You Are Here: Home > Cancer Research > How Cancer is Studied > Clinical Research >
Clinical Trial Background Issues - pg. 3

Clinical Trial Background Issues - page 3

Clinical Trial Design

Randomized Controlled Trials (RCTs)

RCTs have become the gold standard for clinical research. To establish causality between the intervention (e.g., a drug) and the outcome (e. g., overall survival), researchers try to make sure that one can assume that the experimental and control arms are similar in every way except for the intervention.

This is sometimes referred to as balancing the groups, and ensures that no superfluous variables are confounding, (produce results that seem to be caused by the intervention when they really aren't). Three techniques to avoid confounding are discussed below.

Randomization is a method of assigning patients to treatment arms by chance, avoiding any systematic imbalance in characteristics between patients who will receive the experimental intervention versus those that receive the control intervention.

 
 

Patients are usually assigned equally to all arms, although this need not be the case.

With a simple two-arm trial (one experimental and one control arm) randomization can be accomplished with a flip of a coin.

When there are more than two arms, or unequal numbers of patients are to be assigned to different arms, computer algorithms can be used to ensure random assignment.

Blinding

Blinding is a means of ensuring that neither patients, healthcare providers, nor researchers know which group specific patients are assigned.

Trials are said to be single (the patient does not know), double (the patient and the investigator do not know), or triple blinded (the patient, investigator and assessors do not know if participant is receiving the new intervention), depending upon how many of the relevant participants in the trial are unaware of patient assignment.

The purpose of blinding is to minimize the chance that patients receive different care, or their data is interpreted differently, based upon the intervention to which they are assigned.


 
Personalized Medicine
 
Recently Diagnosed
 
Cancer 101
 
Survivorship
 
Research
Advocacy
 
Inspiration
 
Resources
 
Webinars
 
e-Training
 
Latina Navigator Training
 
 
 
 
 
 
 
Stratification

Stratification prior to randomization can be used to ensure that the number of patients assigned to the experimental and control arms are balanced with respect to important attributes (stratification variables).

 
 

Examples of stratification variables are gender or disease stage. The purposes of stratification are two-fold.

  • First, stratification ensures that the stratification variable is not confounded with the intervention, which is especially important when the stratification variable is known to have a large impact on the outcome of interest. In large trials randomization alone typically achieves balance, and stratification may be unnecessary.
  • Second, if adequately planned, stratification allows sub-group analysis, essentially looking at each stratum separately. It is not uncommon, however, for subgroup comparisons to be conducted even when not adequately planned, which can lead to erroneous conclusions.

 

Statistical Inference

Making statistical inference allows researchers to draw conclusions applying to populations of future patients using the limited sample of patients in their trials.

The traditional approach:

  • Determine whether trial results are evidence of a true difference between experimental and control intervention, or a chance occurrence.
  • Errors will be made, but over the long run, the proportion of errors is controlled.

 

Key terms:

  • P-value: is the likelihood the result is due to chance
    • <0.05 or 1 in 20 is the standard value used
  • Confidence interval: used to indicate the reliability of an estimate
  • Sample size: the number of people in the trial
  • Relative risk: the likelihood that cancer will occur within a specific timeframe in one group vs. another

 

 
 
 
1 2 3 4
 
 
 
Copyright © 2006-2017 CISN - All Rights Reserved.
 
Site Design by: Cara M. Caloroso
 
CISN Home Page About Us Services CISN Home Page Contact Site Map CISN Home Page CISN Home Page