Examkin

Question

Difficulty: HardBias and Confounding in Research

A prospective cohort study is conducted to evaluate the risk of incident hepatocellular carcinoma (HCC) in patients treated with a novel antiviral medication compared to standard therapy for chronic hepatitis C infection. The unadjusted (crude) relative risk (RR) for HCC in the novel medication group is 1.61.6 (95%95\% CI: 1.2−2.11.2-2.1).

The investigators suspect that baseline liver cirrhosis may be influencing the results. They perform a stratified analysis based on the presence or absence of cirrhosis at the time of treatment initiation. The stratum-specific results are as follows:

Baseline StatusRelative Risk95% Confidence Interval
Cirrhosis present2.82.82.1−3.72.1 - 3.7
Cirrhosis absent0.90.90.7−1.20.7 - 1.2

Based on these data, which of the following is the most appropriate next step for the investigators to report their findings?

  1. Report the stratum-specific relative risks separatelyAnswer
  2. B
    Perform a multivariable regression to calculate an adjusted relative risk
  3. C
    Utilize propensity score matching to control for baseline differences
  4. D
    Conclude that the novel medication independently increases the risk of HCC
  5. E
    Recommend a crossover study design to eliminate the influence of cirrhosis

Answer

The investigators should report the stratum-specific relative risks separately because the data demonstrates effect modification.
The data presented demonstrates effect modification. We know this because the relative risk for hepatocellular carcinoma varies significantly based on the presence or absence of cirrhosis (2.82.8 vs 0.90.9). When effect modification is present, pooling the data into a single adjusted estimate obscures a critical biological interaction. Therefore, the standard and most appropriate epidemiological practice is to report the distinct findings for each stratum separately.

Step-by-Step Solution

1
Analyze the crude relative risk compared to the stratum-specific relative risks.
The crude RR (1.61.6) is significantly different from the stratum-specific RRs (2.82.8 and 0.90.9). Furthermore, the two stratum-specific RRs are significantly different from each other.
Comparing unadjusted and adjusted/stratified data is the primary method to distinguish between confounding and effect modification.
2
Identify the epidemiological phenomenon occurring based on the numeric patterns.
The pattern indicates effect modification (interaction).
In effect modification, the magnitude of the exposure's effect varies depending on the level of a third variable. In confounding, the stratum-specific RRs would be similar to each other but different from the crude RR.
3
Determine the appropriate statistical handling and reporting method for this phenomenon.
The stratum-specific relative risks must be reported separately.
Effect modification represents a true biological interaction, not a flaw in the study design. Adjusting or pooling the data would hide this important clinical information.

Key Concept

Effect modification occurs when the magnitude of an exposure's effect on an outcome differs depending on the level of a third variable. Unlike confounding, it is not a statistical bias to be adjusted for, but a biological finding that must be reported using stratum-specific estimates.
Estimated Time:1m 30s
Rate this question

Topics

Bias and Confounding in Research

More questions from this topic