week 10 Flashcards
What are clinical trials?
- Studies of human
2. Prospective research studies - follow human studies prospectively/forward over time
What does clinical trial involve?
Comparing different interventions (treatments)
Definition of a clinical trial
Any research study that prospectively assigns human participant or groups of humans to one or more health-related interventions to evaluate the effects on health outcomes
What does clinical trial employ?
Measures designed to eliminate any possible source of bias
Very strong providers of evidence for the effectiveness of treatment
Focus of a very high degree ethical scrutiny - have to conduct trials ethically
What are examples of possible source of bias?
- Blinding
- Randomising
- Accounting for all participants
Who was the first clinical trial in history conducted by?
- James Lind
- 1753
- Looking for the cure for scurvy which was a disease that affected sailors particularly
What is Scurvy caused by?
Vitamin C deficiency
Didn’t know that at the time
James Lind’s Treatise of the Scurvy
- Took 12 sailors and divided them into 6 groups with 2 sailors each
- Each pair of sailors got a different pair of treatments that might be useful in treating Scurvy
What were the different types of treatments?
- 1 quart of cider
- 25 drops of sulphuric acid
- 6 spoonfuls of vinegar
- 1/2 pint of seawater
- 2 oranges and one lemon
- spicy paste and barley water
Which sailors were cured?
One that were given 2 oranges and one lemon
What was the net result of this study?
Sailors in the British Navy was given limes to prevent scurvy
How long did it take to see control groups?
150 years
- concurrent in the study with the active intervention participants
Johannes Fibiger, 1898
Diptheria patients were allocated to treatment (serum injections) or control (routine care) on alternating days
Michigan tuberculosis trial, 1926
Concurrent controls were used because of the ‘‘notorious tendency of the disease to fluctuate naturally’’
What was the control data people used?
- Histological data from previous period of time where people weren’t receiving active intervention
How long did it take for the first modern day clinical trial?
- Another 150 years
- Conducted in 1948 by MRC
- Looking at Streptomycin for treatment of tuberculosis
What was the design of MRC study?
- Done on Austin Bradford
- Randomise the allocation of participants to the intervention or control by shaking a dice
- Allocation completely randomly
What does a randomised controlled trial allocate?
Eligible participants at random either to the active intervention or to a control
What is randomisation used for?
Determine the allocation of participants of active intervention or the controls
What is the idea of randomisation?
- Eligible participants arriving and agrees to take part in the study
- Either allocate them to active treatment (A) or to control (B)
- Have a random element of that decision process
What is simple randomization?
- Toss a coin for each new participant
- Get a computer to make a random selection between letter A/B
- Random sequence of letters A/B down the page representing success of allocations of successive participants in the trial
What is the practical difficulty of simple randomization?
- by chance you might end up with different numbers of people at end of trial allocated to A compared with number allocated to B
What is block randomization?
- Divide sucessive allocation into blocks of even size
- 6 allocations to each box
- 3 out of every 6 block participant are allocated to A and the other 3 are allocated to b
- Which 3 our of 6 participant A’s and which 3 out of 6 B’s are determined randomly
What is Stratified Block Randomization?
- Divide participants into a number of different strata according to characteristics of patients
e. g. divide patients into 2 strata according to age - Each distinct strata - do block randomization
What is the objective of stratified randomization?
Ensure balance of treatment group with respect to various combination of the prognostic variables
Why should the number of strata used be to a minimum?
For good effect
What is randomization by minimization?
- Common alternative to stratification
2. Achieve a balance in a lot of patient’s characteristics
Randomization by Minimisation
Allocates each new participant to the trial arm which will lead to the minimum imbalance on a number of characteristics
Usually also includes a random element
Generally needs specialist software
What does Randomization by Minimisation do?
Divides opinion
Why randomise?
- Random allocation ensures treatment and control groups both represent the same population
- Helps to ensure allocation concealment - concealing future allocations from staff who are recruiting; if they cannot predict the next allocation, it cannot influence who they choose as the next participant
Why randomise (2)?
- Both randomly drawn from the same wider population of individuals
- Both represent the same group of patients
- No systematic difference between people allocated to intervention/control
Historically, how was allocation distributed?
- to trial site in a series of numbered, sealed envelope to ensure concealment
What will modern trial more likely to use?
- internet or telephone based randomisation services
What is related to allocation concealment but is slightly different?
Blinding
What does blinding or masking refer to?
Steps taken to conceal allocations after they have been made
What does allocation concealment refer to?
Stopping people knowing before the allocation - what is it going to be
Who could you blind?
- Patients/participants
- Those involved in patient care
- Outcome assessors
- Those who analyse the data
- Other groups
What single, double, triple blind refer to?
