A retrospective cohort study evaluates the efficacy of a novel oral anticoagulant (NOAC) c… — USMLE Step 2 CK

A retrospective cohort study evaluates the efficacy of a novel oral anticoagulant (NOAC) compared to warfarin for the prevention of ischemic stroke in patients with nonvalvular atrial fibrillation. The investigators analyze the initial data and report a crude relative risk (RR) of 0.70 (95% CI: 0.55-0.88), suggesting that the NOAC is associated with a significantly lower risk of stroke compared to warfarin.

During peer review, concerns are raised that patient age may be distorting the observed association. In response, the investigators perform a stratified analysis, separating the cohort into two groups based on age. The results of the stratified analysis are presented below:

Age Group	Relative Risk (NOAC vs Warfarin)	95% Confidence Interval
$<$ 75 years	0.98	0.85-1.13
$\geq$ 75 years	1.02	0.88-1.18

Which of the following epidemiological phenomena best explains the discrepancy between the crude relative risk and the stratified relative risks?

ConfoundingAnswer
B
Effect modification
C
Ascertainment bias
D
Attrition bias
E
Recall bias

Answer

Confounding

The correct answer is confounding. Confounding is a distortion of the true relationship between an exposure and an outcome by a third variable that is related to both. Stratification is a primary analytical technique used to detect and control for confounding. When a dataset is stratified by a confounding variable, the apparent association seen in the crude analysis disappears or changes significantly, and the relative risks within the individual strata will be similar to one another. In this scenario, the crude RR of 0.70 falsely suggested that NOACs were superior to warfarin. However, once stratified by age, the RRs became approximately 1.0 in both age groups. This indicates that age was a confounder; for instance, younger patients (who inherently have a lower risk of stroke) may have been preferentially prescribed NOACs, creating the illusion of NOAC superiority in the unadjusted crude data.

Step-by-Step Solution

Analyze the crude measure of association.

The crude relative risk is 0.70, indicating a statistically significant protective effect of the NOAC compared to warfarin.

Establishing the baseline finding is necessary to compare it against the stratified results.

Evaluate the stratum-specific measures of association.

The relative risks in both age strata are approximately 1.0 (0.98 and 1.02), indicating no significant difference in stroke risk between NOAC and warfarin within each age group.

Stratification separates the data by the suspected variable to isolate its impact on the association.

Compare the crude and stratified results to identify the epidemiological phenomenon.

Because the stratum-specific relative risks are similar to each other but significantly different from the crude relative risk, age is acting as a confounding variable.

This pattern is the hallmark of confounding, meaning the initial association was driven entirely by the unequal distribution of age between the treatment groups.

Key Concept

In epidemiological studies, confounding is identified when stratification by a third variable reveals stratum-specific measures of association that are similar to each other but differ significantly from the crude measure of association.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic