Introduction
Methods
Procedures
Outcomes
Variable | No VTE | Acute VTE | p value | ||
---|---|---|---|---|---|
Gender | Male | 1447 (49.8) | 118 (50.0) | 1 | |
Female | 1460 (50.2) | 118 (50.0) | |||
Race/ethnicity | American Indian or Alaskan Native | 6 (0.2) | 0 (0.0) | 0.119 | |
Asian or Pacific Islander | 52 (1.8) | 0 (0.0) | |||
Black | 995 (34.2) | 93 (39.4) | |||
White | 1645 (56.6) | 131 (55.5) | |||
Hispanic | 50 (1.7) | 1 (0.4) | |||
Others | 86 (3.0) | 4 (1.7) | |||
Unknown | 72 (2.5) | 7 (3.0) | |||
Age (years) | Mean (SD) | 66.2 (16.4) | 68.0 (16.7) | 0.125 | |
Body mass index (kg/m2) | < 18.5 | 52 (1.9) | 7 (3.3) | 0.329 | |
18.5–24.9 | 487 (17.7) | 44 (20.8) | |||
25–29 | 828 (30.2) | 60 (28.3) | |||
> 30 | 1378 (50.2) | 101 (47.6) | |||
Hospital LOS (days) | Mean (SD) | 8.8 (6.4) | 12.2 (9.2) | < 0.001 | |
Total ICU (days) | Mean (SD) | 1.9 (5.0) | 3.8 (8.1) | < 0.001 | |
Mechanical ventilation (days) | Mean (SD) | 1.1 (3.9) | 2.4 (6.9) | 0.005 | |
Cardiopulmonary resuscitation | 420 (14.4) | 37 (15.7) | 0.675 | ||
Oxygen device | None | 2241 (77.4) | 153 (64.8) | < 0.001 | |
Nasal cannula/non-rebreather mask | 574 (19.8) | 60 (25.4) | |||
Ventilator | 43 (1.5) | 11 (4.7) | |||
Other | 29 (1.0) | 5 (2.1) | |||
High-flow nasal cannula | 7 (0.2) | 7 (3.0) | |||
Vitals on presentation | Oxygen saturation (%) | Mean (SD) | 94.0 (6.2) | 93.4 (7.2) | 0.258 |
Heart rate (beats/minute) | 94.3 (19.7) | 100.7 (20.7) | < 0.001 | ||
Respiratory rate (breaths/minute) | 20.9 (6.4) | 21.3 (6.0) | 0.283 | ||
Diastolic blood pressure (mmHg) | 74.7 (15.6) | 74.6 (16.5) | 0.897 | ||
Systolic blood pressure (mmHg) | 134.1 (24.9) | 130.9 (25.6) | 0.074 | ||
Labs on presentation | White blood cell count (K/uL) | 7.5 (4.5) | 9.4 (6.0) | < 0.001 | |
Lymphocytes (K/uL) | 1.6 (1) | 1.4 (0.9) | 0.007 | ||
Neutrophils (K/uL) | 6.7 (2.5) | 6.6 (2.8) | 0.645 | ||
Hemoglobin A1c (%) | 7.8 (2.2) | 7.4 (2.2) | 0.281 | ||
B-type natriuretic peptide (pg/mL) | 167.2 (338.2) | 189.5 (311.9) | 0.387 | ||
C-reactive protein (mg/dL) | 9.2 (7.6) | 11.6 (8.5) | < 0.001 | ||
D-dimer (μg/mL) | 1.8 (2.3) | 4.6 (5.0) | < 0.001 | ||
Ferritin (ng/mL) | 753.3 (1673.9) | 725.7 (978.0) | 0.716 | ||
Fibrinogen (mg/dL) | 570.5 (168.3) | 554.0 (207.1) | 0.573 | ||
Interleukin-6 (pg/mL) | 76.3 (127.4) | 136.5 (159.8) | 0.094 | ||
)Hemoglobin (gm/dL) | 10.9 (2.9) | 12.4 (5.1) | 0.754 | ||
Lactate (mmol/L) | 1.8 (1.4) | 2.0 (1.3) | 0.384 | ||
Lactate dehydrogenase (U/L) | 322.9 (340.8) | 370.1 (367.3) | 0.079 | ||
Alanine transaminase (U/L) | 53.1 (205.8) | 64.5 (204.0) | 0.436 | ||
Aspartate aminotransferase (U/L) | 59.9 (216.8) | 100.3 (564.9) | 0.298 | ||
Procalcitonin (ng/mL) | 2.1 (8.7) | 1.3 (2.2) | 0.008 | ||
Blood urea nitrogen (mg/dL) | 26.0 (21.3) | 30.3 (26.0) | 0.018 | ||
Creatinine (mg/dL) | 1.6 (2.0) | 1.6 (1.9) | 0.711 | ||
Potassium (meq/L) | 4.0 (0.7) | 4.1 (0.8) | 0.034 | ||
Total bilirubin (mg/dL) | 0.8 (0.6) | 1.0 (2.6) | 0.227 | ||
Platelet count (K/uL) | 218.1 (95.3) | 258.7 (144.2) | < 0.001 | ||
Social history | Smoker | 216 (7.4) | 19 (8.1) | 0.826 | |
Alcohol Use | 45 (1.5) | 7 (3.0) | 0.168 | ||
Marijuana Use | 18 (0.6) | 3 (1.3) | 0.443 | ||
In-hospital medications | Inpatient anticoagulation therapeutic dose | 302 (11.5) | 101 (46.3) | < 0.001 | |
Inpatient anticoagulation prophylactic dose | 2446 (92.8) | 162 (74.3) | < 0.001 | ||
Home medications | Non-steroidal anti-inflammatory drugs | 29 (10.9) | 5 (18.5) | 0.384 | |
Azithromycin | 21 (7.9) | 4 (14.8) | 0.383 | ||
