Propensity

Question 1

Propensity scores (PS) and Traditional methods (TM) and residual confounding ... Which is/are TRUE?

Accepted Answer

TM: Unknown or unmeasured factors may yield to residual confounding, PS: Unknown or unmeasured factors may yield to residual confounding

Question 2

Examples of ATE and ATT... Which is/are TRUE? (ALL TRUE!!)

Accepted Answer

ATT: intervention potentially harmful (ex: drug abuse), ATT: time-intensive rehabilitation programmes, ATE: Medications and drug therapies, ATE: Surgery

Question 3

The best method/s in a randomised study to eliminate confounders ... Which is TRUE?

Accepted Answer

Randomisation

Question 4

Between groups balance in a 2-groups observational design should be addressed by... Which is TRUE?

Accepted Answer

standardized diferences

Question 5

The targeted balance should include... Which is/are TRUE? (two responses)

Accepted Answer

Ratio of variances ≈ 1, ≤ 10% for the absolute standadised difefrences

Question 6

Traditional methods for handling confounders... Which is/are TRUE? (mark "a")

Accepted Answer

a) Include: Adjusting, Stratification, Matching and Restriction

Question 7

Which is/are TRUE? (several options are valid)

Accepted Answer

Collinearity may impair traditional adjusting modeling, Collinearity does not impact the propensity model, Traditional methods always requiere the outcome, Propensity score can be blinded to outcome

Question 8

In this study by Leisman (Crit Care Med. 2019;47:176-185) we can conclude that: (several responses are valid)

Accepted Answer

There is a large amount of unmatched patients (large number of exclusions), External validity may be compromised because of minimal overlapping, There are too many "deterministic" (always or never) patients, Despite potential balance, generalizarion of results is difficult

Question 9

Matching and stratification for traditional methods... Which is/are TRUE?

Accepted Answer

Covariates should be categorical, Limited number of confounders and strata at the same time

Question 10

For IPTW typical formulas of unstabilised weights are: (BOTH TRUE!!)

Accepted Answer

ATE: T/PS + (1 − T)/(1 − PS), ATT: T + PS(1−T) / (1−PS)

Question 11

Propensity scores... Which is/are TRUE?

Accepted Answer

Stratification and adjusting propensity are normally not recommended

Question 12

Hidden and unknown confounders may be handled by... Which is TRUE?

Accepted Answer

None is true

Question 13

Average Treatment (AT) estimand... Which is/are TRUE? (ALL TRUE!!)

Accepted Answer

ATE: effect of moving an entire population from untreated to treated, Average Treatment on the Treated (ATT) is normally addressed by Matching, ATT: effect on those subjects who ultimately received the treatment, Average Treatment Effect (ATE) is normally addressed by IPTW

Propensity

Questions 13

Propensity

Questions 13