Purpose: To measure the extent to which decision aids enable involvement in decision-making of patients with diabetes considering the use of statins to lower cardiovascular risk. Method(s): Sixteen diabetologists and 98 patients with type 2 diabetes participated in the trial. Diabetologists were randomly allocated to decision aid or educational pamphlet (control); patients received the information prior to the visit (from research personnel) or during the visit (from the clinician). We designed the decision aid Statin Choice for interactive use with a health professional. Each visit was videotaped and two researchers (one a clinician, one a qualitative researcher) observed each video and coded the interaction using the OPTION scale, a validated tool to measure the extent to which physicians involve patients in the decision making process. We scored both the statin (aided) and the glucose control (unaided) discussions. Generalized estimating equations assessed the strength of association of the intervention (decision aid vs. control) and method of delivery (during visit with physician or prior to visit with researcher) with the OPTION score for both discussions, while accounting for clustering among providers. Results: Physicians using Statin Choice achieved, on average, 9.9 (95% CI 6.2, 13.5) more points on the OPTION scale than the control encounters. Encounters with patients who had received the decision aid prior to the visit did not have greater OPTION scores than control encounters. All encounters scored similarly for the glucose control decision with physicians using Statin Choice achieved, on average 2.1 (95% CI -0.54, 4.7) more points on the OPTION scale than the control encounters. Conclusions: When physicians use a decision aid during a visit, they are much more likely to involve the patient in the decision making process. However the decision aid only affects the decision it aids. Whether this effect is related to the content of the decision aid or to the way its use affects patient-clinician interaction remains unclear.