Predictive ability of modified Ottawa score for recurrence in patients with cancer-associated venous thromboembolism: From the COMMAND VTE Registry

The external validation of the modified Ottawa score to predict the risk of recurrence in patients with cancer-associated venous thromboembolism (VTE) has not yet been firmly established. The present study aimed to evaluate the utility and limitations of the modified Ottawa score in the risk stratification of recurrent VTE in patients with cancer-associated VTE. The COMMAND VTE Registry is a multicenter retrospective registry enrolling 3027 consecutive patients with acute symptomatic VTE among 29 Japanese centers. The present study population consisted of 614 cancer-associated VTE patients, who were divided into 3 groups; High-risk group: 202 patients (33%) with a modified Ottawa score ≥ 1, Intermediate-risk group: 269 patients (44%) with a score = 0, and Low-risk group: 143 patients (23%) with a score ≤ −1. Recurrent VTE occurred in 39 patients on anticoagulation therapy within 6 months. The cumulative incidence of recurrent VTE substantially increased in the higher risk categories by the modified Ottawa score (high-risk group: 13.6% [95%CI, 8.9%–20.2%], intermediate-risk group: 5.9% [95%CI, 3.5%–9.8%], and low-risk group: 3.0% [95%CI, 1.1%–7.8%], P = .02). The discriminating power of the score was modest with a C-statistic of 0.63. Each score component of the score had a different impact on recurrent events with a variable effect size. The risks of recurrence in patients with cancer-associated VTE substantially increased in the higher risk categories by using the modified Ottawa score, but the discriminating power of the score for recurrence was modest with a variable impact of each score component on recurrent events. •The validity of the modified Ottawa score predicting recurrence is controversial.•We evaluated the score using a large observational real-world database.•The incidence of recurrence increased in higher risk categories by the score.•The discriminating power of the score was modest with a C-statistic of 0.63.•Each score component seemed to have a different impact on recurrence.