespecified plan or hypothesis

To Avoid Inappropriate Uses of Subgroup Analysis
* Prespecify analysis plan.
* Prespecify hypotheses to be tested based on prior evidence.
* Plan adequate power in the subgroups
* Avoid the previous pitfalls.

Problems with Subgroup Analysis
* Low power
* Multiplicity
* Test for interaction
* Comparability of the treatment groups maybe compromised
* Over-interpretation

Blinding
* PATIENTS Single blind.
* CLINICAL THERAPISTS usually double blind.
* Double Dummy
* CLINICAL EVALUATORS. Have to specify.
* Subjective vs. objective assessment

DSMB
* Data and Safety Monitoring Board
* Have duty toward:
* Current study participants (ongoing treatment)
* Future participants.
* Enough evidence to change practice
* Enough evidence to withstand criticism. (Unable to randomize afterward).

Multiple looks
* Alpha spending
DSMB
* Data Safety Monitoring Board
* Early Termination rules
* O’Brien-Fleming
* Early vs. late
* Benefit vs. harm (blinding?)
* Multiplicity
* Rules. Scenarios.
CLINICAL TRIALS JARGON
* Consecutive patients (versus a random sample)
* Baseline characteristics of patients (to see if randomization worked)
* Number of subjects and average duration of follow-up
(versus patient years)
* Interim analysis, problems
* Cumulative incidence (versus incidence density)
* Relative risk (Odds Ratio, or Hazard Ratio) (hopefully <1)is:
rate of outcome in a drug group
rate of outcome in a placebo group
* Relative risk reduction (similar to attributable risk %). But here it is 1-RR.
* Absolute difference in risk (ADR)= risk in control group – risk in intervention group (similar to AR) very important, sometimes not reported
* Relative risk reduction versus absolute difference in risk
* Number needed to treat NNT =1/ADR very useful (remember time)

Descriptions of “Trials”
* 34% relative decrease in the incidence of MI. The decrease is statistically significant. The 95% CI ranges from 55% relative decrease to a 9% relative decrease.

* 1.4% decrease in …. (2.5% versus 3.9%). The decrease is statistically significant. The 95% confidence interval ranges from a 2.5% decrease to a ..

* 77 persons must be treated for an average of just over 5 years to prevent 1 MI.

Ratings of Trials
(-5=harmful,+5=very effective)

Medical Epidemiology Clinical Trials.ppt