Table 2

Statistical power, patient characteristics, select diagnostics, and control outcomes

Noninferiority power analysis in health care database study
 No. patients after 1:1 PSM48,262
 Risk of primary end point per 1,000 patients20.8
 Relative risk assumed by RCT based on equipoise1
 α (two-sided)0.05
 Noninferiority margin1.3
 Statistical power, %98.6
Selected patient characteristics from CAROLINA trial and from 1:1 PSM health care database study and balance statistics
CAROLINA trial (N = 6,041)Health care database study (N = 48,262)
Linagliptin (n = 24,131)Glimepiride (n = 24,131)aSt. Diff.*
 Patient characteristics
  Age, mean (SD), years64.0 (9.5)70.40 (7.76)70.42 (7.71)0.00
  Male sex, n (%)3,622 (60.0)11,519 (47.7)11,512 (47.7)0.00
  White race, n (%)14,408 (73.0)11,396 (71.8)11,485 (72.3)−0.01
  Previous cardiovascular event, n (%)2,084 (34.5)8,468 (35.1)8,537 (35.4)0.00
  Renal dysfunction, n (%)1,099 (18.2)4,385 (18.2)4,164 (17.3)0.00
  Metformin use, n (%)4,982 (82.5)16,009 (66.3)16,136 (66.9)0.00
  Statin use, n (%)3,866 (64.1)17,782 (73.7)17,772 (73.6)0.00
  ACE inhibitors/ARBs, n (%)4,543 (75.2)18,381 (76.2)18,470 (76.5)0.00
  β-Blockers, n (%)2,296 (38.8)9,860 (40.9)10,124 (42.0)0.00
  Baseline glucose-lowering therapy, n (%)
   None557 (9.2)8,736 (36.2)8,836 (36.6)0.00
   1 glucose-lowering agent3,988 (66.0)11,574 (48.0)11,437 (47.4)0.00
   ≥2 glucose-lowering agents1,496 (24.8)3,821 (15.8)3,858 (16.0)0.00
  HbA1c, mean (SD), %27.2 (0.6)8.11 (1.61)8.15 (1.85)0.02
  HbA1c, mean (SD), mmol/mol255 (6.6)65 (17.6)66 (20.2)0.02
  Total cholesterol, mean (SD), mg/dL2177 (44)175.21 (45.85)176.79 (48.86)0.03
  LDL-cholesterol, mean (SD), mg/dL295 (36)88.63 (40.20)87.49 (42.68)0.03
  eGFR, mean (SD), mL/min/1.73 m2277 (20)81.89 (23.92)83.91 (23.00)0.09
  UACR >30–300 μg/mg, n (%)21,281 (21.2)120 (24.9)123 (27.9)0.00
  UACR >300 μg/mg, n (%)2260 (4.3)48 (10.0)26 (5.9)0.03
Post-PSM C-statisticN.A.0.53**
Follow-up, mean (SD) and median (IQR), days222 (221); 131 (62, 294)
Control outcomesN events (IR/1,000 PY)N events (IR/1,000 PY)HR (95% CI)
Control end point 1–Severe hypoglycemia (expected positive association with glimepiride)60 (4.3)160 (10.4)2.38 (1.79–3.13) Ref. = linagliptin
Control end point 2–ESRD (expected null association with linagliptin)30 (2.1)32 (2.1)1.08 (0.66–1.79) Ref. = glimepiride
  • ARB, angiotensin receptor blocker; aSt. Diff., absolute standardized difference; eGFR, estimated glomerular filtration rate; IQR, interquartile range; IR, incidence rate; N.A., not applicable; PY, person-years; Ref., reference.

  • 1Only available in Medicare.

  • 2Only available in commercial data sets for a 7% subset.

  • *Values <0.1 are considered unlikely to cause any meaningful confounding.

  • **A postmatching C-statistic of 0.5 indicates perfect balance across all covariates.