Examkin

Question

Difficulty: EasyBias and Confounding in Research

A large population-based study investigates the relationship between daily moderate alcohol consumption and the incidence of lung cancer. The unadjusted analysis shows a relative risk of 2.1 (95% CI: 1.5-2.8) for lung cancer in individuals who consume alcohol daily compared to those who do not. The investigators subsequently stratify the participants based on their history of tobacco smoking. In the stratified analysis, the relative risk of lung cancer is 1.0 (95% CI: 0.8-1.2) among active smokers and 1.0 (95% CI: 0.7-1.3) among non-smokers. Which of the following phenomena best explains the results of the initial unadjusted analysis?

  1. ConfoundingAnswer
  2. B
    Effect modification
  3. C
    Recall bias
  4. D
    Selection bias
  5. E
    Berkson bias

Answer

Confounding
The scenario describes confounding, which occurs when a third variable (in this case, smoking) is associated with both the exposure (alcohol consumption) and the outcome (lung cancer), but is not on the causal pathway. Because individuals who drink alcohol are also more likely to smoke tobacco, the unadjusted analysis falsely attributes the increased risk of lung cancer to alcohol. Stratification removes this effect, revealing that alcohol consumption itself does not actually increase the risk of lung cancer (RR = 1.0) in either group.

Step-by-Step Solution

1
Analyze the unadjusted (crude) relative risk.
The initial relative risk is 2.1, indicating a statistically significant association between alcohol consumption and lung cancer.
This establishes the baseline finding before controlling for other potential variables.
2
Examine the stratified relative risks.
When separated by smoking status, the relative risk becomes 1.0 for both smokers and non-smokers.
Stratification isolates the data by a potential third variable to observe the true exposure-disease relationship within homogeneous groups.
3
Compare the unadjusted and stratified results to identify the epidemiological phenomenon.
The association disappears in both strata, meaning the initial association was entirely driven by the third variable (smoking).
When stratified measures of association are similar to each other but significantly different from the crude measure, the third variable is a confounder.

Key Concept

Identifying Confounding via Stratification
Rate this question

Topics

Bias and Confounding in Research

More questions from this topic