From Allan Flashcards

1
Q

What is a confounder?

A

A factor, not the intervention being given, that may be responsible for creating a correlation between the intervention and the disorder of interest.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is the problem when using historical controls?

A

Because a therapy being offered now is being compared to a therapy being offered in the past. The difference in time is likely to have an effect because it may reflect differences in patient characteristics, methods of diagnosis or standards of care.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Why might randomised trials not always be the appropriate trial design?

A

1) Sample size - if a large sample size is required to detect a difference between two treatments, it may be too large to be feasible to perform.
2) Bias -

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is a benefit of randomising a phase II trial?

A

It can provide information on the feasibility of a subsequent phase III trial, such as how willing subjects are to be randomised.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What are four key trial design features?

A

Inclusion and exclusion criteria
Randomisation
Blinding
Control group

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the importance of exclusion and inclusion criteria?

A

Unambiguous definitions to make recruiting subjects easier. The criteria produces a population that should have little variability, making them more likely to respond to the intervention in a similar way, making it easier to detect subtle differences between each group.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What may be a problem associated with exclusion and inclusion criteria?

A

If many criteria were used, the population may not be generalisable, and not represent the actual population likely to receive the intervention.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is a problem associated with using only few exclusion and inclusion criteria?

A

Whilst the results would have a more general application, the amount of variability between the patients would be high, making it more difficult to show if a treatment is effective. When there is much variability, sometimes only large effects can be easily detected.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

When is it unethical to give a control group a placebo?

A

When an established treatment already exists.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is randomisation?

A

A process of allocating subjects between different trial interventions. Each subject has the same chance of being allocated to any group, which ensures similarity in characteristics between the arms.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What effect does randomisation have on confounding factors?

A

It reduces the effect of both known and unknown confounding factors.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What effect does randomisation have on bias?

A

Minimises it. If either the researcher or trial subject is allowed to decide which intervention they are to receive, then subjects with certain characteristics could be over-presented in one of the trial arms.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is selection bias?

A

Can occur if choosing a particular subject for the trial is influenced by knowing the next treatment allocation.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is allocation bias?

A

Involves giving the treatment that the clinician or patient feels might be most beneficial. Sometimes, the researcher has access to the list of randomisations from which the next allocation can be determined, possibly creating allocation bias.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

How can allocation bias be avoided?

A

By using a central office, or computer system in charge of randomisation, because the researcher has no control over the process (allocation concealment)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the main limitation of small trials?

A

The interpretation of their results. Due to their small size it is more likely that they will produce false-positive results or overestimate the magnitude of the treatment benefit.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Why are surrogate outcome measures used?

A

If a trial would be impractically large, or long because of a true endpoint with too few events to produce a reliable evaluation of an intervention. Surrogate endpoints are considered as precursors of the true endpoint and produce more events in a shorter space of time.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What surrogate marker is often used in clinical trials and is it a good marker?

A

Tumour response - not a good marker as tumour response often doesn’t correlate well with survival. It is appropriate for use in phase I/II trials, it wouldn’t be used in phase III.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What are the three types of outcome measure?

A

1) Counting people (how many have the health outcome of interest)
2) Take measurements on people
3) Time to event measurements.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Why should both arms of a trial receive their examinations at the same time (when the outcome measure is disease progression)

A

To prevent bias. For disease progression, the date of disease progression is required (time to event). Given that the time to event won’t actually be the day of the examination, the examinations must be at the same time to avoid progression appearing to be sooner/later in one intervention arm compared to the other.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Why are phase III trial patients seen more often? (with additional assessments)

A

To closely monitor the patient’s well being, also to try and get accurate dates of progression/relapse (usually an endpoint)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Why is there a need to balance the frequency of assessments in phase III trials?

A

Increased costs, inconvenience to patients, or increased radiation exposure (scans)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

What are some main purposes of new cancer therapeutic interventions?

A

Cure the cancer
Extend life/ progression free survival (for late stage/advanced cancers)
Reduce symptoms of advanced disease (QOL)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

What type of data is a time-to-event measure, and what is the name of its effect size?

A

Type = Time-to-event. Effect size = Hazard ratio

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

What type of data is tumour response rate and what is its effect size?

A

Counting people in categories (have they had the response). Effect size is relative risk/odds ratio, or risk difference.

26
Q

What type of data is safety/toxicity/adverse event rate? What is its associated effect size?

A

Type = counting people in categories (have they experienced the AE). Effect size = relative risk/odds ratio, or risk difference

27
Q

What type of data is QoL/ biomarker level?

A

Taking measurements on (or from) people. Effect size = difference in the mean/median scores.

28
Q

What comments can be made on OS?

A

It’s the gold standard primary endpoint in phase III trials. It is easy to define, don’t need the cause of death, only the date. It allows for deaths due to drug toxicity.
Disadvantages:
Can take many years to see enough events (long trials). Usually not as many events as other endpoints, so hold less statistical power. If the treatment only effects the cancer of interest, then including other causes of death can dilute the treatment effect. It also doesn’t take into account treatment received post-progression on the intervention.

29
Q

What is the disadvantage of cancer-specific survival?

A

It assumes that the treatment has no effect on other causes of death which may be unlikely. It also requires the cause of death data to be reliable which is not always the case.

30
Q

What is disease-free survival? When is it used?

A

An event is death from any cause or first recurrence of the disease. Its useful when patients are thought to be free from disease and have a good prognosis.

31
Q

What is EFS?

