Statistics Flashcards
Odds ratio
Odds ratio = Positive relationship direction/Negative relationship direction (AD/BC)
Attributable risk
Attributable risk = Incidence of disease in non-exposed persons - Incidence of disease in exposed persons
Reflects additional incidence of disease related to exposure taking into account background rate of disease
Population attributable risk
Used to describe contribution that exposure has on incidence of specific disease in population
Population attributable risk = Attributable risk x Prevalence of exposure to risk factor in population
PAR% = cases caused by exposure/all cases
Number needed to treat
Reciprocal of absolute risk reduction
1/ARR
Error types
Type 1 (a) error - claiming there is true difference when there is not (p value is here)
Type II (B) error - claiming there is no difference when in fact there is (power is here) –> usually occurs when number of subjects is small, so power of study is too weak to detect difference
Confidence intervals
Range around point estimate inside where actual value might lie, with degree of confidence
Definition of selection bias
Occurs when people are selected to go into various exposure groups in a study
Definition of observation bias/measurement bias
Occurs due to use of unreliable or invalid measurement of study variables
Definition of recall bias
Occurs when people with certain disease state or exposure, remember experiences differently from those not affected
Definition of publication bias
Occurs when journals only tend to publish trials with positive results
Definition of lead time bias
Occurs when comparing survival of patients detected through screening vs patients detected from clinical signs/symptoms
Artificially longer survival time creating impression that screening is beneficial
Definition of length time bias
Overestimation of survival duration due to the relative excess of cases detected that are asymptomatically slowly progressing, while fast progressing cases are detected after giving symptoms –> Factually longer survival caused by slower disease progression gives impression that screening was beneficial
Definition of intention to treat analysis
Subjects groups according to treatment they were randomised to, regardless of which treatment they actually received
Provides more realistic estimate of effectiveness of treatment
Phases of clinical trials
1) Toxicity screen for maximum tolerated dose in humans
2) Test for positive effect in humans for fixed dose(s)
3) “Usual” clinical trial to measure relative efficacy - most frequently an RCT
4) Long term surveillance of users of treatment to detect low-incidence adverse effects
Sensitivity measurement
Sensitivity = True positive/All patients who have disease present
