Users versus non-users | Sample size | Number of events | Hazard ratio (95% CI) |
P
|
---|
ACEIs or ARBs | Â | Â | Â | Â |
 Model 1 | 4,307 | 199 | 0.38 (0.27, 0.53) | <0.00001 |
 Model 2 | 4,307 | 199 | 0.60 (0.43, 0.84) | 0.0027 |
 Model 3 | 4,307 | 199 | 0.55 (0.39, 0.78) | 0.0009 |
Statins | Â | Â | Â | Â |
 Model 1 | 5,172 | 243 | 0.36 (0.24, 0.53) | <0.0001 |
 Model 2 | 5,172 | 243 | 0.47 (0.32, 0.70) | 0.0002 |
 Model 3 | 5,172 | 243 | 0.47 (0.31, 0.70) | 0.0003 |
Insulin | Â | Â | Â | Â |
 Model 1 | 4,639 | 201 | 0.48 (0.31, 0.73) | 0.0006 |
 Model 2 | 4,639 | 201 | 0.59 (0.39, 0.89) | 0.0110 |
 Model 3 | 4,639 | 201 | 0.58 (0.38, 0.89) | 0.0119 |
Metformin | Â | Â | Â | Â |
 Model 1 | 2,658 | 129 | 0.38 (0.25, 0.56) | <0.0001 |
 Model 2 | 2,658 | 129 | 0.39 (0.25, 0.61) | <0.0001 |
 Model 3 | 2,658 | 129 | 0.39 (0.25, 0.61) | <0.0001 |
Sulphonylurea | Â | Â | Â | Â |
 Model 1 | 2,291 | 99 | 0.45 (0.29, 0.72) | 0.0008 |
 Model 2 | 2,291 | 99 | 0.44 (0.27, 0.73) | 0.0013 |
 Model 3 | 2,291 | 99 | 0.44 (0.27, 0.73) | 0.0014 |
TZDs | Â | Â | Â | Â |
 Model 1 | 6,074 | 270 | 0.12 (0.03, 0.50) | 0.0033 |
 Model 2 | 6,074 | 270 | 0.17 (0.04, 0.69) | 0.0133 |
 Model 3 | 6,074 | 270 | 0.18 (0.04, 0.72) | 0.0153 |
- Model 1 adjusted for propensity scores (c-statistics = 0.80 for ACEIs or ARBs; 0.80 for statins; 0.79 for insulin; 0.73 for metformin; 0.66 for sulphonylurea; and 0.74 for TZDs), which were calculated using logistic regression procedures with age, sex, body mass index, low- and high-density lipoprotein cholesterols (LDL-C and HDL-C), triglyceride, use of tobacco and alcohol, HbA1c, systolic blood pressure, natural log of the albumin-to-creatinine ratio, estimated glomerular filtration rate, duration of disease, peripheral arterial disease, retinopathy, neuropathy, prior myocardial infarction and stroke as independent variables. Model 2 adjusted for LDL-C-related risk indicators (that is, LDL-C <2.8 mmol/L plus albuminuria or LDL-C ≥3.8 mmol/L); non-linear associations of HDL-C and triglyceride for cancer; body mass index (<24, ≥27.6 kg/m2); age; sex; employment status; use of tobacco and alcohol; duration of disease; systolic blood pressure; and use of statins, ACEIs/ARBs, fibrates, sulphonylurea, metformin and TZDs during follow-up. Model 3 adjusted for all factors in model 2 plus the propensity score of the drug in question. No patient had been exposed to the drug in question for at least 2.5 years prior to enrolment. ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CI, confidence interval; TZD, thiazolidinediones.