Browse the corpus

Impaired fibrinolysis in critically ill COVID-19 patients. BACKGROUND: Critically ill coronavirus disease 2019 (COVID-19) patients present with a hypercoagulable state with high rates of macrovascular and microvascular thrombosis, for which hypofibrinolysis might be an important contributing factor. METHODS: We retrospectively analysed 20 critically ill COVID-19 patients at Innsbruck Medical University Hospital whose coagulation function was tested with ClotPro® and compared with that of 60 healthy individuals at Augsburg University Clinic. ClotPro is a viscoelastic whole blood coagulation testing device. It includes the TPA test, which uses tissue factor (TF)-activated whole blood with added recombinant tissue-derived plasminogen activator (r-tPA) to induce fibrinolysis. For this purpose, the lysis time (LT) is measured as the time from when maximum clot firmness (MCF) is reached until MCF falls by 50%. We compared COVID-19 patients with prolonged LT in the TPA test and those with normal LT. RESULTS: Critically ill COVID-19 patients showed hypercoagulability in ClotPro assays. MCF was higher in the EX test (TF-activated assay), IN test (ellagic acid-activated assay), and FIB test (functional fibrinogen assay) with decreased maximum lysis (ML) in the EX test (hypofibrinolysis) and highly prolonged TPA test LT (decreased fibrinolytic response), as compared with healthy persons. COVID-19 patients with decreased fibrinolytic response showed higher fibrinogen levels, higher thrombocyte count, higher C-reactive protein levels, and decreased ML in the EX test and IN test. CONCLUSION: Critically ill COVID-19 patients have impaired fibrinolysis. This hypofibrinolytic state could be at least partially dependent on a decreased fibrinolytic response.

This retrospective study includes 20 critically ill COVID-19 patients (confirmed by polymerase chain reaction [PCR]) and 60 healthy persons. Remainder samples from routine blood donations made by healthy persons were analysed. All critically ill COVID-19 patients treated between the start of the COVID-19 pandemic and April 17, 2020 with at least one ClotPro measurement (first on April 2, 2020) during their stay in an ICU of Innsbruck Medical University Hospital were included in the analysis. ClotPro measurements were done when the physician felt the need for rapid insight into blood coagulation because extensive hypercoagulability was suspected. Hypercoagulability was defined either by difficulties in reaching the 0.3–0.5 IU ml−1 anti-Xa target range despite high doses of low molecular weight heparin (LMWH) or elevated D-dimer levels of >2000 μg L−1. Only one patient did not fulfil one of these criteria, but his coagulation status was analysed because of developing thrombocytopaenia.

efined either by difficulties in reaching the 0.3–0.5 IU ml−1 anti-Xa target range despite high doses of low molecular weight heparin (LMWH) or elevated D-dimer levels of >2000 μg L−1. Only one patient did not fulfil one of these criteria, but his coagulation status was analysed because of developing thrombocytopaenia. This study was approved by the institutional review board of the Medical University of Innsbruck (Vote #1139/2020). There was no need to obtain oral and written informed consent from the study participants because the data were retrospectively collected and anonymously processed. In order to have normal values for ClotPro analysis, leftover whole blood samples from healthy blood donors were obtained for ClotPro measurements. The institutional review board of Augsburg University Clinic gave its permission for this analysis without the need for obtaining oral or written informed consent from the blood donors (Vote #2018-13).

s for ClotPro analysis, leftover whole blood samples from healthy blood donors were obtained for ClotPro measurements. The institutional review board of Augsburg University Clinic gave its permission for this analysis without the need for obtaining oral or written informed consent from the blood donors (Vote #2018-13). We collected the patient characteristics age, sex, and diagnosed underlying diseases, if applicable. Blood gas and routine laboratory test results (complete blood count, blood chemistry, organ function biomarkers, inflammation and plasma coagulation parameters) at ICU admission and closest to ClotPro measurements were included. ICU charts were used to collect data on organ function, the Simplified Acute Physiology Score 3 (SAPS 3; a prediction model applied at ICU admission to determine probability of death at hospital discharge)15 was calculated for the ICU admission day, and Sequential Organ Failure Assessment (SOFA) scores were calculated for the day of ICU admission and ClotPro assay.

unction, the Simplified Acute Physiology Score 3 (SAPS 3; a prediction model applied at ICU admission to determine probability of death at hospital discharge)15 was calculated for the ICU admission day, and Sequential Organ Failure Assessment (SOFA) scores were calculated for the day of ICU admission and ClotPro assay. ClotPro measurements were routinely carried out in COVID-19 patients in order to gain rapid insight into blood coagulation properties when extensive hypercoagulability was suspected and to assess their fibrinolytic response. ClotPro is a commercially available (enicor GmbH, Munich, Germany), CE-marked viscoelastic in vitro coagulation analyser used mainly in Central Europe as a point-of-care test.16 , 17 It uses pipettes prefilled with starting and modifying agents and 340 μl of citrated whole blood for initiating measurement. A stationary pin is placed in a clockwise and counter-clockwise moving cup, from which the reduction of movement is detected and charted as the amplitude resulting in thromboelastometry curves known from other viscoelastic test methods such as ROTEM. The samples from healthy blood donors and COVID-19 patients were run within 4 h of blood draw.

d in a clockwise and counter-clockwise moving cup, from which the reduction of movement is detected and charted as the amplitude resulting in thromboelastometry curves known from other viscoelastic test methods such as ROTEM. The samples from healthy blood donors and COVID-19 patients were run within 4 h of blood draw. Standard tests used in COVID-19 patients were the EX test (tissue factor (TF)-activated assay), IN test (ellagic acid-activated assay), FIB test (functional fibrinogen assay), and TPA test (r-tPA within an extrinsic pathway-based assay), and the RVV test (Russel viper venom-based assay) if the patient received heparin, or the ECA test (ecarin-based assay) if the patient received argatroban, a direct thrombin inhibitor, as anticoagulant. The RVV test is sensitive to coagulation factor X inhibitors, whereas the ECA test is sensitive to direct thrombin inhibitors. The TPA test measures the fibrinolytic response, which might result in impaired fibrinolysis. The TPA test measures fibrinolysis by adding 650 ng ml−1 r-tPA to TF-activated whole blood. Lysis time (LT) is the time to dissolution of 50% of the clot (defined as maximum clot firmness [MCF]) by r-tPA once MCF is reached. Although the TPA test is currently available exclusively for research use by the manufacturer, this assay is a standardised assay with the reagents already placed in fixed doses in the pipettes. In the context of the COVID-19 pandemic, we use this assay routinely in critically ill COVID-19 patients to evaluate their fibrinolytic response.

