Stuff for 1st Test that no one taught us from the E-Book Flashcards
(15 cards)
Self-selection (volunteer bias)
People who are at greater risk for outcome are more likely to participate
Healthy user bias
Healthier individuals enroll in a study and are less likely to develoop out come of interest
Nonresponse bias
Responders to surveys differ from nonresponders
Interviewer and surveillance bias
Interviewer - different data collection methods results in more aggressive acquisitoon of one group
Surv - more frwquent follow up or diagnositc
Internal validity
Degree to which the results of a study are true for the pop being studied
Chance and bias will impact
Exteneral validity
Generalizability
Degree to which results are true for braoder population
Case control study types of bias of note
More have some degree of selection bias
Ideal for rare conditions or long-term outcomes
Cohort study types of bias
Lower selection bias
UNqiue is surveillance bias
Loss to follow up bias also possible
Type 1
Type 2
Power
Type 1 - Falsely claiming a difference
Type 2 - Falsey claiming the groups are the same
Power - Probability of detecting a truly effective interventione
100%-beta
Allocation concealment
Investigator in charge should be unaware of assignment selection
Simple
Block
Stratified Block design
Coin flip or number generator
Predetermind size is randomized to groups
Patients randomized within subgroups
Cointervention and ascertainment bias
Provision of addtional therapy to one group bc of assignment
MIsinterpret data based on knowledge of the assignment
Meta-analysis vs. systematic reveiw
Sys - selection of studies and summary of individual study results
Meta - extraction of individual study data…pooled average result across all stides…uses a forest plot
Q-test
Heterogeneity test
Minimmal I2 = 25
Moderate =50
Substantial = 75
Publication bias and funnel plots
Positive studies more likely to be published than not
Funnel used to assess for publication bias
