This prospective observational study was approved by the Swedish Ethical Review Authority (Dnr: 2022-02275-01) and all participants gave written informed consent prior to inclusion. The manuscript was written in accordance with the STROBE guidelines [16].

Patients over 18 years of age, reporting to the emergency department during office hours at Skåne University Hospital Lund, Sweden, from September to November 2022, were recruited for the study. The inclusion criterion was any indication for starting treatment with rivaroxaban. Exclusion criteria were previously diagnosed coagulopathy or medication within the last month with anticoagulants including vitamin K antagonists (VKAs) and DOACs, platelet inhibitors (not acetylsalicylic acid) and non-steroidal anti-inflammatory drugs.

Baseline blood samples were taken prior to the first peroral dose of rivaroxaban (Xarelto®, Bayer AG, Leverkusen, Germany;15 mg) and patients were instructed to take another dose of rivaroxaban, 15 mg, the following morning. This is in line with recommendations for treatment of acute onset venous thromboembolism. At their arrival the following morning, they were asked for compliance with the rivaroxaban medication and were excluded if rivaroxaban had not been taken as prescribed. At the follow up visit, a second set of blood samples was taken, with the aim of reaching the maximum plasma concentration of rivaroxaban in the blood. A blood sample for analysis of rivaroxaban concentration was also taken. On both occasions venous blood was drawn in 2.7 ml tubes containing 0.109 M citrate (BD Vacutainer Systems, Plymouth, UK) from an antecubital vein. Baseline characteristics and results from routine laboratory analyses were manually collected from medical records.

Immediately after blood sampling, 1 ml of blood was transferred to each of eight different plastic test tubes (Sarstedt, Micro tube 1.5 ml, Sarstedt AG & Co. KG, Nümbrecht, Germany) and divided into one temperature group and one acidosis group, each consisting of four vials.

To ensure the detectability of any hypothermic effect, even in cases where it may be subtle, we opted for temperatures ranging from 28 to 40 degrees Celsius (°C). The temperature group vials were incubated in water baths for 20 min at their target goal temperatures of 28, 33, 37, and 40 °C. Prior to utilizing the ROTEM devices, the temperature within the ROTEM system was adjusted to the specified testing temperature, allowing sufficient time for proper equilibration. The acidosis group was incubated at 37 °C and mixed with 15 µL of hydrochloric acid (HCl) at concentrations of 0.15 µM, 0.10 µM, 0.05 µM or sterile water, compensating for any dilutional effects from the HCl. Titration with the different concentrations of HCl was tested prior to the current study using blood from healthy volunteers, with the goal of reaching the clinically relevant, pre-defined pH levels of 6.80, 7.00, and 7.20. After the addition of hydrochloric acid and sterile water the pH in each tube was measured using a pH-meter (Testo, Testo 230 pH/Temperature Meter, Testo SE & Co. KGaA, Lenzkirch, Germany). All ROTEM assays were performed within four hours of drawing blood.

After incubation at different temperatures and the addition of HCl/sterile water, EXTEM assays were analysed using newly served, calibrated and tested ROTEM delta (ROTEM, ROTEM delta, TEM innovations, Munich, Germany) or Rotational thromboelastography (roTEG 05, TEM innovations, Munich, Germany). All assays were performed according to the manufacturer’s instructions. Due to time restrictions (4 h after sampling) we chose to use both ROTEM delta and roTEG. CT, AA, and CFT were registered. Maximal clot firmness (MCF) was not registered as previous studies with in-vitro modification to hypothermia were without effect on MCF [7]. All tests were run with the ROTEM instrument set to the incubation temperature.

We wanted to demonstrate a 25% higher synergistic effect of rivaroxaban and hypothermia at 28 °C in the ROTEM EXTEM CT assay compared to the calculated additive effect (see below) with a non-parametric, matched hypothesis test. Given a standard deviation of 19, a 90% power and an alpha-error probability of 0.01 we had to include 13 patients. To allow for dropouts we aimed to include 15 patients.

All continuous variables are presented as medians and ranges (min-max). All numbers are presented as percentages (%).

The temperature and acidosis groups both at baseline and after exposure to rivaroxaban were compared within themselves using the non-parametric Friedman test to evaluate the observed effects of hypothermia, acidosis, hypothermia with rivaroxaban, and acidosis with rivaroxaban on the ROTEM variables CT, AA and CFT. If the Friedman statistics were significant, post hoc Dunn’s multiple comparisons test was performed to investigate differences between different levels of temperature and pH compared to baseline (37 °C and pH 7.4).

To test for possible synergy between rivaroxaban and hypothermia or rivaroxaban and acidosis, a theoretical additive effect was calculated as a sum of the pure temperature or acidosis effect and the pure rivaroxaban effect. The pure temperature or pure acidosis effect was the difference in a variable (CT, CFT or AA) from baseline, where baseline was the value of that variable at 37 °C and pH 7.4. The rivaroxaban effect was calculated by subtracting the value of a rivaroxaban test (CT, CFT or AA) at a specific temperature or pH with the value at baseline defined as the rivaroxaban test at 37 °C and pH 7.4 respectively. This additive effect was then compared to the real observed effect of simultaneous hypothermia/acidosis and rivaroxaban. Hence, if there is no synergy between rivaroxaban or acidosis/hypothermia, there would be no difference between additive and observed effects. If synergy exists however, the observed effect would systematically be greater than the additive effect. After calculating the values of the additive and observed effects, they were compared using the two-tailed, non-parametric Wilcoxon matched-pairs signed rank test. The Wilcoxon test was also used to test possible pH differences between baseline and rivaroxaban samples.

A possible dose-response relationship of the synergy between rivaroxaban and hypothermia or acidosis on ROTEM variables was also investigated. This was only done if a significant synergy could be demonstrated in the pairwise testing of differences between additive and synergistic effects. To verify normally distributed residuals, four tests of normality were used: the D’Agostino-Pearson omnibus test, the Anderson-Darling test, the Shapiro-Wilk test, and the Kolmogorov-Smirnov test. To test if differences between the slopes of the lines, indicating dose-response synergy, a null-hypothesis of parallel lines was assumed generating a p-value to answer the question as to what the probability of the observed slopes is provided that the null hypothesis holds. Analyses for the study were performed using GraphPad Prism version 9.5.0 (GraphPad, San Diego, USA). Due to the large number of statistical tests planned a modified Bonferroni correction was established placing a requirement of p < 0.01 for statistical significance.