t is currently available exclusively for research use by the manufacturer, this assay is a standardised assay with the reagents already placed in fixed doses in the pipettes. In the context of the COVID-19 pandemic, we use this assay routinely in critically ill COVID-19 patients to evaluate their fibrinolytic response. We compared the LT of COVID-19 patients with that of 60 healthy persons who were analysed at Augsburg University Clinic, for whom the 95% confidence interval (CI) of the TPA test LT was 157–393 s. Therefore, decreased fibrinolytic response is defined as a LT >393 s. The critically ill COVID-19 patients were further divided into two groups to compare patients with normal fibrinolytic response (LT within reference range) and those with decreased fibrinolytic response (LT > 393 s). Differences between groups, that is either healthy volunteers vs COVID-19 patients or COVID-19 patients with normal vs decreased fibrinolytic response, were assessed using Fisher's exact test for binary variables and the Wilcoxon rank-sum test for continuous variables. We provide effect sizes as odds ratios or estimated median differences, with corresponding 95% CIs.

We collected the patient characteristics age, sex, and diagnosed underlying diseases, if applicable. Blood gas and routine laboratory test results (complete blood count, blood chemistry, organ function biomarkers, inflammation and plasma coagulation parameters) at ICU admission and closest to ClotPro measurements were included. ICU charts were used to collect data on organ function, the Simplified Acute Physiology Score 3 (SAPS 3; a prediction model applied at ICU admission to determine probability of death at hospital discharge)15 was calculated for the ICU admission day, and Sequential Organ Failure Assessment (SOFA) scores were calculated for the day of ICU admission and ClotPro assay.

ClotPro measurements were routinely carried out in COVID-19 patients in order to gain rapid insight into blood coagulation properties when extensive hypercoagulability was suspected and to assess their fibrinolytic response. ClotPro is a commercially available (enicor GmbH, Munich, Germany), CE-marked viscoelastic in vitro coagulation analyser used mainly in Central Europe as a point-of-care test.16 , 17 It uses pipettes prefilled with starting and modifying agents and 340 μl of citrated whole blood for initiating measurement. A stationary pin is placed in a clockwise and counter-clockwise moving cup, from which the reduction of movement is detected and charted as the amplitude resulting in thromboelastometry curves known from other viscoelastic test methods such as ROTEM. The samples from healthy blood donors and COVID-19 patients were run within 4 h of blood draw.

We compared the LT of COVID-19 patients with that of 60 healthy persons who were analysed at Augsburg University Clinic, for whom the 95% confidence interval (CI) of the TPA test LT was 157–393 s. Therefore, decreased fibrinolytic response is defined as a LT >393 s. The critically ill COVID-19 patients were further divided into two groups to compare patients with normal fibrinolytic response (LT within reference range) and those with decreased fibrinolytic response (LT > 393 s). Differences between groups, that is either healthy volunteers vs COVID-19 patients or COVID-19 patients with normal vs decreased fibrinolytic response, were assessed using Fisher's exact test for binary variables and the Wilcoxon rank-sum test for continuous variables. We provide effect sizes as odds ratios or estimated median differences, with corresponding 95% CIs.

The healthy volunteers were 38 (28–46.25) yr, and critically ill COVID-19 patients were 61.5 (56.25–68) yr old (median [range]; P<0.01). Sex distribution did not differ between healthy persons and COVID-19 patients (P=0.08). All patients in both study populations were Caucasian. The mean ICU length of stay of the COVID-19 patients was 26.5 (10) days with ICU mortality of 20% (4 out of 20). In median, ClotPro was performed on day 8.5 (4.5–15) of ICU stay, and blood gas analyses (BGA) were made within 63.5 (48.5–98.5) min and routine laboratory measures within 8.4 (7.9–10.3) h of ClotPro assays. A hypercoagulable state was seen in TF-activated ClotPro assays in COVID-19 patients as compared with healthy individuals with an EX test MCF of 68 (63–71) mm vs 61 (58–64) mm (P<0.01) and a FIB test MCF of 34 (28–39) mm vs 17 (13–20) mm (P<0.01).

The healthy volunteers were 38 (28–46.25) yr, and critically ill COVID-19 patients were 61.5 (56.25–68) yr old (median [range]; P<0.01). Sex distribution did not differ between healthy persons and COVID-19 patients (P=0.08). All patients in both study populations were Caucasian. The mean ICU length of stay of the COVID-19 patients was 26.5 (10) days with ICU mortality of 20% (4 out of 20). In median, ClotPro was performed on day 8.5 (4.5–15) of ICU stay, and blood gas analyses (BGA) were made within 63.5 (48.5–98.5) min and routine laboratory measures within 8.4 (7.9–10.3) h of ClotPro assays. A hypercoagulable state was seen in TF-activated ClotPro assays in COVID-19 patients as compared with healthy individuals with an EX test MCF of 68 (63–71) mm vs 61 (58–64) mm (P<0.01) and a FIB test MCF of 34 (28–39) mm vs 17 (13–20) mm (P<0.01). Although there was no significant difference in EX test clotting time (CT) of 52 (47–76) s vs 48.5 (45–54) s (P=0.08), IN test CT was longer in COVID-19 patients, namely 188 (167–215) s vs 159 (153–166) s (P<0.01), although still within the reference range (Table 1 ).Table 1ClotPro analysis in healthy individuals compared with critically ill COVID-19 patients.Table 1Total (n=80)Healthy (n=60)COVID-19 (n=20)Estimate with 95% CIaP-valuebNot knownTPA test CT (s)43 (37–51)42 (36–46)50 (42–90)–12 (–38 to –3)<0.010/0 LT (s)238 (198–324)210 (186–261)508 (365–827)–265 (–358 to –186)<0.010/0 MCF (mm)34 (29–41)32 (28–36)55 (38.75–64)–21 (–27 to –13)<0.010/0 ML (%)95 (94–96)95 (94–96)97 (95–98)–2 (–2 to –1)<0.010/0EX test CT (s)49 (45–55)48.5 (45–54)51.5 (47–76)–5 (–16 to 1)0.080/0 A5 (mm)49.5 (46–55)48 (44–52)58 (50–61)–9 (–12 to –5)<0.010/0 A10 (mm)57 (54–62)56 (53–59)65.5 (58–68)–8 (–11 to –4)<0.010/0 A20 (mm)61 (58–65)60 (58–63)68 (62–70)–6.39 (–9 to –4)<0.010/0 MCF (mm62 (59–65)61 (58–64)68.5 (63–71)–6 (–9 to –3)<0.010/0 ML (%)5 (3–8)6 (4–8)3 (2–7)2 (1–4)<0.010/1IN test CT (s)162 (154–176)159 (153–166)188 (168–215)–30 (–47 to –17)<0.010/0 A5 (mm)45 (42–48)44 (41–47)52 (44–57)–8 (–11 to –4)<0.010/0 A10 (mm)54 (50–57)53 (50–55)60 (53–64)–7 (–10 to –4)<0.010/0 A20 (mm)59 (55–61)58 (55–60)64.5 (58–68)–6 (–9 to –3)<0.010/0 MCF (mm)59 (56–62)59 (56–61)64 (59–69)–5 (–8 to –2)<0.010/0 ML (%)5 (3–7)5 (3.5–7)4 (3–8)1 (–1 to 2)0.501/3FIB test CT (s)68 (62–77)68 (63–75)72 (53–108)–4 (–25 to 9)0.570/0 A5 (mm)14 (10–20)12 (9–15)29 (23–32)–16 (–19 to –14)<0.010/0 A10 (mm)16 (11–22)14 (10–17)31 (25–35)–17 (–20 to –14)<0.010/0 A20 (mm)17 (12–24)15 (12–18)33 (27–37)–17 (–21 to –14)<0.010/0 MCF (mm)19 (14–26)17 (13–20)34 (28–39)–17 (–21 to –14)<0.010/0Continuous data are presented as medians (25th to 75th percentile).A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis.aOdds ratios for binary variables and estimated median difference for continuous variables (CI).bAssessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum te

