Fig. 1.

(A) Baseline associations between statin use and study outcomes: bar graphs represent the number of patients with related outcomes; percentages (95% confidence intervals) represent frequencies; P values represent univariable associations between statin use and clinical outcomes assessed by Fisher’s exact test. (B) Baseline distribution balance after propensity score (PS) analyses: PS was computed using a logistic regression model with statin treatment as the dependent variable and the following as explanatory (independent) variables: gender; age; ethnicity; body mass index; arterial hypertension; history of micro- or macrovascular diabetes complications; heart failure; treated obstructive sleep apnoea or chronic obstructive pulmonary disease (COPD); and use of any of the following drugs/drug classes on admission [metformin, dipeptidyl peptidase-4 inhibitors, glucagon-like peptide (GLP)-1 receptor agonists; insulin; ezetimibe; and renin–angiotensin–aldosterone system blockers, including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, mineralocorticoid receptor antagonists (MRA), calcium-channel blockers, anticoagulant agents and corticosteroids]. IPTW, inverse probability of treatment weighting. (C) PS-weighted associations between statin use and clinical outcomes: baseline covariates were used to compute PS in all multivariable models (see B); sensitivity analyses used these baseline variables plus routine HbA1c and routine estimated glomerular filtration rates (eGFR).