To rule out the possibility that the difference in the annualized change in the mean IMT between the two groups could be attributed to the slightly higher shed rate in the Tong Xin Luo capsules group (17.8%) than in the placebo group (15.9%), we performed a sensitivity analysis of the primary outcome using two methods: (1) Assuming random data missing, multiple interpolation method was used to obtain the mean difference of the annual change of the mean IMT between the two groups as -0.01693 (95%CI, -0.02155 to -0.01230, p 0.001); (2) Assuming that the data are not lost at random, the mean difference in the annualized change in the mean IMT between the two groups is −0.01663 (95%CI, −0.02126 to −0.01200, p 0.001) using the model mixing model. Therefore, the results of the two sensitivity analyses were consistent with the primary outcome analysis, that is, there was no evidence of bias due to differences in missing data between the two groups.

Carotid plaque area

Relative to baseline measurements, the mean LS of the maximum plaque area in the long-axis view of the bilateral carotid arteries decreased by 0.513 (95%CI:-1.399 to 0.373) at 24 months in the Tong Xin Luo capsules group, but increased by 1.671 (95%CI:0.793 to 2.548) in the placebo group (Table 2). From baseline to 24 months, there was a significant difference in long-axis plaque area between the two groups (P = 0.003, Figure 3). Similarly, compared with baseline values, the mean LS of the maximum plaque area in the bilateral carotid artery braxial view decreased by 0.234 (95%CI:-1.289 to 0.821) at 24 months in the Tong Xin Luo capsules group, but increased by 3.031 (95%CI:1.995 to 4.066) in the placebo group (Table 2). In the short-axis view, there were also significant differences in patch area change from baseline to 24 months between the two groups