2. Randomised Control Trials

Question 1

What is a clinical trial?

What are the 3 main types of clinical trial?

Answer 1

A clinical trial is an experiment in which a treatment is administered to humans in order to evaluate its efficacy and safety.

1) An uncontrolled trial – everyone gets the treatment
2) A controlled trial–treated group compared with untreated group (placebo), or treated group compared with control group having “usual treatment”. Controls may be geographical, historical, or randomised
3) A randomised controlled trial - Allocation to groups determined by chance

Question 2

What are geographical controls?

What are historical controls?

Can we randomise using alternate allocation?

What are some benefits of randomised controls?

Answer 2

Patients with the same disorder seen at another hospital or clinic where the new intervention is not provided. Selection bias

Patients with the same disorder seen in the past before the use of the new intervention. Selection bias.

No - allocation bias

Helps ensure group receiving treatment A is similar to that receiving treatment B. Avoids selection/allocation bias

Question 3

What is a single or double blind trial?

What are the 2 main types of randomised control trial?

Answer 3

Single blind – patients don’t know what treatment they’re on

Double blind – also the observers don’t know what treatment the patients are on (not always possible)

Parallel group and crossover

Question 4

Would you use a parallel group or cross-over randomised control trial for the following:

a) Efficacy of analgesics for chronic pain (eg osteoarthirits)
b) Efficacy of aspirin to prevent heart attacks
c) Efficacy of chemotherapy to treat cancer
d) Efficacy of statins to lower cholesterol

Ultimately, when do you use parallel group, and when do you use cross-over?

Answer 4

a) Cross-over
b) Parallel group
c) Parallel group
d) Cross-over

Parallel group when effect of treatment is irreversible; cross-over when it’s reversible

Question 5

What are some advantages and disadvantages of cross-over trials?

What is a cluster randomised trial?

Answer 5

Advantages: each patient is their own control, smaller sample size to get same number of observations, better for subjective measurements

Disadvantages: more time consuming, carry-over effects – carry over effect of one treatment into other treatment period - so add in washout period

Randomise pre-existing groups e.g. villages to one of two treatments. Avoids contamiation and enhances compliance

Question 6

Which should be designed as a cluster randomised trial?

1. Trial of aspirin versus placebo to prevent pre-eclampsia
2. Trial of addition of flocculant disinfectant to drinking water in the prevention of diarrhoea
3. Trial of text message reminders versus standard care to improve drug adherence

Answer 6

2

Question 7

Why is a factorial trial useful?

What is the aim of a non-inferiority trial?

What are the phases in developing and evaluating a new drug?

Answer 7

Assess 2 interventions using the same number of patients as 1 intervention.

Aim to show a new intervention is no worse than a current intervention. Analysis based on confidence interval only.

1) PRECLINICAL – Non-human study to obtain preliminary efficacy, toxicity and pharmacokinetic information

2) PHASE 0 – First in-human trials, small number of subjects given subtherapeutic dose of drug to determine pharmacodynamics and pharmacokinetics

3) PHASE 1 – Screening for safety, testing of drug on (usually) healthy volunteers for dose ranging.

4) PHASE 2 – Assess efficacy and safety - determine whether drug can have a therapeutic effect. Case series or randomised controlled trial

5) PHASE 3 – Assess efficacy and safety - randomised controlled trial on large number of patients to determine what the therapeutic effect is

6) PHASE 4 – Post-marketing surveillance

Question 8

What are some advantages of registering clinical trials?

Answer 8

• Assist in the planning of new trials
• Avoid unnecessary duplication of research
• Avoid subjecting patients to trials seeking evidence that is already available
• Encourage collaboration between research groups
• Facilitate optimal use of research funds by funding agencies
• Facilitate patients’ access to information and improve recruitment
• Improve opportunities for methodological research
• Reduce discrepancies between published results and original trial protocol
• Help to detect publication bias in meta-analyses

Question 9

How do you calculate the relative risk of e.g. death in a treatment group compared to a control group?

What is the relative risk if there is no effect of treatment?

Answer 9

[Pic]

RR = 1

Question 10

How would you calculate the relative risk of low IQ using the data below?

Answer 10

Question 11

Which is the correct interpretation of a relative risk of 0.86?

1. Children of mothers in the treatment group were 14% less likely to have an IQ≤85 than children of mothers in the control group
2. The risk of a child having an IQ≤85 if the mother received treatment is 86%
3. Children of mothers in the control group were 14% less likely to have an IQ≤85 than children of mothers in the treatment group

Answer 11

1. If no difference, relative difference =1, so 1 - 0.84 = 14%

Question 12

What is intention-to-treat analysis?

Answer 12

Comparison of all subjects based on the treatment group assigned, regardless of whether they complied. This is the primary analysis (rather than on-treatment)

Question 13

What is on-treatment analysis?

Answer 13

Comparison of subjects who actually took treatment

Question 14

Poor compliance on an intention-to-treat analysis reduces ability to detect treatment difference (if one exists) - how could you maximise compliance?

Answer 14

– Selection of patients (not too ill)

– Double blind design

– Run in period where all get treatment (to identify those who can’t tolerate it)

Question 15

How many patients would need to be treated to prevent one patient getting the disease/disorder? Use the scenario below - what calculations would you do?

Answer 15

Question 16

Risk of death in treatment group = 20%

Risk of death in the placebo group = 30%

Which is the correct number needed to treat?

1. 30
2. 20
3. 10
4. 1.5

Answer 16

10. (100/(30-20) = 10)

Question 17

Relative risk of death of treatment A compared to B = 2. Is the result statistically significant in each of the trials 1, 2 and 3?

Answer 17

1) No - trial not big enough for significance
2) No but not far off p = 0.05 and CI only just crosses 1
3) Yes - CI doesn’t cross 1 and p < 0.05

Question 18

What percent is statistical power set at?

In terms of CI and p-value, when is a result statistically significant?

What is the purpose of meta-analysis?

In what form are the results usually displayed in?

Answer 18

85%

Statistically significant if 95% CI for relative risk does not include 1, and if p-value <0.05

To bring together all the evidence to more powerfully estimate the effect size, can be done for cross-sectional, case-control, cohort studies and randomised trials.

Forest plot

Question 19

What do the following mean in a Forest plot:

a) dots
b) lines
c) box

Answer 19

a) RR
b) CIs
c) study size

Question 20

Describe the 2 issues with meta-analysis

Answer 20

1) Heterogeneity - variation in study results. Caused by difference in study design, participant characteristics, intervention e.g. drug dose, and chance. Can be assessed statistically – is there greater heterogeneity than you would expect by chance?

2) Publication bias - studies with significant or favourable results more likely to be published. Caused by investigators, journal editors and peer reviewers. Can be assessed statistically and graphically – funnel plot