Evidence - Sem 4

Question 1

What is a clinical trial? (4)

Answer 1

Any form of planned experiment which involves patients and is destined to explain the most appropriate method of treatment for future patients with a given medical condition.

Question 2

Describe the purpose of a clinical trial. (5)

Answer 2

Provide reliable evidence of the:

• efficacy - ability of the intervention to improve health under gold standard conditions.
• safety - the ability of an intervention to not cause harm under gold standard conditions.

Question 3

Describe the inclusion criteria of a study. (1)

Answer 3

Characteristics that place an individual in the experimental group.

Question 4

Describe the exclusion criteria of a study. (1)

Answer 4

Characteristics that would prevent an individual from takin part in the study.

Question 5

Explain the difference between primary and secondary outcomes. (4)

Answer 5

A primary outcome is the main thing you’re investigating and is used to calculate sample size. This needs to be pre-determined so you can write the aims of the study.
A secondary outcome is a different outcome of interest, often occurrence of side effects. This can be written later based on what you find out.

Question 6

Describe the three types of outcome that can come out of a clinical trial. Give an example for each. (6)

Answer 6

Pathophysiological - tumour size
Clinically defined - death
Patient-focused - quality of life.

Question 7

Give 5 examples of baseline data you might need to collect on your participants for further research later. (5)

Answer 7

Age
Gender
Previous experience 
Other drugs 
Other disease states

Question 8

Define clinical equipoise. (4)

Answer 8

The state in which one should exist to be able to conduct a clinical trial: genuinely not knowing which treatment option in a trial is better. This is because testing new drugs could cause harm, so you have to believe that it could be a new amazing drug to make it worth the risk.

Question 9

Describe the stages in drug development. (10)

Answer 9

Pre-clinical lab studies: testing on cell cultures and animals.
I - volunteers: pharmacokinetics/-dynamics, side effect testing on 100 healthy volunteers.
II - treatment studies: effect, dose, side effect testing on 1000 patients.
III - clinical trials: comparison with standard treatment on 10,000 patients.
IV - surveillance: potential new use analysis and monitoring while rolled out to the whole population.

Question 10

Why are clinical trials randomised? (4)

Answer 10

Minimises allocation bias - equal chances of being in each group.
Minimises confounding - groups should end up similar in size and characteristics by chance (known and unknown confounders).

Question 11

Describe ways of executing non-randomised clinical trials. (3)

Answer 11

Location - Hospital 1 gets drug 1.
Numerical order - first 1000 patients get drug 1.
Historical - everyone gets drug 2 now and we compare to the people who got drug 1 last year.

Question 12

Describe the flaws with using a historical, non-randomised clinical trial. (2)

Answer 12

The standard treatment group (in the past) might not have been treated as standard because they weren’t in a trial at the time, and it might have more confounders that weren’t adjusted for.

Question 13

Define open label allocation and describe the effects of this. (4)

Answer 13

When both the doctor and the patient knows which drug they are getting.
Can lead to:
- behaviour effects - patients change their behaviour
- non-treatment effects - clinicians change their behaviour
- measurement bias - clinicians change the way results are taken / recorded.

Question 14

Explain the three types of blinding. (6)

Answer 14

Single - one out of clinician, patient (normally me) and researcher doesn’t know what group people are in.
Double - two don’t know (normally clinician and patient)
Triple - three don’t know (rare because researcher often is clinician).

Question 15

Give 5 examples of where blinding can be difficult. (5)

Answer 15

Surgical procedures.
Psychotherapy vs antidepressant medications
Alternative (acupuncture) vs Western drugs
Lifestyle interventions (exercise) vs drugs
Prevention programs.

Question 16

Describe the features in which a placebo must match the active drug. (6)

Answer 16

Size
Taste
Texture
Dose regime
Warnings
Appearance

Question 17

Explain the two main ethical issues surrounding placebos. (4)

Answer 17

To give an inactive placebo, there must be no other available drug, because it is unethical to leave patients without treatment.
Patients must be aware of the possibility if receiving a placebo to take part in the trial with informed consent.

Question 18

Describe the two types of losses to follow up seen in clinical trials. (2)

Answer 18

Appropriate - their condition necessitates their removal from the trail.
Unfortunate - the patient chooses to withdraw.

