Risk of hypoglycemic event among patients taking metformin plus insulin versus metformin plus sulfonylurea in the propensity score–matched cohort
| Variable | Metformin + sulfonylurea n = 12 180 | Metformin + insulin n = 2 436 |
|---|---|---|
| Persistent exposure required* | ||
| Composite first hypoglycemic event, no. | 466 | 121 |
| Person-years, no. | 18 922 | 3 919 |
| Unadjusted rate/1000 person-years | 24.6 (22.5–26.9) | 30.9 (25.9–36.8) |
| Unadjusted HR (95% CI) | Reference | 1.26 (1.03–1.54) |
| Adjusted HR† (95% CI) | Reference | 1.30 (1.06–1.59) |
| Composite recurrent hypoglycemic events, no. | 585 | 159 |
| Person-years, no. | 19 593 | 4 070 |
| Unadjusted rate/1000 person-years | 30.0 (27.6–32.3) | 39.1 (33.5–45.5) |
| Unadjusted HR (95% CI) | Reference | 1.31 (1.06–1.63) |
| Adjusted HR† (95% CI) | Reference | 1.39 (1.12–1.72) |
| Persistent exposure not required‡ | ||
| Composite first hypoglycemic event, no. | 1 016 | 261 |
| Person-years, no. | 36 169 | 7 081 |
| Unadjusted rate/1000 person-years | 28.1 (26.4–29.8) | 36.9 (32.7–41.5) |
| Unadjusted HR (95% CI) | Reference | 1.31 (1.15–1.50) |
| Adjusted HR† (95% CI) | Reference | 1.34 (1.17–1.54) |
| Composite recurrent hypoglycemic events, no. | 1 470 | 448 |
| Person-years, no. | 38 248 | 7 654 |
| Unadjusted rate/1000 person-years | 38.4 (36.6–40.4) | 58.5 (53.5–64.0) |
| Unadjusted HR (95% CI) | Reference | 1.52 (1.27–1.83) |
| Adjusted HR† (95% CI) | Reference | 1.58 (1.32–1.89) |
Note: CI = confidence interval, HR = hazard ratio.
↵* Primary analysis requires persistence on metformin; patients are censored after 90 days without metformin in hand.
↵† Adjusted HR is derived from Cox proportional hazards model for time to outcome for matched cohort, adjusted for all baseline covariates included in propensity score model except network of care.
↵‡ Sensitivity analysis does not require persistence on regimen or refills within 90 days. Patients can add another regimen or stop.