A randomized controlled trial comparing two antihypertensive agents (amlodipine vs. lisinopril) enrolls 300 patients with Stage 2 hypertension. Patients are randomized 1:1 using a computer-generated sequence with stratification by age and baseline systolic BP. Baseline characteristics show appropriate balance in age (mean 58 years), sex (55% male), and BP (156/94 mmHg in both groups). However, at the end-of-study analysis, the amlodipine group has a significantly higher prevalence of diabetes (45% vs. 25%, p=0.03) and elevated HbA1c (7.8% vs. 6.9%, p=0.02), which were not documented as baseline differences. The amlodipine group also has numerically higher diastolic BP reduction (12 mmHg vs. 8 mmHg), but the difference is not statistically significant (p=0.08). Which of the following best explains why the observed difference in diabetes prevalence limits the internal validity of the trial?

1. A)Differential loss to follow-up between treatment groups compromised the randomization process
2. B)Uncontrolled confounding by diabetes status may bias the estimate of treatment effect on blood pressure outcomesGABARITO
3. C)The investigators failed to measure HbA1c and diabetes status at baseline, resulting in detection bias
4. D)Measurement error in fasting glucose levels was more common in the amlodipine group due to protocol violations
5. E)Regression to the mean in the amlodipine group led to spurious diabetes classification