Question 19

Explain 4 ways to minimise the loss to follow up. (4)

Answer 19

Make the follow-up practical and minimise inconvenience
Be honest initially about commitments
Avoid coercion or inducements
Maintain contacts with participants.

Question 20

Give 5 examples of reasons that a patient might give to not comply with their treatments. (5)

Answer 20

Misunderstanding 
Dislike
Ineffectivity  
Alternative treatments 
Laziness

Question 21

Describe 4 methods for increasing compliance to treatments. (4)

Answer 21

Simplify instructions
Ask about adherence
Ask about effects and side effects
Monitor adherence if possible - tablet count, urine levels.

Question 22

Describe an “as treated trial”. (2)

Explain why this is not ideal. (4)

Answer 22

Analyses only those who completed follow-up and complied fully with treatments.
Negates the effects of randomisation because non-compilers are likely to be systematically different to compilers, creating selection bias. Also tends to give unrealistic large sizes of effect.

Question 23

Describe an “intention to treat” trial. (2)

Explain why this is better than an “as treated” trial. (3)

Answer 23

Analysis completed according to group allocation regardless of whether the patient adhered or completed follow up.
Preserves the effects of randomisation, is more realistic to general practice, and gives more realistic ideas of effect.

Question 24

Explain the concept of a scientifically robust clinical trial. (3)

Answer 24

Actually will get a reasonable answer to a question that needs answering while maintaining patient health and study reliability.

Question 25

Explain the concept of inappropriate inclusion. (4)

Answer 25

The inclusion of communities unlikely to benefit in the long run (eg testing on people who can’t afford the drugs), or those at increased risk anyway (eg pregnancy, other diseases).

Question 26

Explain the concept of inappropriate exclusion. (4)

Answer 26

The exclusion of communities that are difficult to get informed consent from (eg the elderly, the mentally ill - they’re still people you still have to include them), and the exclusion of people who differ from your “ideal” group (eg women, non-whites).

Question 27

Explain what it means if a trial is voluntary. (2)

Give two categories of things that could compromise this, and give examples in each. (4)

Answer 27

Free to withdraw at any time and without repercussions.
Coercion - non-access to “best” treatment.
Manipulation - exploitation of state, distortion of info.

Question 28

Define collective ethic. (2)

Answer 28

All patients should have treatments tested for efficacy and safety.

Question 29

Describe the pillars of individual ethic, and explain why clinical trials could be seen to be going against these ideas. (8)

Answer 29

Beneficence (no guaranteed benefit), non-maleficence (can cause harm), autonomy (chance), justice (benefits and burdens random).

Question 30

Explain the difference in confidence intervals worked out using relative risk data and absolute risk data. (2)

Answer 30

Relative risk makes the null hypothesis = 1

Absolute risk makes the null hypothesis = 0.

Question 31

Describe the hierarchy of evidence types.

6

Answer 31

Most reliable 
Systematic reviews of randomised trials 
Randomised control trials
Non-randomised control trials 
Observational studies - case control, cohort, cross sectional 
Case studies 
Expert opinion 
Lease reliable

Question 32

List 4 common critique criteria for randomised control trials. (4)

Answer 32

Blinding appropriate?
Intention to treat analysis?
Randomised appropriately?
Appropriate inclusion and exclusion criteria?

Question 33

Explain the purpose of the Cochrane Collaboration. (3)

Answer 33

To act as a worldwide meta analysis that is able to review and collate all the evidence, where the overall confidence interval and p value can be presented, which are both normally tiny.

Question 34

Describe the flaws in using RCTs in all “good” sources of evidence. (3)

Answer 34

Not everything needs a RCT to show its effectiveness (eg blood after a haemorrhage)
Not everything can be tested with a RTC (eg smoking)
Not every question has been around long enough to get many RTCs on it to meta analyse yet.

Question 35

Explain the PICOS method for searching PubMed. (5)

Answer 35

Patient group or population
Investigations or interventions considered
Control 
Outcomes considered
Study design

Question 36

Describe a systematic review. (2)

Answer 36

A systematic review used to summaries and communicate large quantities of research on a similar topic.

Question 37

Describe a meta-analysis. (2)

Answer 37

A quantitative synthesis of the results of two or more primary studies that addressed the same hypothesis in the same way. Basically combines evidence.

