Patient Disposition and Baseline Characteristics
Overall, 814 patients on OAM treatment were randomized to receive BI (n = 410) and MMI (n = 404), of whom 663 patients completed the 48-week trial. Patient disposition is shown in Fig. 1. Baseline characteristics were similar for the treatment groups and have been presented previously [14]. Most of the patients in both groups had poorly controlled diabetes with a mean (± standard deviation, SD) HbA1c of 9.8 ± 1.61%. The mean (± SD) duration of diabetes was 9.4 ± 5.77 years, the mean (± SD) FPG was 11.4 ± 3.47 mmol/L, and the mean (± SD) PPG was 15.8 ± 4.50 mmol/L. The gender distribution was similar in both groups, with 233 (56.80%) and 223 (55.20%) male patients in the BI and MMI groups, respectively.
Glycemic Control (Baseline to Week 48)
As reported previously, both the BI and MMI groups demonstrated good glycemic control at week 24 [14]. The least-squares (LS) mean standard error (SE) of the change in HbA1c from baseline to week 24 was similar between the MMI and BI groups (MMI: − 2.15 [0.07] vs. BI − 2.00 [0.07]; LS mean difference [95% CI]: − 0.15 [− 0.34, 0.04]; P = 0.13). With an extended 6 months of treatment, however, a statistically greater decrease in HbA1c at week 48 was observed in the MMI group (MMI: − 2.03 [0.06] vs. BI: − 1.82 [0.06]; LS mean difference [95% CI]: − 0.21 [− 0.39, − 0.03]; P < 0.05) (Fig. 2) than in the BI group. Similar to the 24-week results, there was a greater reduction in FPG in the BI group than in the MMI group from baseline to week 48 (LS mean [SE]: MMI: − 2.53 [0.14] vs. BI: − 3.19 [0.14]; P < 0.01), and the reduction in PPG post breakfast was similar in both groups with no significant difference between the groups (LS mean [SE]: MMI: − 4.35 [0.22] vs. BI: − 4.33 [0.23], P > 0.05). The proportion of patients who achieved HbA1c < 7% at week 48 was comparable between the groups (MMI: 23.3% vs. BI: 23.4%, P > 0.05) and was lower compared with week 24 [14].
The changes in these effectiveness parameters after the additional 6 months of treatment showed that there was better glucose control in the MMI group and a limited change in insulin dose from week 24 to week 48. The mean (SD) dose of BI was 14.6 (7.07) U/day at week 24, with minimal increments to 15.4 (7.09) U/day at week 48. The dose of MMI reduced from 24.8 (10.25) U/day at week 24 to 24.2 (11.24) U/day at week 48. There were similar changes in body weight (kg) at week 24 and week 48 in both groups. Table 1 presents the changes in HbA1c, FPG, PPG, and body weight from baseline to week 48 and the daily insulin doses at baseline and week 48.
Table 1 Changes in glycated hemoglobin, fasting plasma glucose, postprandial glucose, and body weight from baseline to week 48 and the daily insulin dose at baseline and week 48 A significant intergroup difference in OAM use was detected (P < 0.0001): more patients increased their OAM use in the BI group whereas decreased their OAM use in the MMI group through week 48 (Table 2). The BI group demonstrated increased use of alpha-glucosidase inhibitors, glinides, and insulin secretagogues compared with the MMI group (Table 3). The proportions of patients who discontinued, switched, or intensified insulin treatment through week 48 were comparable between both the groups (P = 0.35, Table S1 in the Supplementary Information).
Table 2 Oral antihyperglycemic medication usage at baseline and week 48 Table 3 Number of patients using different categories of oral antihyperglycemic medications at week 48 Safety Outcomes
As in the 24-week analysis [14], both BI and MMI showed no major safety issues at week 48. The overall safety findings were comparable between the groups (Table S2 in the Supplementary Information). No severe hypoglycemia event was reported over the 48 weeks of treatment in either group. The incidence of total hypoglycemia was higher in the MMI group compared with the BI group (MMI: 124% [30.7] vs. BI: 76% [18.5], P < 0.0001), while the incidence of noctural hypoglycemia was similar between the groups (MMI: 42% [10.4] vs. BI: 38% [9.3], P = 0.64) (Table S3 in the Supplementary Information).
Cost Outcomes
A total of 590 patients (BI: 289, MMI: 301) were included in the cost analysis. During the 48-week treatment period, the mean daily total medication cost per patient per day was 38.5% higher in the BI group compared with the MMI group (mean [SD], BI: 17.81 [7.07] CNY vs. MMI: 12.86 [5.73] CNY; P < 0.0001). Similarly, the mean daily OAM cost per patient per day was 51.0% higher (mean [SD], BI: 9.71 [6.32] CNY vs. MMI: 6.43 [5.16] CNY; P < 0.0001) and the daily insulin cost per patient per day was 25.8% higher (mean [SD], BI: 8.10 [2.95] CNY vs. MMI: 6.44 [2.46] CNY; P < 0.0001) in the BI group.
The total medication cost of a 1% HbA1c reduction was 52.9% higher in the BI group (BI: 3376 CNY vs. MMI: 2208 CNY). Similarly, for each patient who achieved the HbA1c target of < 7%, the cost was 38.6% higher in the BI group (BI: 27,010 CNY vs. MMI: 19,484 CNY).