A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis.aOdds ratios for binary variables and estimated median difference for continuous variables (CI).bAssessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum te st for continuous variables. ClotPro analysis in healthy individuals compared with critically ill COVID-19 patients. Continuous data are presented as medians (25th to 75th percentile). A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis. Odds ratios for binary variables and estimated median difference for continuous variables (CI). Assessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. COVID-19 patients had a median BMI of 28.8 (24.3–31.0). Their medical history showed that 50% had hypertension, 40% cardiovascular diseases, 25% diabetes mellitus, 10% solid oncologic diseases, and 5% immunologic diseases. Two patients (10%) also had a history of thromboembolic events before contracting SARS-CoV-2 (Table 2 ).

9 patients had a median BMI of 28.8 (24.3–31.0). Their medical history showed that 50% had hypertension, 40% cardiovascular diseases, 25% diabetes mellitus, 10% solid oncologic diseases, and 5% immunologic diseases. Two patients (10%) also had a history of thromboembolic events before contracting SARS-CoV-2 (Table 2 ). Median SAPS 3 (on ICU admission) for our critically ill COVID-19 patients was 56 (53–64) points; SOFA score was 6.5 (3–8.2) at ICU admission and 6.5 (6–8.2) on the day of ClotPro assay (Supplementary file 1, Table S3).Table 2Baseline characteristics of COVID-19 patients.Table 2Total (n=20)TPA LT ≤393 s (n=6)TPA LT >393 s (n=14)Estimate with 95% CIaP-valuebNot knownAge (yr)61.5 (56.25–68)66 (61–70.25)61 (54.75–65.25)6 (–4 to 15)0.270/0Female sex6/20 (30%)2/6 (33.3%)4/14 (28.6%)0.81 (0.07–12.38)10/0Height (cm)174 (170–176)166 (163–169)174 (170–177)–10 (–25 to 4)0.194/5Weight (kg)82 (74.2–97)82 (80–98)84 (73.9–95)4.5 (–18 to 23)0.711/4BMI (kg m−2)28.8 (24.3–31.0)28.3 (25.9–30.7)28.8 (25.6–30.8)2.0 (–9.2 to 13.3)0.714/6History of thromboembolic events2/20 (10%)0/6 (0%)2/14 (14.3%)Inf (0.08 to Inf)10/0Medical history Cardiovascular8/20 (40%)4/6 (66.7%)4/14 (28.6%)0.22 (0.01–2.24)0.160/0 Central nervous system4/20 (20%)1/6 (16.7%)3/14 (21.4%)1.34 (0.08–85.67)10/0 Coagulation2/20 (10%)0/6 (0%)2/14 (14.3%)Inf (0.08 to Inf)10/0 Diabetes mellitus5/20 (25%)0/6 (0%)5/14 (35.7%)Inf (0.42 to Inf)0.260/0 Gastrointestinal3/20 (15%)0/6 (0%)3/14 (21.4%)Inf (0.17 to Inf)0.520/0 Haematologic1/20 (5%)0/6 (0%)1/14 (7.1%)Inf (0.01 to Inf)10/0 Hepatologic3/20 (15%)2/6 (33.3%)1/14 (7.1%)0.17 (0–4.11)0.200/0 Hypertension10/20 (50%)2/6 (33.3%)8/14 (57.1%)2.54 (0.26–37.18)0.630/0 Immune1/20 (5%)1/6 (16.7%)0/14 (0%)0 (0–16.71)0.30/0Scores on ICU admission SAPS 3 (points)56 (53–64)56 (55–66)55 (53–61.5)3 (–6 to 14)0.381/0 SOFA Respiratory system (points)3 (3–4)3 (3–3)3 (3–4)0 (–1 to 1)0.790/0 SOFA Respiratory system >217/20 (85%)6/6 (100%)11/14 (78.6%)0 (0–5.81)0.520/0 SOFA Coagulation (points)0 (0–0)0 (0–0)0 (0–0)0 (0–0)0.260/0 SOFA Liver (points)0 (0–0)0 (0–0.75)0 (0–0)0 (0–1)0.440/0 SOFA Cardiovascular system (points)3 (0–3.25)0 (0–2.25)3 (0.75–3.75)–1 (–3 to 0)0.200/0 SOFA Cardiovascular system >212/20 (60%)2/6 (33.3%)10/14 (71.4%)4.57 (0.45–70.73)0.160/0 SOFA Nervous system (points)0 (0–0)0 (0–0)0 (0–0)0 (0–0)0.590/0 SOFA Renal (points)0 (0–1)0 (0–0)0 (0–1)0 (–1 to 0)0.360/0 SOFA (points)6.5 (3–8.25)3.5 (3–6.25)7 (3.5–8.75)–2 (–5 to 1)0.300/0Binary data are presented as no./total no.

