A 52-year-old man with type 2 diabetes mellitus is enrolled in a randomized controlled trial comparing a new antidiabetic agent (Drug X) versus metformin for glycemic control. He is randomly assigned to the Drug X arm. At baseline, his HbA1c is 8.2%, fasting blood glucose 156 mg/dL, and BMI 28.5 kg/m². After 12 weeks of treatment, his HbA1c drops to 7.1%, and he reports excellent adherence. However, the trial coordinator notes that 18% of participants in the metformin arm discontinued the study due to gastrointestinal side effects, while only 2% in the Drug X arm withdrew. Which of the following best describes the primary threat to the validity of this trial's findings?

Explanation

## Identification of Bias Type The scenario describes a **differential dropout rate** between the two treatment arms—18% in the metformin group versus 2% in the Drug X group. This is a classic example of **attrition bias** (also called loss-to-follow-up bias), a form of selection bias that occurs *after* randomization. ## Why This Threatens Validity **Key Point:** Attrition bias occurs when participants drop out *differentially* between arms, and the reasons for dropout are related to the outcome or the treatment effect. In this trial: - Metformin's GI side effects caused selective dropout of those who were intolerant. - This removes from the metformin arm participants who might have had worse outcomes (or who were at higher risk of non-response). - The remaining metformin participants may be a healthier, more compliant subset. - This artificially narrows the apparent difference between Drug X and metformin, or inflates Drug X's apparent benefit. ## Impact on Internal Validity Attrition bias compromises the **internal validity** of the trial because: 1. The groups are no longer comparable despite initial randomization. 2. The intent-to-treat (ITT) analysis may be undermined if many participants are excluded. 3. The per-protocol analysis becomes biased because dropout is non-random. **High-Yield:** The hallmark of attrition bias is *differential* loss across arms—if dropout rates were equal and random, this would be less problematic. ## Prevention Strategy **Clinical Pearl:** To minimize attrition bias, trials should: - Conduct intent-to-treat analysis (analyze all randomized participants in their assigned groups, regardless of adherence or dropout). - Perform sensitivity analyses on completers vs. dropouts. - Report reasons for withdrawal by arm. - Use strategies to maximize retention (frequent follow-up, incentives, flexible visit scheduling).