Examkin

Question

Difficulty: Very hardMeasures of Disease Frequency and Association

Researchers are evaluating the association between a novel oral anticoagulant (NOAC) and the development of drug-induced liver injury (DILI). They query a large national health registry and identify 200 patients who have been diagnosed with DILI (cases) and 800 patients without DILI (controls). Upon reviewing pharmacy records, they ascertain the prior use of the NOAC in both groups. The findings are summarized in the table below:

Exposure StatusDILI (Cases)No DILI (Controls)
NOAC prescribed50100
No NOAC prescribed150700

During a secondary analysis, the researchers stratify the participant data by the concurrent use of a cytochrome P450 3A4 (CYP3A4) inhibitor. After stratification, the measure of association is found to be 1.05 in patients taking CYP3A4 inhibitors and 1.10 in patients not taking CYP3A4 inhibitors.

Based on the appropriate calculation for this specific study design, which of the following represents the correct measure of association and the most accurate interpretation of the stratified analysis?

  1. Odds ratio of 2.33; the concurrent use of a CYP3A4 inhibitor is a confounding variable.Answer
  2. B
    Relative risk of 1.89; the concurrent use of a CYP3A4 inhibitor is a confounding variable.
  3. C
    Odds ratio of 2.33; the concurrent use of a CYP3A4 inhibitor is an effect modifier.
  4. D
    Relative risk of 1.89; the concurrent use of a CYP3A4 inhibitor is an effect modifier.
  5. E
    Odds ratio of 2.33; the positive predictive value of NOAC use for predicting DILI would increase if the prevalence of DILI decreases.

Answer

Odds ratio of 2.33; the concurrent use of a CYP3A4 inhibitor is a confounding variable.
This study identifies patients based on the presence or absence of an outcome (DILI) and retrospectively examines their exposure history (NOAC use). This defines a case-control study. The appropriate measure of association for a case-control study is the odds ratio (OR), calculated using the cross-product ratio of the 2x2 contingency table: OR=a×db×cOR = \frac{a \times d}{b \times c}. Using the provided data, OR=50×700150×100=3500015000=2.33OR = \frac{50 \times 700}{150 \times 100} = \frac{35000}{15000} = 2.33.

In the secondary analysis, the crude OR (2.33) is significantly different from the stratified ORs (1.05 and 1.10). Because the stratified ORs are very similar to each other but differ from the crude OR, this pattern indicates that the concurrent use of a CYP3A4 inhibitor is a confounding variable, artificially inflating the crude measure of association. If CYP3A4 inhibitor use were an effect modifier, the stratified ORs would be significantly different from each other.

Step-by-Step Solution

1
Identify the study design based on how participants were selected.
Participants were selected based on their outcome (DILI vs. no DILI), making this a case-control study.
Determining the study design dictates which measure of association can be appropriately calculated.
2
Select and calculate the appropriate measure of association.
The correct measure is the Odds Ratio (OR). OR=50×700150×100=2.33OR = \frac{50 \times 700}{150 \times 100} = 2.33.
Relative risk cannot be calculated in case-control studies because the investigator artificially fixes the number of cases and controls, so true incidence rates are unknown.
3
Interpret the results of the stratified analysis.
The crude OR is 2.33, while the stratified ORs are 1.05 and 1.10. The variable is a confounding factor.
When a crude measure of association is significantly different from the stratified measures, and the stratified measures are similar to each other, it indicates confounding. If the stratified measures were significantly different from each other, it would indicate effect modification.

Key Concept

In case-control studies, the odds ratio (OR) is the appropriate measure of association. Confounding is demonstrated when a crude measure of association differs significantly from stratified measures that are similar to one another.
Estimated Time:2m 0s
Rate this question

Topics

Measures of Disease Frequency and Association

More questions from this topic