FA Cardiovascular system >212/20 (60%)2/6 (33.3%)10/14 (71.4%)4.57 (0.45–70.73)0.160/0 SOFA Nervous system (points)0 (0–0)0 (0–0)0 (0–0)0 (0–0)0.590/0 SOFA Renal (points)0 (0–1)0 (0–0)0 (0–1)0 (–1 to 0)0.360/0 SOFA (points)6.5 (3–8.25)3.5 (3–6.25)7 (3.5–8.75)–2 (–5 to 1)0.300/0Binary data are presented as no./total no. (%), continuous data as medians (25th to 75th percentile).CI, confidence interval; TPA LT, lysis time in TPA test; SAPS 3, Simplified Acute Physiology Score 3; SOFA, Sequential Organ Failure Assessment.aOdds ratios for binary variables and estimated median difference for continuous variables (CI).bAssessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. Baseline characteristics of COVID-19 patients. Binary data are presented as no./total no. (%), continuous data as medians (25th to 75th percentile). CI, confidence interval; TPA LT, lysis time in TPA test; SAPS 3, Simplified Acute Physiology Score 3; SOFA, Sequential Organ Failure Assessment. Odds ratios for binary variables and estimated median difference for continuous variables (CI). Assessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables.

CI, confidence interval; TPA LT, lysis time in TPA test; SAPS 3, Simplified Acute Physiology Score 3; SOFA, Sequential Organ Failure Assessment. Odds ratios for binary variables and estimated median difference for continuous variables (CI). Assessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. Critically ill COVID-19 patients presented with impaired fibrinolysis compared with healthy individuals (Table 1). This was indicated by longer fibrinolytic response reflected by the LT in the TPA test than in healthy persons: 508 (365–827) s vs 210 (186–261) s (P<0.01), respectively (Fig. 1 a). Less lysis was also seen in the EX test, where maximum lysis (ML) was significantly lower in the COVID-19 population than in healthy persons: 3% (2–7%) vs 6% (4–8%) (P<0.01).Fig 1Decreased fibrinolytic response in critically ill COVID-19 patients. The numbers below the box plots depict median (25th to 75th percentile) and estimated median differences with corresponding 95% confidence intervals (CIs) and P-value. (a) Prolonged lysis time (LT) in the TPA test assay in critically ill COVID-19 patients as compared with healthy persons. Fibrinogen levels (b), platelet counts (d), and levels of C-reactive protein (e) were significantly elevated in patients with decreased fibrinolytic response, whereas D-dimers did not differ between patients with and without impaired fibrinolysis (c).Fig 1

tically ill COVID-19 patients as compared with healthy persons. Fibrinogen levels (b), platelet counts (d), and levels of C-reactive protein (e) were significantly elevated in patients with decreased fibrinolytic response, whereas D-dimers did not differ between patients with and without impaired fibrinolysis (c).Fig 1 Decreased fibrinolytic response in critically ill COVID-19 patients. The numbers below the box plots depict median (25th to 75th percentile) and estimated median differences with corresponding 95% confidence intervals (CIs) and P-value. (a) Prolonged lysis time (LT) in the TPA test assay in critically ill COVID-19 patients as compared with healthy persons. Fibrinogen levels (b), platelet counts (d), and levels of C-reactive protein (e) were significantly elevated in patients with decreased fibrinolytic response, whereas D-dimers did not differ between patients with and without impaired fibrinolysis (c). Of the 20 critically ill COVID-19 patients with suspected extensive hypercoagulability, 14 (70%) showed decreased fibrinolytic response as shown in the r-tPA challenge test (LT >393 s), whereas only six patients had a normal response to fibrinolysis activation (LT ≤393 s). Decreased fibrinolytic response (LT >393 s in the TPA test) was also associated with hypofibrinolysis as shown by a significantly lower EX test ML of 2% (1–3%) and IN test ML of 3% (2–4%) compared with 8% (5–9%) and 8% (5–10%) in critically ill COVID-19 patients with normal fibrinolytic response (P<0.01 and P=0.04), respectively.

ibrinolytic response (LT >393 s in the TPA test) was also associated with hypofibrinolysis as shown by a significantly lower EX test ML of 2% (1–3%) and IN test ML of 3% (2–4%) compared with 8% (5–9%) and 8% (5–10%) in critically ill COVID-19 patients with normal fibrinolytic response (P<0.01 and P=0.04), respectively. COVID-19 patients with decreased fibrinolytic response (prolonged TPA LT) also showed increased clot strength as reflected by a higher EX test MCF than did those within normal fibrinolytic response (70 [68 to 72] mm vs 60 [57–63] mm; P<0.01) (Table 3 ).Table 3ClotPro analysis of critically ill COVID-19 patients with decreased fibrinolytic response compared with critically ill COVID-19 patients with normal fibrinolytic response.Table 3Total (n=20)TPA LT ≤393 s (n=6)TPA LT >393 s (n=14)Estimate with 95% CIaP-valuebNot knownTPA test CT (s)49.5 (41.5–90.2)73.5 (46.5–109.5)47.5 (40.5–84.25)12 (–19 to 66)0.270/0 LT (s)508 (364.75–826.75)321.5 (278.75–347)573 (503–982)–296.5 (–688 to –160)<0.010/0 MCF (mm)55 (38.75–64)34 (27.5–43.5)62 (55–64.75)–22.98 (–36 to –10)<0.010/0 ML (%)97 (95–98)95.5 (95–96.75)97 (96.25–98)–1 (–3 to 0)<0.010/0EX test CT (s)51.5 (46.75–76.25)72 (46.25–106)51.5 (47–65.5)13.49 (–15 to 62)0.590/0 A5 (mm)58 (50–61)49 (47.25–52.25)59.5 (57.25–64)–10 (–15 to –3)0.010/0 A10 (mm)65.5 (58–67.5)56.5 (55–59.5)66 (64.5–69)–9 (–14 to –3)<0.010/0 A20 (mm)67.5 (62–70.25)60 (57.25–62.75)69.5 (67.25–71.75)–8.9 (–13 to –3)<0.010/0 MCF (mm)68.5 (62.75–71.25)60 (57.25–62.75)70.5 (68.25–72)–9 (–14 to –4)<0.010/0 ML (%)3 (2–7)7.5 (4.75–8.75)2 (1–3)4.6 (1–7)<0.010/1IN test CT (s)188 (167.75–214.75)203 (183.75–250)180 (164–208.25)23 (–18 to 71)0.110/0 A5 (mm)52.5 (44–56.75)44 (41.75–45.5)54.5 (51.5–59)–10.75 (–15 to –3)0.010/0 A10 (mm)60.5 (53–64.5)53 (50–54.5)62 (60.25–66)–9 (–14 to –3)0.010/0 A20 (mm)64.5 (58–68.25)57.5 (54–58.75)66.5 (64.25–69)–9 (–13 to –4)<0.010/0 MCF (mm)64.5 (59–69)58 (54–59)66.5 (64.25–69)–9.99 (–14 to –3)<0.010/0 ML (%)4 (3–8)7.5 (4.75–9.5)3 (2–4)4 (0–7)0.040/3FIB test CT (s)72 (53–108)94 (64–119)66 (52–104)15 (–18 to 61)0.280/0 A5 (mm)29 (23–32)22 (18–25)32 (28–34)–10 (–15 to –3)<0.010/0 A10 (mm)31 (25–35)24 (20–27.25)34 (31–36.5)–10 (–16 to –4)<0.010/0 A20 (mm)33 (27–37)25 (22–28)36 (32–38)–11 (–16 to –4)<0.010/0 MCF (mm)34 (28–39)26 (24–30)38 (34–40)–11 (–17 to –4)<0.010/0Continuous data as medians (25th to 75th percentile).A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clo