Hydroxychloroquine | 4 (1.5) | 1 (3.7) | 0.949 | ||
Angiotensin-converting enzyme inhibitors | 66 (24.7) | 9 (33.3) | 0.455 | ||
Angiotensin—ll receptor blockers | 42 (15.7) | 5 (18.5) | 0.919 | ||
Beta blockers | 91 (34.1) | 7 (25.9) | 0.52 | ||
Diuretics | 85 (31.8) | 9 (33.3) | 1 | ||
Statins | 115 (43.1) | 8 (29.6) | 0.252 | ||
Warfarin | 12 (4.5) | 2 (7.4) | 0.839 | ||
Aspirin | 88 (33.0) | 5 (18.5) | 0.187 | ||
P2Y12 inhibitors | 10 (3.7) | 2 (7.4) | 0.685 | ||
Direct-acting oral anticoagulants | 17 (6.4) | 0 (0.0) | 0.359 | ||
Other anticoagulants | 7 (2.6) | 0 (0.0) | 0.85 | ||
Corticosteroids | 26 (9.7) | 4 (14.8) | 0.619 | ||
Proton pump inhibitors | 70 (26.2) | 10 (37.0) | 0.329 | ||
Oxygen requirement prior to admission | Yes | 18 (6.7) | 0 (0.0) | 0.319 | |
Unknown | 5 (1.9) | 1 (3.7) | |||
Lab values (Maximum and Minimum) | White blood cell count Max | Mean (SD) | 8.8 (5.3) | 10.4 (6.2) | < 0.001 |
Lymphocytes Min | 12.3 (7.6) | 11.0 (7.5) | 0.013 | ||
Neutrophils Max | 84.0 (10.6) | 86.0 (10.5) | 0.007 | ||
Hemoglobin A1c Min | 7.8 (2.2) | 7.4 (2.2) | 0.291 | ||
B-type natriuretic peptide Max | 168.8 (333.9) | 197.0 (318.8) | 0.283 | ||
C-reactive protein Max | 7.6 (5.8) | 9.8 (8.1) | < 0.001 | ||
D-dimer Max | 2.3 (2.5) | 4.7 (3.8) | < 0.001 | ||
Ferritin Max | 871.4 (2476.0) | 773.6 (1094.9) | 0.284 | ||
Fibrinogen Max | 594.3 (168.9) | 571.8 (196.6) | 0.423 | ||
Interleukin-6 Max | 79.3 (137.6) | 138.3 (160.5) | 0.104 | ||
Hemoglobin Min | 10.0 (2.4) | 10.6 (0.5) | 0.445 | ||
Lactate Max | 2.0 (1.5) | 2.4 (2.7) | 0.165 | ||
Lactate dehydrogenase Max | 380.0 (369.0) | 438.1 (378.4) | 0.036 |
Statistical analysis
Initial data cleaning and analysis
Data cleaning
Model building
-
Multiple linear regression (MLR)
-
Multiple logistic regression (LR)
-
Decision tree
-
Random forest
Results
Prediction model for VTE
Multiple linear regression | Multiple logistic regression | Decision tree | Random forest |
---|---|---|---|
D-dimer | D-dimer | Therapeutic anticoagulation inpatient | D-dimer |
Nonalcoholic steatohepatitis | History of venous thromboembolism | History of venous thromboembolism | Therapeutic anticoagulation inpatient |
History of venous thromboembolism | Mechanical ventilation | D-dimer | Platelet count |
Therapeutic anticoagulation inpatient | Therapeutic anticoagulation inpatient | Age | Blood urea nitrogen |
High-flow nasal Cannula | High-flow nasal Cannula | Race/ethnicity | Age |
Mechanical ventilation | Nonalcoholic steatohepatitis | Blood urea nitrogen | WBC count |
Coronary artery bypass graft | Nonalcoholic steatohepatitis | Hospital length of stay | Systolic blood pressure on presentation |
Heart rate on presentation | Thyroid disease | Lymphocytes | |
Alanine aminotransferase | Nasal Cannula or non-rebreather Mask | Alanine aminotransferase | |
Chronic kidney disease | Coronary artery bypass graft | Abnormal potassium level (higher or lower) | |
Chronic kidney disease | B-type natriuretic peptide | ||
Ferritin | C-reactive protein | ||
Creatinine | |||
Lactate dehydrogenase | |||
Neutrophils | |||
Heart rate on presentation | |||
Total bilirubin | |||
Aspartate transaminase | |||
Diastolic blood pressure on presentation | |||
Venous thromboembolism | |||
Ferritin | |||
Oxygen saturation on presentation |
Model | Misclassification Rate | R-square | AIC | BIC | Sensitivity | Specificity | PPV | NPV | AUC | ||
---|---|---|---|---|---|---|---|---|---|---|---|
TS | VS | TS | VS | ||||||||
Multiple linear regression | NA | NA | 25.39% | 16.29% | 50 | 355 | NA | NA | NA | NA | NA |
