To the Editor: Dr Howard and colleagues1 reported a 0.012-mm reduction in carotid intimal media thickness (IMT) among participants randomized to aggressive blood pressure/low-density lipid cholesterol–lowering treatment and a 0.038-mm increase in carotid IMT among participants randomized to standard treatment after 36 months in the Stop Atherosclerosis in Native Diabetics Study (SANDS). Randomization is assumed to equally distribute all known and unknown confounders between experimental groups. However, potential confounders may be unevenly distributed between groups in small trials or within strata of large trials and subsequently induce bias despite randomization.2 Such accidental bias is possible in SANDS because at baseline 82% of participants in the aggressive treatment group concomitantly received oral hypoglycemic medications compared with 73% in the standard treatment group (P = .02). Oral hypoglycemic medications such as metformin and pioglitazone have demonstrated beneficial effects on carotid IMT independent of glycemic control,3,4 and both medications are common components of oral hypoglycemic regimens.5 Therefore, the observed effect for the aggressive therapy group, in which a higher proportion of participants used oral hypoglycemic medications, may be susceptible to a positive bias away from the null. Furthermore, prevalent use of oral hypoglycemic medication may account for the lower than expected progression of subclinical disease within the standard treatment group.
Oral Hypoglycemic Use and the SANDS Trial. JAMA. 2008;300(4):389-390. doi:10.1001/jama.300.4.389-a