.5)–10 (–16 to –4)<0.010/0 A20 (mm)33 (27–37)25 (22–28)36 (32–38)–11 (–16 to –4)<0.010/0 MCF (mm)34 (28–39)26 (24–30)38 (34–40)–11 (–17 to –4)<0.010/0Continuous data as medians (25th to 75th percentile).A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clo tting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis; TPA LT, lysis time in TPA test.aOdds ratios for binary variables and estimated median difference for continuous variables (CI).bAssessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. ClotPro analysis of critically ill COVID-19 patients with decreased fibrinolytic response compared with critically ill COVID-19 patients with normal fibrinolytic response. Continuous data as medians (25th to 75th percentile). A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis; TPA LT, lysis time in TPA test. Odds ratios for binary variables and estimated median difference for continuous variables (CI). Assessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables.

A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis; TPA LT, lysis time in TPA test. Odds ratios for binary variables and estimated median difference for continuous variables (CI). Assessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. BMI and sex distribution were not significantly different between the groups with normal fibrinolytic response and decreased fibrinolytic response in the TPA test (P=0.71 and P=1). Also, medical history showed no significant difference between the groups for normal and decreased fibrinolytic response (Table 2). There were no significant differences in SAPS 3 and SOFA scores between patients with normal and prolonged LT in the TPA test (P=0.38 and P=0.30, respectively). Critically ill COVID-19 patients with decreased fibrinolytic response had significantly higher fibrinogen measurements. Fibrinogen was 674 (578–796) mg dl−1 vs 530 (390–568) mg dl−1 (P=0.02), and FIB test MCF was 38 (34–40) mm vs 26 (24–30) mm (P<0.01) in the groups with prolonged vs normal LT, respectively (Fig 1b). There was no difference in D-dimers between the groups. Patients with impaired fibrinolysis had D-dimer levels of 4860 (2440–7330) μg L−1 and patients with normal fibrinolysis time 6540 (5440–12 810) μg L−1 (P=0.46; Fig. 1c).

mm vs 26 (24–30) mm (P<0.01) in the groups with prolonged vs normal LT, respectively (Fig 1b). There was no difference in D-dimers between the groups. Patients with impaired fibrinolysis had D-dimer levels of 4860 (2440–7330) μg L−1 and patients with normal fibrinolysis time 6540 (5440–12 810) μg L−1 (P=0.46; Fig. 1c). Platelet number was higher in patients with decreased fibrinolytic response: 274 (198–344) G L−1 compared with 154 (151–173) G L−1 (P=0.04) (Fig. 1d). Furthermore, patients with impaired fibrinolysis showed higher levels of C-reactive protein (CRP) at ICU admission (23 [17–29] mg dl−1 vs 10 [5–13] mg dl−1, P<0.01) and at time of ClotPro assay (20 [13–28] mg dl−1 vs 4 [3–11] mg dl−1, P=0.02), as depicted in Fig. 1e. Systemic hypoperfusion parameters did not differ between those with normal and impaired fibrinolytic response as shown by lactate 8 (8–14) vs 9 (8–11) mg dl−1 (P=1) and base excess (BE) 6.4 (4.4–7.2) vs 6.25 (2.2–8.4) mmol L−1 (P=0.90), respectively (Supplementary file, Table S4). Three patients had a Horovitz Index <100 mm Hg at ClotPro analysis, all of whom showed impaired fibrinolysis (LT >393 s). In addition, although no significant difference in death rate was seen between the two COVID-19 groups (P=0.27), all four patients who died had impaired fibrinolysis.

Systemic hypoperfusion parameters did not differ between those with normal and impaired fibrinolytic response as shown by lactate 8 (8–14) vs 9 (8–11) mg dl−1 (P=1) and base excess (BE) 6.4 (4.4–7.2) vs 6.25 (2.2–8.4) mmol L−1 (P=0.90), respectively (Supplementary file, Table S4). Three patients had a Horovitz Index <100 mm Hg at ClotPro analysis, all of whom showed impaired fibrinolysis (LT >393 s). In addition, although no significant difference in death rate was seen between the two COVID-19 groups (P=0.27), all four patients who died had impaired fibrinolysis. There was no difference in number of thromboses between patients with impaired fibrinolysis and those without. In each group, one patient experienced thrombotic events. One patient with normal fibrinolytic response to r-tPA in vitro suffered from pulmonary embolism, which was the reason for ICU admission. The ClotPro assessment was performed 4 days after the event with successful lysis therapy with tenecteplase, and therefore correlation between this thromboembolic event and ClotPro assay results is limited. The other patient experienced three thrombotic events, a spleen infarction, and bilateral jugular vein thrombosis diagnosed on day 7 after ClotPro analysis. The TPA test showed a decreased fibrinolytic response to r-tPA in this patient.

correlation between this thromboembolic event and ClotPro assay results is limited. The other patient experienced three thrombotic events, a spleen infarction, and bilateral jugular vein thrombosis diagnosed on day 7 after ClotPro analysis. The TPA test showed a decreased fibrinolytic response to r-tPA in this patient. On the day of ClotPro analysis, 16 patients received anticoagulation with enoxaparin, an LMWH, at a median dose of 80 (60–100) mg day−1 with corresponding peak plasma levels of 0.30 (0.23–0.32) IU ml−1. Target anti-Xa levels were set at 0.3–0.5 IU ml−1, and patients who reached these levels received a medium LMWH dose of 100 (80–100) mg day−1, whereas patients who did not reach this target level also received 100 (80–120) mg day−1. There was no statistical difference in any of the routinely measured laboratory parameters between patients who reached the target level and those who did not.

