G.2 Randomised clinical trials Flashcards
Clinical trials
A clinical trial is defined as a prospective study comparing the effect and value of interventions (s) against control in human subjects
Name some interventions
Therapeutic agent Prophylactic agent diagnosis agent Devices Surgical procedures Health service strategies or activity
What would be a control?
No intervention
Placebo
Best current standard treatment
What are the different phases of clinical trials?
Phase I: Healthy volunteers
Phase II: Small sample of patients
Phase III: Extended sample of patients
Phase IV: Large cohort of patients
What is a double-blind randomised controlled trial
A type of cohort study
Intervention for each participant is assigned randomly by the investigator
Intervention unknown both by patient and investigators (judges outcome)
How is a clearly focused research question established ?
Established using PICO P: population/patient/problem I: intervention C: Comparison/ Control O: Outcome
How would you undergo a selection of study cohort?
Define inclusion criteria appropriate for research question
Define exclusion criteria that helps control errors
Design and adequate sample size and plan that recruitment accordingly
How is sample size determined and what are risks associated with having too many or too little patients?
Determined by:
- Statistical power
- p-value
- Difference magnitude
Too little: may not be able to answer the question central to the trial
Too large: some patients may be unnecessarily exposed to an inferior treatment
What are the types of error ?
Type I error: When you conclude that there was a difference in your group when in fact there was none
a(alpha)= probability that we make such an error (typically 0.05 or 5%
Type II error= When you conclude that there is no error but in fact there is one.
b (beta)= probability that we make those error (probability of type II error)
What is power
The probability of concluding there was a difference when there is one
we want to maximise this value in out study and increasing the number of participants will increase this power
power= 1- beta
Overall power calculation depends on alpha, treatment effect, the distribution of treatment effect under the null hypothesis and the alternative hypothesis
What is the purpose of random allocation?
To eliminate selection biases (by patient or clinicians)
To ensure that each patient has an equal probability f receiving either treatment
To balance treatment and control groups with respect to prognostic factors identified
What are the two commonly used methods for treatment allocations?
Simple randomisation Restricted randomisation (block randomisation, stratified randomisaton)
What is simple randomisation?
equivalent to tossing a coin :
- random numbers
- random number tables
- calculator
- computer software
advantages:
simple, not predictable (slightly predicable)
disadvantage:
Unequal numbers in each group, particularly for small t trials
Imbalance in numbers leads to loss of power
What is block randomisation?
Choose block length L which must be a multiple of the number of treatment N, to ensure equal number of participants allocated to n treatment
What is Stratified randomisation?
Select a variable that is considered as a prognostic factor
- continuous variables need to be grouped e.g. age bands to create strata
- construct separate randomisation lists for each strata using block randomisation
- stratification may be extended to two or more factors but number of separate randomisation list rapidly becomes very large
