Descriptive statistics of patients on T2DM second-line drug classes for the BMI outcome
| Drug class | Patients (n) | Treatment change* | Missing outcome* | Average age† | % Female* |
| Sulfonylurea | 18 170 | 2967 (16%, 2e−109) | 8611 (47%, 2e−16) | 60.9 (4e−17) | 48.4% (0.01) |
| Thiazolidinedione | 2691 | 640 (23.8%, 3e−6) | 1503 (56%, 2e−29) | 59.2 (7e−05) | 49.3% (–) |
| Glucagon-like peptide-1 receptor agonists | 1172 | 293 (25%, 5e−5) | 441 (38%, 3e−8) | 52.9 (3e−86) | 66.6% (7e−34) |
| Dipeptidyl peptidase 4 | 6295 | 1852 (29%, 3e−92) | 2352 (37%, 2e−49) | 58.7 (2e−29) | 50.9% (0.003) |
Per-confounder statistics appear in online supplementary table 3.
*Proportion test. Missing entries (–) are not significant with FDR <0.05.
†Wilcoxon rank-sum test. Missing entries (–) are not significant with FDR <0.05.
BMI, body mass index; FDR, false discovery rate; T2DM, type 2 diabetes mellitus.