Clinical Trials

Question 1

Define clinical trials

Answer 1

• An experiment that attempts to answer a medical question.
• The experiment is conducted under conditions determined in advance, using a specific methodology to test a hypothesis.
• Usually, the question is non-trivial and is addressed in an unambiguous, reliable and easily interpreted manner.*

Question 2

What are goals of clinical trials

Answer 2

• To identify promising new treatments
• To determine if such treatments really reduce mortality and morbidity
• Statistically, to obtain an unbiased estimate of the effect of treatment for the groups being studied
• To extrapolate the results to the general population

Question 3

What are the phases of a clinical trial

Answer 3

1. Determine the “Safe Dose”
   
   1. Increase dose in subject groups
   2. Stop escalating if too toxic
   3. Carefully monitor all toxicity
2. Pharmacokinetics
3. Efficacy
4. Help define experimental endpoints
5. Phase 2
   
   1. Objective: estimate activity of an intervention; decide whether the treatment should be studied in large-scale comparative trials.
   2. Key role in the drug development process because results determine whether or not to proceed to phase III trials.
   3. Usually a single cohort (arm) gets exposed to the same intervention (same fixed dose)
   4. Dichotomous endpoints are the norm
6. Phase 3
   
   1. An experiment planned by investigator, who assigns treatment (does not occur “naturally” as in an observational study)
   2. Compare intervention (new treatment, program or management procedure) with another group (e.g., standard therapy, procedure or placebo) to assess efficacy of treatment in humans, animals or herds
   3. Study groups are comparable, NOT necessarily representative
   4. Participants followed for a defined outcome
7. Phase 4
   
   1. Look for adverse events
   2. Post-marketing sureveillance

Question 4

Explain treatment assignment in a RTC

Answer 4

• Random assignment (or allocation) implies each individual has equal chance of receiving each possible treatment, independent of another individual receiving treatment
• Validity of trial depends on randomization achieving similarity of treatment groups on measured and unmeasured factors at baseline

Question 5

Explain creation of treatment groups and control groups in clinical trial

Answer 5

• Random assignment vs. random selection
  
  • Internal validity and external validity
• Advantages of random assignment:
  
  • Tends to produce comparable groups for known and unknown factors
    (use baseline data to assess)
  • Removes potential for allocation bias (investigator or participant choice)

Question 6

Alternatives to Random Assignment to Treatment and Control Groups

Answer 6

• Non-randomized, concurrent controls; e.g.:
  
  • Participants at different institutions (hospitals, clinics, etc.) receive different treatments
  • Even and odd number assignment or other systematic approach
  • Groups may not be comparable
  • May never have all information on important prognostic factors
• Historical controls
  
  • Non-randomized, non-concurrent.
  • Assume observed change is due to new therapy
    
    • Groups may not be comparable, many factors change over time
    • Change may be attributable to patient population shifts, management changes, not the intervention
    • Problems with accuracy, completeness, consistency of retrospective data
  • Advantages: Economical, more ethical

Question 7

What are good allocation practices?

Answer 7

• Mask assignment to patients, physicians and all others until needed
• Make sure future assignments can’t be predicted:
  
  • Birth dates, SSN, odd-even or systematic schemes
• Use method where the order of allocations is entirely reproducible:
  
  • Not coin tosses
  • Computer-generated assignment (with fixed seed)

Question 8

What is a factorial design

Answer 8

• Goal is to answer two separate research questions in a single cohort
• Example: Physician’s Heart Study
  
  • H1: Effect of aspirin on MI
  • H2: Effect of β-carotene on cancer
• Randomly assigned to four groups:
  
  • Treatment (Aspirin) and Treatment (β-carotene)
  • Treatment (Aspirin) and Placebo (β-carotene)
• 3) Placebo (Aspirin) and Treatment (β-carotene)
• 4) Placebo (Aspirin) and Placebo (β-carotene)

Question 9

How is a factorial design designed?

Answer 9

• Randomly assigned to 4 groups
• Each of the two hypotheses is tested by comparing
  two halves of the study cohort:
• Analysis 1: All on Treatment (Aspirin) are compared to Placebo (Aspirin)
  
  • Disregards that half of each group received b-carotene
• Analysis 2: All on Treatment (b-carotene) are compared to Placebo (b-carotene)
  
  • Disregards that half of each group received Aspirin
• Very efficient study design
  
  • Two trials for the price of one!
• Chief limitation: possibility of interactions between the treatment and outcome
  
  • What if b-carotene had a significant impact on MI?
    
    • Reduces statistical power
    • Makes interpretation more complex
• Best reserved for two unrelated research questions
• Factorial designs can be used to study interactions (with increased sample size)

Question 10

what is a cross-over design?

Answer 10

• Half of the subjects are randomly assigned to start off with the control and then switch to active treatment
• The other half start with treatment and switch to control
• Permits both between-groups and within-groups analysis
• Most optimally employed when:
  
  • Number of study subjects is very limited
  • Carryover effects are not an issue

Question 11

What are the advantages and disadvantages of a cross-over design?

Answer 11

Advantages

• All subjects serve as own controls and error variance is reduced thus reducing sample size needed
• All subjects receive treatment (at least some of the time)
• Statistical tests assuming randomization can be used
• Blinding can be maintained

Disadvantages

• Doubles (at least) duration of the study
• Major possibility of carryover effect
  
  • Residual influence of treatment on the outcome after it has been stopped
• May require a washout period
  
  • Could be lengthy or, worse, unknown
• Cannot be used for treatments with permanent effects (e.g., surgery)

Question 12

what is stratified randomization in controlled trials

Answer 12

• Based on data collected prior to randomization
  
  • Post-stratification is a process used in data analyses, not in the execution of the trial
• Stratified randomization is used to reduce or eliminate variation in outcome measure due to the pre-specified stratification variable(s)
• Practical limit to number of variables stratified
• Limit to variables that are not likely to have large recording errors:
  
  • e.g., randomize within clinic in multi-site trials à comparable groups even if large population differences by clinic
• The more restrictive the patient selection criteria, the less need for stratification (also the less generalizable)
• Stratification can complicate randomization process because assignments cannot be made until all data needed for stratification are obtained
  
  • e.g., minimum delay if gender is the stratification variable, a lot more delay if the stratification variable is a laboratory-generated value