Table 4.
Sub‐cohorts | No of events | Hazard ratio [95% CI]* | Hazard ratio [95% CI]† propensity score adjusted |
---|---|---|---|
Association between Metformin use and incident LA | |||
Patients with CLD | 9 | 0.79 [0.12–5.19] | 0.70 [0.10–4.76] |
Patient without CLD | 21 | 1.38 [0.26–7.23] | 0.97 [0.28–3.31] |
Patient with CKD | 21 | 0.66 [0.18–2.45] | 1.04 [0.33–3.28] |
Patient without CKD | 9 | 1.45 [0.29–7.33] | 1.38 [0.27–6.96] |
Patients without CKD or CLD | 7 | 1.18 [0.23–6.19] | 1.44 [0.27–7.76] |
Association between CKD and incident LA | |||
Patients with CLD | 9 | 14.27 [2.91–69.97] | 14.79 [3.00–72.90] |
Patient without CLD | 21 | 5.04 [1.86–13.63] | 4.79 [1.77–12.94] |
The covariates which caused at least a 10% shift in the risk estimate for the univariate analysis were adjusted in the final model. The COX model for each analysis was constructed separately.
We categorized the continuous propensity score into 20 strata of 5% each for the distribution of scores. The covariates which caused at least a 10% shift in the risk estimate after given by stratum of propensity score were adjusted in the final model.