drnieves's version from 2017-06-20 02:27


Question Answer
Correct resultThere is an effect when one exists.
Correct resultNull hypothesis rejected
Correct resultThere is no effect when none exists
Correct resultNull hypothesis accepted.
Type 1 errorAlternative hypothesis accepted when null hypothesis is correct
Type 1 errorFalse positive
Type 2 errorNull rejected accepted when is false
Statistical power1-B
Statistical powerProbability of rejecting the null hypothesis when it is false.
Increase power (decrease B)Increase sample size, expected effect size, precision of measurement.
NPVProportion of negative test result that are true negative
Probability that a person actually is disease free given a negative result.NPV
NPVInversely proportion to pretest probability or prevalence.
PPVProportion of positive test results that are true positive.
Probability that a person actually has the disease given a positive test.PPV
PPVDirectly proportional to pretest probability or prevalence
Fixed properties of a testSpecificity and sensitivity
high sensitivity testScreening in diseases with low prevalence.
High specificity testFor confirmation after a positive test result.
Results bias`Confounding and lead time bias
Confounding biasWhen a factor is related to exposure and outcome, but not on a casual pathway. The factor distorts effect of exposure on outcome.
Reduce confoundingMultiple/repeated studies
Cross-over studiesTo reduce confounding bias
Reduce confoundingMatching (pts with similar characteristics in both groups)
Performing study biasRecall, measurement, procedure, observer-expectancy bias
Recall biasAwareness of disorder alters recall of subjects
In retrospective studiesRecall bias
How to reduce recall biasReduce time from exposure to follow up
Measurement biasGetting information in a way that distorts it
Reduce measurement biasUse standardised method for data collection
Procedure biasGroups are not treated in the same way
How to reduce procedure biasSame as observer-expectancy
Observer-expectancy biasPygmalion effect
Researcher's belief is tx changes the outcomeObserver-expectancy
How to reduce observers expectancy biasBlinding and placebo (lack of awareness)
Recruiting biasSelection bias
Selection biasSampling bias. Unrepresentative sample
Berkson biasHospital population vs general
Healthy worker effectHealthier than general
Non response biasSubjects differ from nonrespondents in meaningful ways.
How to avoid selection biasRandomization
Latent periodExposure to a risk factor sometimes occurs years before clinical manifestations of a disease are present.
Lead time biasWhen a test dx disease at an earlier time than another test, but does't impact the natural course of the disease.
HawthorneStudy subjects change behaviour due to awareness.
ANOVA (analysis of variance)to determine whether there are any differences between the means of 2 or more independent groups.
Null hypothesis rejected when there are at least 2 means that are significantly different from one anotherANOVA
Meta-analysispooling of data from several studies to perform an analysis with greater statistical power than individual studies alone.
Case fatality ratenumber of fatal cases/ total people with condition.
Question Answer
Normal distribution68- 95-99.7
RRR (relative risk reduction)absolute risk control-absolute risk tx/ absolute risk control

Recent badges