Examkin

Question

Difficulty: MediumBias and Confounding in Research

An observational study investigates the association between the regular use of a newly approved prescription sleep medication and the incidence of daytime motor vehicle collisions. Unadjusted analysis reveals a significant positive association (RR = 2.4, 95% CI 1.7-3.1). During peer review, a reviewer expresses concern that underlying obstructive sleep apnea (OSA) is distorting the results. OSA independently increases the risk of motor vehicle collisions and is strongly associated with a higher likelihood of being prescribed sleep medications. Which of the following epidemiological methods are appropriate to control for the influence of OSA in this study? (Select all that apply)

  1. Incorporating obstructive sleep apnea severity as a covariate in a multivariable regression modelAnswer
  2. Restricting study enrollment exclusively to patients who do not have a diagnosis of obstructive sleep apneaAnswer
  3. C
    Reporting the relative risks separately for each obstructive sleep apnea stratum without calculating a pooled overall estimate
  4. D
    Blinding the outcome assessors to the patients' underlying obstructive sleep apnea status during data collection
  5. E
    Increasing the study's sample size to naturally randomize the distribution of obstructive sleep apnea between the groups

Answer

The correct methods to control for confounding in this scenario are incorporating OSA severity as a covariate in a multivariable regression model, and restricting study enrollment exclusively to patients without an OSA diagnosis.
Confounding occurs when a third variable is associated with both the exposure and the outcome, distorting their true relationship. It can be controlled during the study's design phase through restriction (limiting participants to a single category of the confounder) or matching. It can also be controlled during the analysis phase using multivariable regression (which mathematically adjusts for covariates) or stratification with data pooling.

Step-by-Step Solution

1
Identify the epidemiological phenomenon caused by the third variable.
The scenario describes obstructive sleep apnea (OSA) as being independently associated with both the exposure (sleep medication) and the outcome (collisions), causing a distortion of the overall results. This matches the definition of confounding.
Correctly identifying the problem as confounding rather than effect modification or bias determines the appropriate methodological solutions.
2
Recall the valid methods for controlling confounding in observational studies.
Confounding can be controlled during the design stage (via restriction or matching) or during the analysis stage (via stratification with data pooling, or multivariable regression).
The question asks for appropriate methods to control for the identified variable.
3
Evaluate the provided options against the known methods for controlling confounding.
Restriction (limiting the study to patients without OSA) and multivariable regression (adjusting for OSA as a covariate) are correct methods. Unpooled stratification addresses effect modification, blinding addresses observer bias, and increasing sample size only improves precision.
This process of elimination isolates the specific strategies that successfully neutralize confounding.

Key Concept

Confounding can be controlled at the design stage via restriction, matching, or randomization, and at the analysis stage via stratification (with pooling) or multivariable regression.
Rate this question

Topics

Bias and Confounding in Research

More questions from this topic