Multiple logistic regression | 5.74% | 9.64% | 41.12% | 3.79% | 436 | 742 | 0.76 | 0.76 | 0.87 | 0.85 | 0.80 |
Decision tree | 7.11% | 9.65% | 19.89% | 11.35% | NA | NA | 0.69 | 0.65 | 0.78 | 0.79 | 0.77 |
Random forest | 6.84% | 8.40% | 58.89% | 18.76% | NA | NA | 0.68 | 0.82 | 0.26 | 0.97 | 0.83 |
Performance of the model
Discussion
Study | Country | Study type, time period | Total number of cases | Venous thromboembolism incidence rate | Analysis performed | Identified predictors |
---|---|---|---|---|---|---|
Cohen et al. | United States | Retrospective, March 1st to April 27th, 2020 | 9407 | 2.9% (2.4% in medical ward and 4.9% in ICU) | Multivariate analysis | For VTE or mortality: 1.Advanced age 2.Increasing Charlson Comorbidity Index 3.History of cardiovascular disease 4.ICU level of care, and 5.Elevated maximum D-dimer with a cutoff at least four times the upper limit of normal |
Dalager- Pedersen et al. | Denmark | Retrospective, January 27th to June 1st, 2020 | 1540 | 5% VTE (both ICU and general ward) | 30-day absolute risks | This Study compared COVID-19 and non-COVID-19 patients showed COVID-19 patients had a higher risk of VTE |
Freund et al. | France, Spain, Belgium, Italy, Chile, Canada | Retrospective, February 1st to April 10th, 2020 | 974 | 15% (only PE, DVT not studied) | Multivariable binary logistic regression | 1. Male gender 2. Age > 48 3. Heart rate 4. Prior history of VTEs 5. Clinical signs of DVT 6. Recent immobilization |
Mei et al. | China | Retrospective, January 1st to March 23rd, 2020 | 616 | 2% VTE (DVT and/or PE) | χ2 test, Fisher exact test, t test, and Mann–Whitney U test | This study compared Padua score in COVID-19 pneumonia and community-acquired pneumonia |
Poissy et al. | France | Retrospective case series, February 27th to March 31st, 2020 | 196 (ICU patients only) | 6.1% (PE only) | Simple descriptive analysis | None |
Rieder et al. | China | Retrospective, March 26th to April 20th, 2020 | 49 | 6.1% | Spearman test | The level of D-dimers at hospital admission and the maximum level during follow-up were correlated with days at the hospital, days in ICU, days on non-invasive ventilation, or days on invasive ventilation |
Study | Country | Study type, time period | Total number of cases | Venous thromboembolism incidence rate | Prediction model | Performance |
---|---|---|---|---|---|---|
Kampouri et al. | Switzerland | Retrospective, February 28th to April 30th, 2020 | 491 | 9.3% | Wells score for PE ≥ 2 points and D-dimer value ≥ 3,000 ng/mL | PPV: 18.2% NPV: 98.5 Accuracy: 0.905 |
Dujardin et al. | Netherlands | Retrospective, March 13th to April 9th, 2020 | 127 | 41.7% | Binary linear regression model; D-dimer is > 9 μg/mL and C-reactive protein > 280 mg/mL | Predicted probability: 92% |
Tsaplin et al. | Russia | Retrospective, April 30th to May 29th, 2020 | 168 | 6.5% | Modified Caprini score > 12; D-dimer > 3 upper limit of normal | Sensitivity: 73%; Specificity: 84% |
Spyropoulos et al. | United States | Retrospective, March 1st, 2020 to April 27th, 2020 | 9407 | 2.9% | The International Medical Prevention Registry on Venous Thromboembolism and D‐Dimer (IMPROVE‐DD) risk assessment model | AUC: 70%; sensitivity: 97%; specificity: 22% |
Freund et al. | France, Spain, Belgium, Italy, Chile, and Canada | Retrospective, February 1st to April 10th, 2020 | 974 | 15% | Revised Geneva score and D-dimer [D-dimer below the age-adjusted threshold (i.e., 500 µg/mL under 50 years and age × 10 over 50 years)] | AUC: 0.81 |