Main analysis
The INR results were measured from 8 weeks before statin initiation to 12 weeks after statin initiation. We graphically depicted the changes in INR values by mapping median, interquartile and 10thand 90th percentile during this window. To formally assess the INR changes, we estimated the increase in mean INR by comparing the latest INR result in the before window to the first INR result within week 3-6 after statin initiation (if any), by using a paired t test. We further assessed the median effect by estimating median changes in INR levels for all statin treated patients, as well as those treated with a high- (≥40mg) and low-dose (<40mg) simvastatin in secondary analyses. Finally, we calculated the proportion of patients with an INR above the therapeutic interval (for most patients INR between 2-3); defined as INR > 4 and > 5 by comparing the proportion of patients 1-4 weeks prior to initiating simvastatin to the proportion 3-6 weeks after.
All analyses were performed using STATA Release 14.1 (Stata- Corp, College Station, TX, USA). The study was approved by the Danish Data Protection Agency. Since data is based on anonymized register data neither approval from Ethics Committee nor collection of informed consent was needed.