CAT Paper Flashcards
Clearly focussed?
PICO/PEO
Appropriate?
Design
Acceptable recruitment?
Co CC SR
Cohort - represent, selection bias, all included who should
Case-control - both represent pop
SR - all papers
Acceptable Assignment?
RCT
RCT - how, random, concealed
Accurate measurement?
Co CC
Cohort - exposure and outcome
Case-control - exposure
Follow-up?
Cohort - complete/ long enough
Case-control - treated equal
SR - reasonable to combine
RCT - exclude/ lost (accounted for, explained, ITT)
Confounding?
Co CC
ID and accounted for, list important and missing
Sound method?
RCT
Blinding, clearly defined and similar groups, clearly defined and similar care
Results?
Co
Bottom line
Rate between
Strength of association (RR)
ARR
CI
Size of effect
Chance/ bias/ confounding
Flaw in design/ method
Bradford-Hills
Results?
CC
Bottom line
Appropriate analysis
Strength of association (OR)
Adjusted for confounding (still explain, big OR change?)
Size of p
CI
All variables considered
Effect of refusal (towards null)
Size of effect
Chance/ bias/ confounding
Flaw in design/ method
Bradford-Hills
Results?
SR
Bottom Line
Expressed how (NNT, OR)
CI
Results?
RCT
(IMPRESSED)
Comprehensive reporting?
- Identify bias
- Missing/ incomplete data
- Power + P-values
- Results all reported
- Expressed how
- (clearly) Specified outcome
- (which) Stats tests
- Effect (absolute and relative)
- Differential drop-out (effect)
CI
Size of effect
Harms/ unintended reported
Cost-effectiveness analysis
Significantly different pop/ setting
All
Study design
Co RCT
Co - appropriate for my Q
RCT - limitations
Important outcomes not studied
SR RCT
Local harm: benefit
Cohort
Case-control
SR - talked about? Your opinon?
RCT - better than existing? resources needed?
Implications
Co CC
Cohort - fit with other evidence, one not robust enough
Case-control - fit with other evidence
Cohort - measurement of exposure
Measurement or classification bias
- subjective or objective
- validated (measure what you want)
- same procedure to classify into groups
Cohort - measurement of outcome
Measurement or classification bias
- subjective or objective
- validated (measure what you want)
- reliable system to detect all cases
- similar measure method in both groups
- subjects/ assessor blinded to exposure
Cohort - follow-up
long enough for outcome to happen
effect of drop outs
link between people leaving
