Examkin

Question

Difficulty: HardBias and Confounding in Research

An epidemiological investigation evaluates the risk of severe hepatotoxicity associated with a newly approved antifungal agent compared to standard therapy. The research team collects data from multiple national registries over a two-year period. Initial analysis reveals a crude relative risk (RR) of 1.91.9 (95% CI: 1.4−2.51.4-2.5) for hepatotoxicity in patients receiving the new agent.

Suspecting a third variable may be influencing these results, the analysts stratify the cohort by the presence of underlying non-alcoholic fatty liver disease (NAFLD). The stratified analysis yields the following results:

Patient SubgroupRelative Risk (95% CI)
With NAFLD3.83.8 (2.9−4.92.9-4.9)
Without NAFLD1.11.1 (0.8−1.40.8-1.4)

Based on these findings, which of the following is the most appropriate statistical or methodological action for the investigators?

  1. Present the stratum-specific relative risks independently rather than calculating a pooled estimateAnswer
  2. B
    Utilize a multivariable regression model to adjust the final relative risk for NAFLD status
  3. C
    Redesign the investigation as a case-control study to eliminate the disproportionate outcome distribution
  4. D
    Conclude that NAFLD is a positive confounder and report the crude relative risk as the primary outcome
  5. E
    Apply a restriction protocol to exclude patients without NAFLD in subsequent analyses to increase internal validity

Answer

Present the stratum-specific relative risks independently rather than calculating a pooled estimate
The data provided in the table clearly demonstrates effect modification (also known as interaction). This phenomenon occurs when the effect of an exposure on an outcome varies significantly depending on the presence or level of a third variable. In this scenario, the risk of hepatotoxicity from the antifungal is greatly increased in patients with NAFLD (RR=3.8RR = 3.8) but is negligible in those without it (RR=1.1RR = 1.1). Because the stratified relative risks differ significantly from one another, NAFLD is an effect modifier, not a confounder. (If NAFLD were a confounder, the stratified relative risks would be similar to each other, but distinct from the crude relative risk). The standard epidemiological rule is that confounding is a distortion that should be eliminated via adjustment (e.g., multivariable regression), whereas effect modification is a biological reality that should be described. Therefore, the most appropriate action is to report the stratum-specific risk estimates independently.

Step-by-Step Solution

1
Analyze the relationship between the crude and stratified relative risks.
The crude RR is 1.9, while the stratified RRs are 3.8 and 1.1.
Comparing crude and stratified estimates is the primary method for identifying the influence of a third variable.
2
Evaluate the stratified estimates against one another.
The RR for patients with NAFLD (3.8) is significantly different from the RR for patients without NAFLD (1.1), and their confidence intervals do not overlap.
If the estimates differ significantly across strata, the third variable is interacting with the exposure.
3
Identify the specific epidemiological concept.
The data demonstrates effect modification (interaction), not confounding.
Effect modification occurs when the magnitude of an exposure's effect varies depending on the level of a third variable.
4
Determine the standard methodological protocol for handling this concept.
The investigators must report the stratum-specific estimates separately.
Unlike confounding, which is a nuisance to be adjusted for, effect modification is a biological finding that should be highlighted by presenting the distinct risks for each subgroup.

Key Concept

Distinguishing effect modification from confounding through stratified analysis, and applying the correct reporting protocol.
Estimated Time:1m 30s
Rate this question

Topics

Bias and Confounding in Research

More questions from this topic