Week 3 Flashcards
performance bias
systematic differences in the care provided to the comparison groups other than the intervention being tested
reducing performance bias
blinding
detection bias
outcome assessment differs systematically between comparison groups
reducing detection bias
blinding of outcome assessors
measurement bias
systematic error in the measurement of information on the exposure or outcome
measurement bias aka
information bias and observation bias
misclassification
incomplete medical records, recording errors, misinterpretation of records, errors in the questionnaire including recall bias by the participant
non differential misclassification
information is incorrect, but same across the two groups
underestimates the strength of association between exposure and disease
differential misclassifaction
information errors differ between the two groups
either a decrease or increase in the perceived association between exposure and outcome
reduce measurement bias
standardised research protocols, obtaining information from multiple sources, blinding of researchers and staff to the exposure/disease status of the participant, training of interviewers
reduce recall/response bias
defining research question carefully
devise high quality questionnaires
highly trained research staff
attrition bias
loses to follow up or dropouts
intention to treat analysis (ITT)
analysing all participants recruited from the randomisation process even if they have not completed the study
done to avoid the effects of crossover and dropout
methods for dealing with missing data
baseline values or last outcome measured by participant carries over
sensitivity analysis
mixed models
worst and best cases scenarios
reducing attrition bias
create project identity
good communication between staff and participants
keep follow up interviews brief
accessibility to clinics
per-protocol analysis
only data for participants who complete the study are analysed
as treated analysis
middle grounds where non adhering participants moved to control group
negative confounding
confounding variable biases towards the null hypothesis
positive confounding
confounding variable biases away from null hypothesis
Reduce confounding
limit participation of certain subgroup who share the confounding factor
analysis of the data through stratification
filter for certain groups
matching in case control studies
randomise individuals to different groups
how much attrition is too much (attrition bias)
less than 5% leads to little bias
more than 20% threatens study validity
probability can be defined by
relative frequency
parameter
unknown characteristic of interest in the true population
sample result
estimated value of the parameter based on a random sample taken from the study population
descriptive epidemiology
hypothesis generation; person, place, time
analytical epidemiology
hypothesis testing; study design, methods
aetiology
cause of a disease
measured score
true’ score +/- Error
bias vs confounding
Bias creates an association that is not true whereas confounding describes an association that is true, but potentially misleading; potential to correct for confounding once study completed
test-retest reliability
test results are consistent over time by administering the same test twice over a period of time to same group of individuals
inter-rater reliability
assess the degree to which different judges or raters agree in their assessment decision
construct validity
does test measure what is claims to be measuring
negatively skewed
mean < median < Mode
positively skewed
mean>median>mode
controlling confounders in design stage
randomisation
restriction
matching
controlling confounders in analysis stage
stratification
statistical modelling