Clinical Trials Flashcards
What is the advantage of the randomised trial over clinical trials in general?
The potential to limit/eliminate confounding
NB: this is the only advantage, but it is a big one
Aims of clinical trials?
- unbiased evaluation of effects of an intervention
- ethical practice: no person is harmed
Outline types of trials?
(Phase 0: drug studies that aim to find how a drug behaves in humans. Usually low doses of the drug small numbers)
Phase I: Safety studies
Phase II: Exploratory (pilot) trials to see if it can be done
Phase III: Larger explanatory trials to see if it efficacious in controlled conditions, or effective in real-world clinical setting?
Phase IV: post-marketing surveillance
Potential problems in trials?
- lack of power (sample size)]- type II error vs cost
- selection criteria
- outcome choices
- mutiple aims and measures]- type I error
- drop out
- self selection of participants
- researcher bias
- S/Es
Define Type I error and Type II error?
Type I error: false positive]- this result in ppl getting treatments which are ineffective
Type II error: false negative
Potential solutions to the problems with trials?
- sample size: power calculation (90%)
- selection criteria (explanatory-narrow; pragmatic-broad)
- outcome choices (literature, stakeholders, qualitative data)
- outcome measures (single primary outcome, trial registration)
- self-selection: recruitment strategy, CONSORT diagram
NB: CONSORT = Consolidated Standards of Reporting Trials
What to consider when calculating sample size?
- dichotomous or continuous outcome data
- estimate of the size of the effect that is clinically important
- power of a study: 1-β (where β is probability of Type II)
- level of statistical significance - α (probability of null rejection)
- randomisation ratio
What is intention to treat (ITT)?
Include ALL those randomised to intervention and control treatments when analysing data
DON’T EXLUDE DROPOUTS
Compliant ppl may be different to non-compliant and these differences may be associated w/ better outcomes
E.g. a completely useless treatment might look efficacious if you only compare before and after for those who complete the study and ignore the drop outs.
Potential solutions for observer bias?
- blind researchers to participants exposure status
- blind participants (placebos)
- blind researchers to study hypothesis
- use objective criteria to assess study outcome (routine data from independent records e.g. hospital admission)
- train researchers to use validated questionnaires
Solutions to:
drop out rate:
reporting bias:
publication bias:
fraud:
Solutions to:
drop out rate: rigorous follow up, regular contact, next of kin
reporting bias: publication of trial protocol, open access to study data
publication bias: transparency
fraud: commercial interest “study 329”, publish trial data for independent analysis
Relevance of pharmacological industry sponsorship vs non-pharm sponsorship?
4 times more likely to get a positive result for pharm-industry funded…
This highlights the need to declare unpublished data too (otherwise you can just not publish the trials that don’t give the conclusions that you want)
Ethical principles refresher?
4 Ethical Principles
- non-maleficience (inconvenience and confidentiality)]- you cant do a study to prove something “is bad” as it’s unethical; trials can only prove if something “is better”
- beneficience (study will benefit pt)
- patient autonomy (pt choice)
- justice (fair risk and benefit distribution)
Importance of side effects in clinical trials?
- standard operating procedures
- DMEC: to review side effects only
- rules to stop trials if certain criteria are met
- same can occur for psychological therapies
- access to unblinded data
What to consider when critically appraising a trial?
[see lecture as this is incomplete]
- randomisation procedures explained
- clearly defined population
- clearly defined intervention and control treatment
- researchers and participants masked
- etc…