Question 38

Explain a situation where a systematic review may not include a meta-analysis. (2)

Answer 38

If the clinical heterogeneity is too great - eg if that particular meta-analysis was too different to the others included.

Question 39

Explain the point of a meta-analysis. (3)

Answer 39

Facilitate the collection and collation of a large number of study results.
To reduce problems in interpretation due to sampling variation.
To quantify the size of effect and uncertainty as a pooled estimate.

Question 40

Describe the quality criteria of meta-analysis. (4)

Answer 40

The complete set of studies
Identification of the common variable between studies
Standardised methods of data extraction
Analysis allowing for variation.

Question 41

Explain how confidence intervals are pooled. (2)

Answer 41

Each confidence interval is weighted according to how big it is (smaller CI = more weight).

Question 42

Describe a forest plot. (10)

Answer 42

Each study’s odds ratio is indicated by a box. The size of the square represents the weight given to that study.
Each study has a horizontal line indicating the confidence interval.
The Y axis is the null hypothesis (an odds ratio of 1 normally)
The X axis is the odds ratio.
The diamond is the pooled estimate, with the centre (with the dotted line) representing the pooled odds ratio and the width of the diamond representing the pooled confidence interval.

Question 43

Describe 3 problems with meta-analysis. (4)

Answer 43

Heterogeneity between studies (fixed effect model vs random effects model)
Variable quality of the studies
Publication bias in the selection of the studies.

Question 44

Describe the two models of variation. (5)

Answer 44

The fixed effect model assumes that all studies are estimating the same “true effect”.
The random effects model assumes that the studies are estimating similar but not the same “true effect” value, making the really “true effect” value the “true mean effect”

Question 45

Describe the differences in measuring odds ratio, confidence interval and weighting of studies in the different models of variation. (3)

Answer 45

Odds ratio - similar in both.
Confidence interval - wider in the random effects model.
Weighting - more equal in the random effects model (more weight given to smaller studies).

Question 46

Describe how a forest plot would change if the variable was continuous rather than a measured effect. (3)

Answer 46

The Y axis of the forest plot sits at 0 on the X axis rather than 1, because the null hypothesis becomes mean difference (where null would be no difference) rather than odds ratio (where no difference would be 1:1).

Question 47

Describe how variation between studies can be assessed. (5)

Answer 47

Sub group analysis is used to explain heterogeneity:
Study characteristics - year of publication, length of study, % female
Participant profile - where data is analysed separately by type of participant - male female, adult child.

Question 48

Describe three things that a study can have low quality due to. Indicate which is the hardest to measure (*). (4)

Answer 48

Poor study design.
Poorly designed protocols
Poorly followed protocols. *

Question 49

List four study designs in an order of susceptibility to bias. (6)

Answer 49

Randomised control trials - least susceptible to bias.
Non-randomised control trials
Cohort studies
Case control studies - more susceptible to bias

Question 50

Describe the two approaches to helping to combat bias or low study quality. (5)

Answer 50

1 - define a basic quality standard and only accept studies that meet that eg Cochrane Reviews only include RTCs.
2 - score each study for quality then:
- incorporate the score into the weighting so higher quality studies have more weight.
- use sub-group analysis to investigate differences in study quality.

Question 51

Describe characteristics that can be used to assess the quality of a study. (5)

Answer 51

Allocation method eg randomisation.
Blinding and outcome assessment
Patient drop out - ideally less than 10% and “intention to treat”
Appropriate statistical analysis.

Question 52

Describe the concept of publication bias and explain the effect that this has. (5)

Answer 52

Studies with statistically significant or “favourable” results are more likely to get published (especially relevant with size of study).
Any systematic review or meta-analysis is therefore flawed because they are biased to the demonstration of any effect.

Question 53

Describe a Funnel Plot. (4)

Answer 53

Plots 1 / standard error on the Y
Log (odds ratio) on the X
Plots the results for each of the study results as a point on the graph. Should fit a funnel shape with the wide edge on the X axis.

Question 54

Describe a biased funnel plot. (2)

Describe an unbiased funnel plot. (2)

Answer 54

Biased = gaps in the funnel shape - not symmetrical. 
Unbiased = no gaps and symmetrical.