hed these levels received a medium LMWH dose of 100 (80–100) mg day−1, whereas patients who did not reach this target level also received 100 (80–120) mg day−1. There was no statistical difference in any of the routinely measured laboratory parameters between patients who reached the target level and those who did not. The other four critically ill COVID-19 patients received argatroban, with three of them having an available argatroban plasma concentration measurement (anti-IIa measured via diluted thrombin time) on the day of ClotPro assay. Two patients did not reach the argatroban target level of 0.3–0.6 μg ml−1 at that time. They received 0.1 and 0.56 μg kg−1 min−1 argatroban with corresponding plasma levels of 0.29 and 0.13 μg ml−1, respectively. The patient who reached the target range received 0.42 μg kg−1 min−1 with a corresponding argatroban level of 0.50 μg ml−1. The reason for argatroban administration was that these patients did not respond to LWMH well enough as seen from the fact that the anti-Xa target range could not be reached despite high doses of LMWH. These patients did not suffer from suspected or proven heparin-induced thrombocytopenia (HIT).

The healthy volunteers were 38 (28–46.25) yr, and critically ill COVID-19 patients were 61.5 (56.25–68) yr old (median [range]; P<0.01). Sex distribution did not differ between healthy persons and COVID-19 patients (P=0.08). All patients in both study populations were Caucasian. The mean ICU length of stay of the COVID-19 patients was 26.5 (10) days with ICU mortality of 20% (4 out of 20). In median, ClotPro was performed on day 8.5 (4.5–15) of ICU stay, and blood gas analyses (BGA) were made within 63.5 (48.5–98.5) min and routine laboratory measures within 8.4 (7.9–10.3) h of ClotPro assays.

A hypercoagulable state was seen in TF-activated ClotPro assays in COVID-19 patients as compared with healthy individuals with an EX test MCF of 68 (63–71) mm vs 61 (58–64) mm (P<0.01) and a FIB test MCF of 34 (28–39) mm vs 17 (13–20) mm (P<0.01).

A hypercoagulable state was seen in TF-activated ClotPro assays in COVID-19 patients as compared with healthy individuals with an EX test MCF of 68 (63–71) mm vs 61 (58–64) mm (P<0.01) and a FIB test MCF of 34 (28–39) mm vs 17 (13–20) mm (P<0.01). Although there was no significant difference in EX test clotting time (CT) of 52 (47–76) s vs 48.5 (45–54) s (P=0.08), IN test CT was longer in COVID-19 patients, namely 188 (167–215) s vs 159 (153–166) s (P<0.01), although still within the reference range (Table 1 ).Table 1ClotPro analysis in healthy individuals compared with critically ill COVID-19 patients.Table 1Total (n=80)Healthy (n=60)COVID-19 (n=20)Estimate with 95% CIaP-valuebNot knownTPA test CT (s)43 (37–51)42 (36–46)50 (42–90)–12 (–38 to –3)<0.010/0 LT (s)238 (198–324)210 (186–261)508 (365–827)–265 (–358 to –186)<0.010/0 MCF (mm)34 (29–41)32 (28–36)55 (38.75–64)–21 (–27 to –13)<0.010/0 ML (%)95 (94–96)95 (94–96)97 (95–98)–2 (–2 to –1)<0.010/0EX test CT (s)49 (45–55)48.5 (45–54)51.5 (47–76)–5 (–16 to 1)0.080/0 A5 (mm)49.5 (46–55)48 (44–52)58 (50–61)–9 (–12 to –5)<0.010/0 A10 (mm)57 (54–62)56 (53–59)65.5 (58–68)–8 (–11 to –4)<0.010/0 A20 (mm)61 (58–65)60 (58–63)68 (62–70)–6.39 (–9 to –4)<0.010/0 MCF (mm62 (59–65)61 (58–64)68.5 (63–71)–6 (–9 to –3)<0.010/0 ML (%)5 (3–8)6 (4–8)3 (2–7)2 (1–4)<0.010/1IN test CT (s)162 (154–176)159 (153–166)188 (168–215)–30 (–47 to –17)<0.010/0 A5 (mm)45 (42–48)44 (41–47)52 (44–57)–8 (–11 to –4)<0.010/0 A10 (mm)54 (50–57)53 (50–55)60 (53–64)–7 (–10 to –4)<0.010/0 A20 (mm)59 (55–61)58 (55–60)64.5 (58–68)–6 (–9 to –3)<0.010/0 MCF (mm)59 (56–62)59 (56–61)64 (59–69)–5 (–8 to –2)<0.010/0 ML (%)5 (3–7)5 (3.5–7)4 (3–8)1 (–1 to 2)0.501/3FIB test CT (s)68 (62–77)68 (63–75)72 (53–108)–4 (–25 to 9)0.570/0 A5 (mm)14 (10–20)12 (9–15)29 (23–32)–16 (–19 to –14)<0.010/0 A10 (mm)16 (11–22)14 (10–17)31 (25–35)–17 (–20 to –14)<0.010/0 A20 (mm)17 (12–24)15 (12–18)33 (27–37)–17 (–21 to –14)<0.010/0 MCF (mm)19 (14–26)17 (13–20)34 (28–39)–17 (–21 to –14)<0.010/0Continuous data are presented as medians (25th to 75th percentile).A5, amplitude after 5 min; A10, amplitude after 10 min; A20, amplitude after 20 min; CI, confidence interval; CT, clotting time; LT, lysis time; MCF, maximal clot firmness; ML, maximum lysis.aOdds ratios for binary variables and estimated median difference for continuous variables (CI).bAssessed with Fisher's exact test for categorical variables and the Wilcoxon rank sum te

Critically ill COVID-19 patients presented with impaired fibrinolysis compared with healthy individuals (Table 1). This was indicated by longer fibrinolytic response reflected by the LT in the TPA test than in healthy persons: 508 (365–827) s vs 210 (186–261) s (P<0.01), respectively (Fig. 1 a). Less lysis was also seen in the EX test, where maximum lysis (ML) was significantly lower in the COVID-19 population than in healthy persons: 3% (2–7%) vs 6% (4–8%) (P<0.01).Fig 1Decreased fibrinolytic response in critically ill COVID-19 patients. The numbers below the box plots depict median (25th to 75th percentile) and estimated median differences with corresponding 95% confidence intervals (CIs) and P-value. (a) Prolonged lysis time (LT) in the TPA test assay in critically ill COVID-19 patients as compared with healthy persons. Fibrinogen levels (b), platelet counts (d), and levels of C-reactive protein (e) were significantly elevated in patients with decreased fibrinolytic response, whereas D-dimers did not differ between patients with and without impaired fibrinolysis (c).Fig 1

