The two patient groups (Rivaroxaban 20 and Rivaroxaban 15) were compared according to demographics. Patients in the group with the reduced Rivaroxaban dose were older, weighed less, and had higher CHA2DS2VASc and HAS-BLED scores than patients in the full dose group (Table 1). In addition, creatinine (µmol/L) and creatinine clearance (CrCl) (mL/min), were previously reported to differ between the dose groups [5].

There were no statistically significant differences between patients with and without bleeds regarding age, gender, weight, hypertension, diabetes, chronic heart failure, peripheral artery obstructive disease, ischemic heart disease, cerebrovascular disease, cerebrovascular insult, CHA2DS2VASc score or HAS-BLED score (Table 2). This was true for patient both on Rivaroxaban 20 and 15.

PT DOAC (CTR) was determined in 178 trough and 177 peak samples. Trough CTR was 0.93–3.36 with the median at 1.33, while peak CTR was 1.80–7.03 with the median at 3.57. The CTR was significantly higher in peak samples as compared to trough samples (p < 0.001), indicating that the effect of rivaroxaban can be measured using PT DOAC.

Neither trough nor peak CTR differed between patients on Rivaroxaban 20 vs. patients on Rivaroxaban 15 p = 0.663 and p = 0.443 respectively (Fig. 1), indicating that the effect of rivaroxaban was similar in both patient groups at the time right before the next dose (trough) and that both doses give the same effect in the patients (peak).

Clot time ratio in trough and peak samples. Comparison of mean CTR at trough (A, C) and peak (B, D) measured in plasma samples from 30 patients on Rivaroxaban 20 (A-B) and 30 patients on Rivaroxaban 15 (C-D) using PT DOAC. The bottom and top of the box represent the first and third quartiles (Q1-Q3), while the band inside the box represents the median

We used the first to third quartile range (Q1-Q3) of CTR to evaluate the association with clinical outcomes. No overlap was seen between the trough CTR Q1-Q3 range (TQ1−Q3) and the peak Q1-Q3 (PQ1−Q3) for either dose-group (Fig. 1). The PQ1−Q3 for the lower dose was however wider than for the full dose (CTR 2.7–4.6 and CTR 3.2–4.2 respectively), indicating a larger variation in the effect obtained from the adjusted dose compared to the full dose of Rivaroxaban.

CTR at repeated samplings (through 1, 2, 3, and peak 1, 2, 3) did not vary for patients on Rivaroxaban 15, or at peak for Rivaroxaban 20 (Table 3), while there was a significant difference for CTR at trough for Rivaroxaban 20 at sampling 1 compared to sampling 2 (Table 3). A post-hoc Bonferroni t-test revealed the difference between trough sampling 1 and 2. Further analysis of the results at trough sampling 1 disclosed that one patient (#20) had a very high CTR at trough sampling 1 versus trough sampling 2 and 3. Accordingly, this patient had a higher rivaroxaban concentration at the first sampling than the following samplings (285 ng/mL vs. 35 and 57). Exclusion of this sample from the analysis did however not remove the difference between trough sampling 1 and 2.

Intra-patient CV at trough and peak did not differ between the dose groups (Table 3). Combined CV (both doses) for CTR at trough was lower than CV for CTR at peak (p < 0.001).

CTR at peak did not correlate with age or body weight, and while mean peak CTR was lower in patients with cerebrovascular disease (p = 0.025) compared to patients without, no differences were seen between patients with/without any of the following patient characteristics: hypertension, diabetes, chronic heart failure, peripheral artery obstructive disease, ischemic heart disease, CHA2DS2VASC score, HAS-BLED, bleedings or thrombotic events. Mean trough CTR did not correlate to any of the evaluated parameters.

During the study, 28 patients (13 on 20 mg and 15 on 15 mg rivaroxaban) suffered bleeding. Three patients suffered major bleedings, and 25 suffered 1–4 minor ones [5]. There was no difference in mean peak- or trough CTR between patients with- and without bleeds, p = 0.445 and p = 0.202 respectively. When looking at the results for the two dose groups separately it was found that patients on Rivaroxaban 20 with bleeds had higher mean peak CTR than patients without bleeds (p = 0.040). In comparison, there was no significant difference in mean peak CTR between patients on Rivaroxaban 15 with or without bleeds (p = 0.803) or at trough for any of the dose groups (p = 0.209, p = 0.648, Fig. 2).

Clot time ratio in patients without and with bleeds. Individual CTRs (white) from sampling 1, 2 and 3 and the mean CTR (grey) for patients on Rivaroxaban 20 without bleeds (circles) and with bleeds (diamonds). The dashed lines represent the first and third quartiles (PQ1−Q3) for patients on 20 mg Rivaroxaban

Five out of seven patients (71%) on Rivaroxaban 20 with mean peak CTR above the dose specific PQ1−Q3 suffered bleeds while 7/16 patients (44%) with mean peak CTR within, and 1/7 patients (14%) with mean peak CTR below the PQ1−Q3 suffered bleeds indicating a higher risk of suffering bleeds on full dose rivaroxaban reflected in higher CTR at peak. In addition, among the rivaroxaban 20 non-bleeders only 1 had peak CTRs that were consistently above the PQ1−Q3 while the remaining 5 patients had a single peak CTR above the PQ1−Q3. On the other hand, among the Rivaroxaban 20 bleeders, 4 patients had peak CTRs consistently above the PQ1−Q3, one patient had 2/3 peak CTRs above the PQ1−Q3, while 2 patients had a single CTR above PQ1−Q3. Figure 2 shows the distribution of peak CTRs for individual samples at sampling 1, 2 and 3 as well as the patient mean CTRs for patients on Rivaroxaban 20 without and with bleeds.

For patients on Rivaroxaban 15, 3/7 patients (43%) with mean peak CTR above the dose specific PQ1−Q3, 8/16 (50%) with mean peak CTR within the PQ1−Q3 and 4/7 patients (57%) with mean peak CTR below the PQ1−Q3 suffered bleeds. The risk of suffering bleeds was similar regardless of CTR for this patient group.

Three patients suffered thrombotic events during the study (two on Rivaroxaban 20 and one on Rivaroxaban 15). One of the patients on Rivaroxaban 20 also suffered a minor bleed. As only three patients were suffering from thrombotic events, and their mean CTR at trough and peak varied with the respective Q1-Q3, no conclusions can be drawn regarding CTR and thrombotic events.

Rivaroxaban plasma levels measured by LC-MS/MS (ng/mL) or anti-FXa (ng/mL) were available for 355 samples. There was a significant correlation between CTR, and the rivaroxaban plasma levels (r = 0.905, p < 0.001 for LC-MS/MS and r = 0.892, p < 0.001 for anti-FXa. The sensitivity, specificity, positive- and negative predictive values, and the area under the curve of the ROC for PT DOAC CTR at the upper limit of the reference interval to predict a rivaroxaban concentration > 50 ng/mL are shown in Table 4. The resulting ROC curves for the prediction of DOAC plasma levels > 50 ng/mL as determined with LC-MS/MS and anti-FXa methods are shown in Fig. 3.

ROC curves. ROC curves for the prediction of Rivaroxaban concentration > 50 ng/mL A; LC-MS/MS and B; Anti-FXa