Prophylactic anticoagulation to prevent venous thromboembolism (VTE) in hospitalized medical patients has been implemented at most institutions as standard of care. Despite this, there is significant interest in developing risk stratification models to help predict those at highest risk of VTE to help prioritize patient selection for prophylaxis, and, thereby, potentially reduce the risk of bleeding complications. Although models predicting VTE risk have been developed, some are biased by high rates of cancer patients or suboptimal control selection, and, thus, none has been validated for predicting VTE risk in hospitalized patients; that is, until the model developed by Barbar and colleagues from the University of Padua, Italy.
In a prospective cohort study, these investigators assessed VTE risk for 1,180 inpatients admitted over a two-year period between January 2007 and December 2008, using a risk assessment model (RAM), the Padua Prediction Score, they developed. Individual risk scores were assigned based on a 20-point quantitative scoring system of risk factors, including cancer, myocardial infarction, stroke, congestive heart failure or respiratory failure, trauma, surgery, mobility, acute infectious or rheumatologic disease, thrombophilic condition, obesity (BMI ≥ 30), or hormone use. Renal disease, thrombocytopenia, major or recent bleeding, or pregnancy were exclusions. A total of 469 (39.7%) patients had a RAM score of 4 or greater, considered indicative of “high-risk,” while the remaining 711 (60.3%) had RAM scores of less than four and were considered “low-risk.”
Among the 469 high-risk patients, only a minority, 186 (39.7%), received thromboprophylaxis with unfractionated heparin, low-molecular-weight heparin, or fondaparinux during hospitalization. Of these, 2.2 percent (4) developed confirmed VTE over the next 90 days. In contrast, 11.8 percent (31) of the high-risk patients not receiving thromboprophylaxis developed VTE. Among the 711 low-risk patients, 52 (7.3%) received pharmacological thromboprophylaxis and only two (0.3%) developed confirmed VTE. Bleeding complications, including gastrointestinal, intramuscular, and cerebral, occurred in three (1.6%) of the 469 high-risk patients but were non-fatal.
Comparing these results to a previous scoring system1 revealed that the Padua Score identified twice as many high-risk patients (469 vs. 243). Among these additional subjects, nine (3.7%) developed VTE, indicating the Padua Score affords greater predictive value and protection against VTE.
Padua Risk Assessment Model*
| Points . | Condition . |
|---|---|
| 3 | CA, past VTE, mobility, thrombophillic condition |
| 2 | Trauma or surgery in past month |
| 1 | ≥70, CHF, AMI, Ischemic CVA, BMI ≥30, hormones, other* |
| Points . | Condition . |
|---|---|
| 3 | CA, past VTE, mobility, thrombophillic condition |
| 2 | Trauma or surgery in past month |
| 1 | ≥70, CHF, AMI, Ischemic CVA, BMI ≥30, hormones, other* |
RAM Score ≥ 4 = high risk of VTE
*Acute infectious or rheumatologic disorder
Prospective Cohort Study
| . | Group A . | Group B . | Group C . |
|---|---|---|---|
| Patient Groups . | High-Risk AC . | High-Risk no AC . | Low-Risk . |
| Padua Score | RAM ≥ 4 | RAM ≥ 4 | RAM < 4 |
| VTE Frequency | 4/186 (2.2%) | 31/283 (11.0%) | 2/711 (0.3%) |
| Hazard Ratio | 0.13 | 32.0 | |
| 95% CI, 0.04-0.40 | 95% CI, 4.1-251.0 | ||
| Group A vs. C | Group B vs. C | ||
| Bleeding Risk | - | 3/186 (1.6%) | - |
| . | Group A . | Group B . | Group C . |
|---|---|---|---|
| Patient Groups . | High-Risk AC . | High-Risk no AC . | Low-Risk . |
| Padua Score | RAM ≥ 4 | RAM ≥ 4 | RAM < 4 |
| VTE Frequency | 4/186 (2.2%) | 31/283 (11.0%) | 2/711 (0.3%) |
| Hazard Ratio | 0.13 | 32.0 | |
| 95% CI, 0.04-0.40 | 95% CI, 4.1-251.0 | ||
| Group A vs. C | Group B vs. C | ||
| Bleeding Risk | - | 3/186 (1.6%) | - |
In Brief
Several findings merit mention. First, it is of interest that even after thromboprophylaxis was discontinued after hospital discharge, the protection continued; that is, none developed recurrence. Although not immediately apparent, the authors surmise this is because correction of factors precipitating hospitalization provided protection. Further, it is of note that two of 17 (11.8%) high-risk patients continuing VTE prophylaxis post-hospitalization developed bleeding complications. This finding supports the importance of a risk-benefit assessment for decisions regarding long-term anticoagulation. Second, it is also of interest that only 40 percent of patients determined to be high risk received VTE prophylaxis. In many hospitals in the United States, including mine, the rate would have been higher, as the electronic medical record admissions orders require VTE prophylaxis for all medical patients unless a reason is provided to avoid it. Quantitative risk assessment tools, such as the Padua Score, offer a potential alternative to “all-inclusive” prophylaxis strategies for which ultimate risk-benefit, specifically bleeding complications, remain to be seen. Furthermore, with approval of newer safer oral anticoagulants (e.g., dabigatran, rivaroxiban) prophylaxis strategies will likely continue to evolve.