On the day of ClotPro analysis, 16 patients received anticoagulation with enoxaparin, an LMWH, at a median dose of 80 (60–100) mg day−1 with corresponding peak plasma levels of 0.30 (0.23–0.32) IU ml−1. Target anti-Xa levels were set at 0.3–0.5 IU ml−1, and patients who reached these levels received a medium LMWH dose of 100 (80–100) mg day−1, whereas patients who did not reach this target level also received 100 (80–120) mg day−1. There was no statistical difference in any of the routinely measured laboratory parameters between patients who reached the target level and those who did not. The other four critically ill COVID-19 patients received argatroban, with three of them having an available argatroban plasma concentration measurement (anti-IIa measured via diluted thrombin time) on the day of ClotPro assay. Two patients did not reach the argatroban target level of 0.3–0.6 μg ml−1 at that time. They received 0.1 and 0.56 μg kg−1 min−1 argatroban with corresponding plasma levels of 0.29 and 0.13 μg ml−1, respectively. The patient who reached the target range received 0.42 μg kg−1 min−1 with a corresponding argatroban level of 0.50 μg ml−1. The reason for argatroban administration was that these patients did not respond to LWMH well enough as seen from the fact that the anti-Xa target range could not be reached despite high doses of LMWH. These patients did not suffer from suspected or proven heparin-induced thrombocytopenia (HIT).

We analysed the coagulation properties, especially fibrinolysis, of critically ill COVID-19 patients with suspected extensive hypercoagulability admitted to the ICU of Innsbruck Medical University Hospital. We found that, in addition to a hypercoagulable state, 70% of these patients suffered from impaired fibrinolysis accompanied by a decreased response to r-tPA-induced clot lysis as measured with viscoelastic whole blood coagulation assays (ClotPro). In critically ill COVID-19 patients, a prothrombotic coagulopathy is commonly found.18 In addition to increased factor levels18 and endotheliopathy,19 impaired fibrinolysis (lysis at 30 min in TEG) was observed in this particular patient group, and this was associated with increased thrombosis risk and need for dialysis.18 , 20 , 21 The problem with impaired fibrinolysis is that already formed thromboses and microthromboses cannot be completely resolved and therefore might contribute to the high rate of thrombotic complications, such as pulmonary embolism, in COVID-19 patients.18

ed with increased thrombosis risk and need for dialysis.18 , 20 , 21 The problem with impaired fibrinolysis is that already formed thromboses and microthromboses cannot be completely resolved and therefore might contribute to the high rate of thrombotic complications, such as pulmonary embolism, in COVID-19 patients.18 Our ICU patients with decreased fibrinolytic response also showed hypofibrinolysis in the EX test. Therefore, it can be concluded that this hypofibrinolytic state might be at least partially caused by a decreased response to the pro-fibrinolysis factor r-tPA and not merely by less lysis activation inside the clot. This finding is supported by a recent study by Nougier and colleagues,14 who report that the hypofibrinolytic state in critically ill COVID-19 patients was mainly attributable to increased fibrinolysis inhibitor levels of PAI-1 and TAFI.14 Inflammation itself promotes local release of tPA and PAI-1 from endothelial cells.22 This increase in PAI-1 levels might affect fibrinolysis more than the simultaneous increase in tPA arising in COVID-19 patients.10 , 14 Also, platelets provide a major source of PAI-1, which can be released by various triggers such as hypoxaemia23 or activation via thrombin.24 Our patients with hypofibrinolysis had a higher platelet count and therefore potentially higher releasable amounts of PAI-1. Although platelet count does not indicate the activity state of platelets, platelet inhibition leads to improved lung function in COVID-19 patients.25

s hypoxaemia23 or activation via thrombin.24 Our patients with hypofibrinolysis had a higher platelet count and therefore potentially higher releasable amounts of PAI-1. Although platelet count does not indicate the activity state of platelets, platelet inhibition leads to improved lung function in COVID-19 patients.25 Regardless of their source, PAI-1 and TAFI were identified as predictive biomarkers for decreased fibrinolytic response in non-COVID-19 patients suffering from pulmonary embolism, whereas fibrinogen, α2-antiplasmin, plasminogen, thrombin time, and D-dimer were not.26 Although D-dimer is formed during fibrinolysis, it gives no information about the state of fibrinolysis (hypo-, normo- or hyper-fibrinolysis) as it is not known how much fibrin is formed in vivo and the fibrinolytic system might fail to clear the huge amount of fibrin that has already formed.27 This might be the reason why we could not find a difference in D-dimers between patients with and without impaired fibrinolytic response, although fibrinogen levels were significantly higher in patients with impaired fibrinolysis. Another explanation as to why D-dimers were not lower in patients with impaired fibrinolysis might be that plasminogen not only cleaves fibrinogen, but also misfolded proteins and necrotic tissue,27 which could be present in a larger amount in patients with a higher state of inflammation, as shown in our patients with impaired fibrinolysis.

to why D-dimers were not lower in patients with impaired fibrinolysis might be that plasminogen not only cleaves fibrinogen, but also misfolded proteins and necrotic tissue,27 which could be present in a larger amount in patients with a higher state of inflammation, as shown in our patients with impaired fibrinolysis. Thrombus formation with impaired fibrinolysis not only leads to thrombosis and microthrombosis in the vascular system, but also to fibrin deposition in the alveoli. Increased levels of PAI-1 in bronchoalveolar lavage fluid are found in multiple pulmonary disorders,28 and in patients with pneumonia intra-alveolar fibrin deposition was associated with high levels of CRP.29 Our COVID-19 patients showed an association between higher levels of CRP and impaired fibrinolysis, which might contribute to intra-alveolar fibrin deposition. Inflammation and subsequent fibrin deposition within the alveolar space might impair gas exchange.30 In our patients, it is notable that three patients had a Horovitz index <100 mm Hg at ClotPro analysis, all of whom showed impaired fibrinolysis. As fibrin deposition in the alveolar space appears to be part of the COVID-19 pathomechanism, fibrinolytic therapy was applied in only few patients, either with intravenous tPA31 32 or nebulised plasminogen.33 These studies showed transient or permanent improvement in lung function and thus oxygenation.33

