EPI II

Question 1

Overmatching

Answer 1

1. The first refers to matching that harms statistical efficiency, such as case-control matching on a variable associated with exposure but not disease.
2. The second refers to matching that harms validity, such as matching on an intermediate between exposure and disease.
3. The third refers to matching that harms cost-efficiency.

Question 2

Active comparator, benefits

Answer 2

1) they provide themost clinically relevant comparisons (i.e., clinicians arenot deciding between no treatment versus treatmentwith a TNF inhibitor),
2) they increase the probability that patients in the 2 exposure groups are similar with respect to unmeasured confounders, such as disease severity, and
3) they allow for greater confidence that the diagnosis of RA is correct (a patient with a diagnosisof RA who is not receiving a DMARD has a lowerprobability of actually having RA compared with apatient who is also receiving a DMARD)”

Question 3

New user design, benefits

Answer 3

1. Deals with problem of depletion of susceptibles
2. Ensuring appropriate confounding adjustment by capturing pretreatment variables
3. Reduces risk of immortal time bias

Question 4

Type 1 error

Type 2 error

Answer 4

Type 1: When we reject the null hypothesis when the opposite is true (false positive)
Type 2: When we fail to reject the null hypothesis when it is true (False negative)

Question 5

Types of missingness

Answer 5

Missing completely at random
Missing at random (de med missing skiljer sig från de med info men man kan beräkna missing med annan info)
Missing not at random (beror på icke-observerade faktorer och går typ inte att hantera)

Question 6

Ways of handling missingness

Answer 6

Last observation carried forward
Mean value imputation
Multiple imputation
Complete case analysis

(Worst observation carried forward)
(Non-responder imputation)

Question 7

Key questions when data are missing

Answer 7

(1) Why are data missing? (2) How do patients with missing and complete data differ? and (3) Do the observed data help predict the missing values?

Question 8

Hills criteria

Answer 8

Strength: Stronger association mor likely to be causal
consistency, across different studies
Temporality,
Biological gradient: dose response
Specificity: The exposure causes a specific outcome
Plausibility: When there is a biological hypothesis
Coherence: Should not contradict what is known about the biology beforhand
Experimental evidence: More plausible if there are experimental studies also
Analogy: Has a similar association been observed with a similar exposure/disease