A

Event free survival: Event is a recurrence of the original cancer, death from any cause or a new tumour.

32
Q

What is the benefit of using PFS?

A

As soon as the patient gets recurrence, they are managed and treated differently compared to those on the trial. Therefore what happens after progression will be difficult to associate with the original trial treatment. PFS prevents this problem. Also useful for advanced disease where patients have not been cured and are expected to get worse in the near future. Gives a large number of events in a smaller amount of time.

33
Q

When is tumour response rate useful?

A

In advanced cancers or pre-surgical therapies

34
Q

What is the need for measuring treatment compliance?

A

Both compliance of the new drug but also compliance of the standard drug if given at the same time as the new drug. Does the new drug reduce compliance to standard drug which is known to work.

35
Q

What are some definitions of compliance?

A

Stop taking the drug completely.
Delay drug delivery
Dose reduction

36
Q

What is the problem with a small sample size?

A

You may miss important differences because of chance variation.
The 95% confidence intervals might be wide
Difficult to make robust conclusions
You may see spurious associations

37
Q

What is the problem with a large sample size?

A

If too large, you waste resources because you could have found a clear answer sooner

38
Q

What is the p value?

A

A significance test that tells us how likely it would be to observe an effect as large as the one reported simply by chance, IF there were really no effect. It can also be thought of the probability of making the wrong conclusion that a treatment effect exists when in fact there is no real effect.

39
Q

What information is needed to calculate the sample size in a phase III trial?

A

The expected effect of the new intervention
The expected effect in the control group
The significance level (usually 5% at the two sided level)
Power (usually 80 - 90%)

40
Q

When would the 5% significance level only be one sided?

A

If the trial objective is non-inferiority or that it can only be better.

41
Q

What is power?

A

The chance of finding an effect of the magnitude specified, if it really does exist.

42
Q

What three things would cause a sample size to get larger and why is this done?

A

1) The effect size gets smaller - Its harder to detect small differences than larger ones
2) The power goes up - Increases the chance of picking up the effect if it really exists
3) Significance level goes down - Decreases the chance of saying there is an effect when there is no true effect .
The number of events is also important. A large study with few events can have low power to detect small or moderate treatment effects.

43
Q

What are three possible reasons for a result that shows no statistical significance

A

There really is no difference
There is a difference, but by chance the sample of patient for the trial did not show the effect
There is a difference, but the study had insufficient power to show it reliably (because it was too small)

44
Q

What is the no effect value for risk difference and the difference between two means?

A

0

45
Q

What is the no effect value for hazard ratios and relative risk/odds ratios?

A

1

46
Q

What is likely if the observed medians do not match with the observed medians in survival?

A

If the control group median differs significantly then it may be due to the patient population being either lower or higher risk than expected

47
Q

When would patients care most about grade 1/2 AEs?

A

If sever AEs are rare, for early stage disease with good prognosis, if the drug is used for long term maintenance.

48
Q

What is a predictive marker?

A

A biological, imaging, or histological factor that can be used to determine whether a new intervention has a different effect between different groups of patients (eg ER positive tumours in breast cancer).

49
Q

What are some general considerations when evaluating a predictive factor?

A

How easy is it to measure
Is the marker reliable
What is the extent of measurement error - minimised by a central lab being used, standard operating procedures etc
Financial cost of measuring the marker

50
Q

If a phase III trial is not blinded, what sort of outcome measure should be used?

A

One that is not subjective, as this would be effected by the lack of blinding.

51
Q

Why might an interim analysis be performed in a phase III trial?

A

Early looks at the data aim to possibly end the trial early because a large treatment effect is found, or the effect observed is so small that there is unlikely to be a clinically important effect if the trial were to be continued.

52
Q

What is stratified randomisation?

A

A technique that attempts to guarantee balance for some important stratification factors.

53
Q

Why is recruiting centre or region used as a stratification factor?

A

each region may have subtly different standard of care which may influence the outcome of treatment, and environmental and genetic differences between patients from particular regions may also influence the prognosis and response to certain treatment.

54
Q

What is minimisation?

A

Technique that aims to ensure balance between the treatment groups for pre-specified prognostic factors. The treatment allocation for the first few subjects can be made using simple randomisation. However the allocation of each subsequent subject depends on the distribution of the stratification factors among those who have already been randomised, and not using random numbers.

55
Q

Why might unequal randomisation be used?

A

More subjects needed in one arm because more reliable data might be needed on the effects of the new treatment, or to make it more appealing for subjects to participate in the trial, knowing that they have a higher chance of receiving the intervention.

56
Q

How can you check if randomisation has worked?

A

Examining the baseline characteristics

57
Q

When might risk difference at a pre-specified time be more suitable than using a HR?

A

Because HR assumes that the treatment effect is similar over time. If this is not the case then risk-difference at pre-determined points may be more suitable.

58
Q

How does a studies size affect conclusions

A

In a large trial there is a small standard error, this produces narrow CIs which give a precise estimation and firm conclusions.
In a small trial the standard error is large, producing wide confidence intervals, giving imprecise estimates and no firm conclusions.

59
Q

What is the main benefit of intention to treat?>

A

It maintains the balance in baseline characteristics. It also means the effect size reflects what would happen in practice, because not all people will take the intervention, or some might stop early due to side effects.

60
Q

When would the number of affected subjects rather than the number of events be reported? (adverse events)

A

When people can suffer several (perhaps related) side-effects. As this might cause the extent of harm in one trial arm to be over-estimated.