fibrinolysis. As fibrin deposition in the alveolar space appears to be part of the COVID-19 pathomechanism, fibrinolytic therapy was applied in only few patients, either with intravenous tPA31 32 or nebulised plasminogen.33 These studies showed transient or permanent improvement in lung function and thus oxygenation.33 Other reasons for fibrin deposition and the high rate of thrombosis during COVID-19 should not be ignored. Endothelial dysfunction contributes to the formation of microthrombi, tissue oedema, and tissue hypoxia.34 Endothelial activation with TF expression promotes plasma coagulation and, if the endothelial line is disrupted, exposure to subendothelial structures such as collagen activates platelets. The endothelial dysfunction in COVID-19 is caused, on the one hand, directly by SARS-CoV-2 when it enters the cell19 and also by the inflammatory response and hypoperfusion due to thrombosis and microthrombosis formation. Furthermore, systemic hypoperfusion and capillary leakage and shock were reported to be present in COVID-19 patients.35 At any rate, in our critically ill COVID-19 patients, hypoperfusion caused by circulatory impairment seemed not to play a role as systemic hypoperfusion markers did not indicate such a state, but local hypoperfusion and tissue hypoxia might still be an issue.

nd shock were reported to be present in COVID-19 patients.35 At any rate, in our critically ill COVID-19 patients, hypoperfusion caused by circulatory impairment seemed not to play a role as systemic hypoperfusion markers did not indicate such a state, but local hypoperfusion and tissue hypoxia might still be an issue. Known difficulties encountered in administering sufficient anticoagulation are also present in COVID-19 patients and might contribute to the high rate of thrombotic complications, as these patients seem to need higher doses of, for example LMWH because of heparin resistance.36 At Innsbruck Medical University Hospital, argatroban is used as an alternative anticoagulant in critically ill patients.37 This is also done in COVID-19 patients, and in this particular patient population argatroban might have an additional benefit because this direct thrombin inhibitor seems to enhance fibrinolysis more than heparin does.38 As of now, impaired fibrinolysis seems to be only one pathophysiological aspect of COVID-19, and other drivers of the disease need to be further investigated and targeted. Nevertheless, it is a mechanism that can be treated (or counteracted), and knowing whether a patient suffers from decreased fibrinolytic response might be of greater importance when treatment with tPA is considered.31 , 32

We conclude that critically ill COVID-19 patients are in a hypercoagulable and hypofibrinolytic state, which is at least partly dependent on a decreased response to tPA-induced fibrinolysis. The decreased fibrinolytic response was associated with higher fibrinogen, platelet count, and CRP. Nevertheless, possible relationships of impaired fibrinolysis measured in vitro to clinical outcomes need to be further investigated. A major limitation is that our study was conducted retrospectively in a small number of patients. In addition, the healthy volunteers were much younger than the COVID-19 patients. Therefore, it is not clear whether the differences found were attributable to the older age or to the disease itself. Also, the healthy population could not be matched for other variables such as comorbidities, and therefore different ClotPro results might not solely attributable to COVID-19 infection. Furthermore, our findings rely on comparison of results from two different centres, which includes some risk of bias between the centres. However, from the magnitude of the difference we observed, we do not expect inter-centre variability to have a significant effect.

ight not solely attributable to COVID-19 infection. Furthermore, our findings rely on comparison of results from two different centres, which includes some risk of bias between the centres. However, from the magnitude of the difference we observed, we do not expect inter-centre variability to have a significant effect. Another point is that it does not reflect coagulation throughout COVID-19, but presents only a momentary picture during the phase of critical illness when ClotPro was performed (median, day 8.5 of ICU stay). Furthermore, ClotPro was not performed from blood drawn at the same time as the blood draw for routine laboratory tests, although one might argue that most parameters do not zigzag to the extreme during this period (median 8.4 h). Unfortunately, given these low numbers of thrombotic events in this study population and the lack of ClotPro assays at ICU admission, it is not possible to assess whether a patient with LT >393 s is more likely to experience a thromboembolic event. The inconsistent timing of the ClotPro assay during hospital admission is a weakness of our study because it makes it difficult to compare test results between patients. Furthermore, ClotPro was not performed in all critically ill COVID-19 patients, but in those already suspected of being in a hypercoagulable state or heparin-resistant, which may cause an indication bias.

A major limitation is that our study was conducted retrospectively in a small number of patients. In addition, the healthy volunteers were much younger than the COVID-19 patients. Therefore, it is not clear whether the differences found were attributable to the older age or to the disease itself. Also, the healthy population could not be matched for other variables such as comorbidities, and therefore different ClotPro results might not solely attributable to COVID-19 infection. Furthermore, our findings rely on comparison of results from two different centres, which includes some risk of bias between the centres. However, from the magnitude of the difference we observed, we do not expect inter-centre variability to have a significant effect.

Idea development and concept of analysis: MB, DF Statistical analysis and figure/tables: TH ClotPro assays: MB, JB, DS Collection of COVID patient-related data: MB, DS, JB COVID patient data and samples: DF, JB, MS, SK, VS, GL, MJ, DF Data from healthy persons: AG

MB has received research funding and travel grants from LFB Biomedicaments, Baxter GmbH, CSL Behring GmbH, Mitsubishi Tanabe and non-financial support from TEM International outside the submitted work. MJ reported receiving grants from Baxter; grants and personal fees from Fresenius Kabi; and speaking, consulting honoraria, or both from Sphingotec, CLS-Behring, Fresenius and Astute Medical outside the submitted work. CT reports grants and personal fees from BrainLab, grants and personal fees from DePuySynthes, grants and personal fees from Intrinsic Therapeutics, grants from TETEC AG, personal fees from Aesculap, grants and personal fees from Signus Medizintechnik, grants and personal fees from Medtronic, grants and personal fees from Icotec AG, grants and personal fees from Edge Therapeutics, grants from BIT-Pharma, outside the submitted work. DF has received study funding, honoraria for consultancy and board activity from Astra Zeneca, AOP orphan, Baxter, Bayer, BBraun, Biotest, CSL Behring, Delta Select, Dade Behring, Edwards, Fresenius, Glaxo, Haemoscope, Hemogem, Lilly, LFB, Mitsubishi Pharma, NovoNordisk, Octapharm, Pfizer, Tem-Innovation outside the submitted work. The other authors declare no conflicts of interest.