Hadpop Flashcards
(60 cards)
Define necessary
A cause always precedes the disease
Define a cause
Something that increases probability of getting a disease
Define sufficient
A cause alone can lead to the disease
Possible explanations for systematic and random variation
Confounding Bias: Healthy worker effect Group comparison Information bias Chance measured by the p value
What are the Bradford hills criteria for causality?
Strength of association Specificity of association Consistency of association Reversibility Dose response Temporal sequence Coherence of theory Biological plausibility
What are the 2 purposes of a clinical trial?
Safety - ability of health care intervention to not harm a population under certain conditions
Efficacy - ability of a health care intervention to benefit a defined population under certain conditions
What does a clinical trial new to be in order to give a fair comparison of effect and safety?
Fair
Large trial
Controlled
Reproducible
What is a non randomised control trial?
Where new patients with new treatment are compared with the old group receiving the standard treatment
What are the disadvantages of non randomised control trials?
Selection bias
Confounding
What are historical control trials?
Involves comparison groups who had the treatment with a group receiving the new treatment
What are the disadvantages of historical trials?
Selection less defined and vigorous
Treats differently
Less info
Unable to control confounders
What are the factors to consider with randomised control trials?
Treatment Outcomes Patients taking part Patients not to be considered Disease of interest Bias and confounders
How do you conduct randomised control trials?
Identify Invite Consent Allocate Follow up Minimise losses to follow up Maximise compliance Analyse
What is the comparison of the outcomes in a randomised control trial?
Is there an observed difference?
Is it due to chance?
Is it due to the treatment?
How big is the difference?
Give some examples of randomisation
Number tables
Computer generated
Define single blind
When only one person is blinded
Clinician/patient/assessor
Define double blind
When two people are blinded
Clinician/assessor/patient
Define triple blinded
When three people are blinded
Clinician/assessor/patient all are
Only pharmacist will know treatment allocation
What is the placebo effect?
Inert substance made to look like the active one
Same packaging
Only used when no standard treatment available
It’s deception so may patients be aware that they may be getting a placebo
How do you conduct a case control study?
Identify a group of cases
Identify a suitable group of non cases
Ascertain previous exposure status of everyone
Compare level of exposure in cases and controls
What are the features of a case control study?
Cheaper than cohorts Quicker than cohorts Not good at rare exposures Good at rare diseases Study a range of exposures for that outcome Compares exposure to disease status
How do you work out the odds ratios for case control?
OR = ad
bc
How do you work out the error factor for the odds ratio for the case control?
EF = exp { 2 x square root of [(1/a) + (1/b) + (1/c) + (1/d) ]}
What happens if the number of controls is increased?
Effor factor decreases as b and d will be larger