How many different groups of people you are keeping blinded
Single, double, or triple blind?
These terms are ambiguous, and are now discouraged
Always specify exactly which groups were blinded
What is one way to blind?
Use a control which looks, tastes or feels exactly the same as the active treatment
What helps to eliminate source of bias in a trial?
- Randomisation
- Allocation concealment (before allocating)
- Blinding (after allocating)
Phase I of a trial
- First-in-human studies
- Small number of participants
- Dose in relation to side-effects
Phase II of a trial
- Patients with the condition
- Dose in relation to benefit
- May be randomised but more explanatory than phase 3
Phase III of a trial
- Large enough to show evidence of efficacy/effectiveness
2. Usually randomised
What does explanatory look at ?
Efficacy of a drug
What does pragmatic conditions look at?
Effectiveness
Can the intervention work under ideal conditions?
Efficacy
Will the intervention work when delivered in practice?
Effectiveness
Features of Explanatory?
- High risk or highly engaged participants
- Tightly defined procedures
- Outcomes may be surrogate
- Control may be placebo
- Analysis may focus on compliers
Features of Pragmatic?
- Everyone with the condition participates
- Procedures reflect practice
- Outcomes are meaningful to patients
- Control may be routine care
- Analysis is by intention to treat
What can clinical trials be done on?
Educational interventions
Pragmatic
- All patients with moderate parkinson’s disease
- Participants self-injected
- Outcome was a patient-reported quality of life measure
- Analysis by intention to treat
-3.5 points
Estimate the effect of exenadtide compared with placebo (a value of 0 would indicate no effect)
Confidence interval
A range of plausible values for the true effect
P value
A measure of the statistical significance
The strength of the evidence for an effect of exenatide
low p = strong evidence
A result is described as statistically significant when
- The p value is low (usually taken to be p <0.05)
2. The 95% confidence interval rules out zero as a possible effect (for a difference measure of effect)
A finding may be?
Statistically significant but not clinically significant
Narrow confidence interval
Some trials will pin the effect down quite precisely
Wide confidence interval
Others may be more inconclusive
What does investigators have ethical responsibility to plan?
The number of participants in a trial
- not too many (people exposed unnecessarily to medical research)
- Not too few (too little data to produce conclusive results)
What does power depend on?
How big the true effect is
What is a trial designed to have?
Certain level of statistical power (the chance that the trial will produce a statistically significant finding when the data are analysed)
How can you generally increase the power?
Having a larger sample size
How to detect tiny effects?
Do bigger trials
A sample size of 60 patients
Required to detect a difference of 5.8 MDS-UPDRS points between 2 groups based om a common SD of 13
What do most report of trials follow?
Consort statement
What does the CONSORT statement recommend?
A flow chart showing numbers of participants (by trial arm where appropriate) at the end of:
- Enrolment
- Treatment allocation
- Follow up
- Analysis
What is CONSORT statement?
A set of recommendations for reporting clinical trials
Intention to treat
- Analyse all
2. According to the treatment that was allocated rather than received
Per protocol
Analyse only those who completed the treatment they were allocated
Casual analysis
Estimate the effect of treatment if only given to compliers
What is ethical framework for trials?
- These emphasise the protection of human rights of participants in medical research - in particular that they should give informed consent
Ethical framework for trials
- 1947 Nuremburg Code
- 1964 - Declaration of Helsinki
- 1997 - ICH - Good clinical practice
- 2001 - EU directive on clinical trials
- 2004 - UK legislation - Medicine for human use (clinical trials) - Human Tissue Act
What do most peer-reviewed journals now require?
- Patient consent
- Registration
- Ethical approval
Patient consent
whether and how consent was obtained
Registration
Where the trial was registered
before the trial began
Ethical approval
the ethics committee or Institutional Review Board who approved the trial
What are the risks of stopping a trial early?
- Goes against the protocol
- May introduce an unconscious bias
- Difficult to do a correct statistical analysis
What are the advantages of stopping a trial early?
- Concerns about safety if the trial continues
- Treatment seems effective - could give it to everybody
- Treatment seems ineffective - could stop wasting participants time
Who makes the decision to stop?
- DSMB - Data Safety + Monitoring Board
2. DMEC - Data Monitoring + Ethics Committee
DSMB + DMEC
- Independent of the trial
- They may access unblinded interim data
- They can recommend to the sponsor that the trial be stopped
- Remit established before the trial begins
- Not required for every trial
Ethical conduct of clinical trials
Key principles:
- Justifying sample size
- Obtaining informed consent
- Registering and following a protocol
- Independent oversight
