Association of IVCF usage and brain metastases with 30-day mortality in patients diagnosed with melanoma, kidney, breast, or lung cancers and a VTE by the presence or absence of active bleeding
Variable∗ . | HR (95% CI) . |
---|---|
Active bleeding† (no interaction between IVCF and brain metastasis) | |
Brain metastases | |
Yes | 1.27 (0.98-1.65) |
No | Reference |
IVCF | |
Yes | 0.53 (0.42-0.68) |
No | Reference |
No active bleeding‡ (interaction between IVCF & brain metastasis) | |
Brain metastases | |
IVCF | |
Yes | 0.72 (0.60-0.85) |
No | Reference |
No brain metastases | |
IVCF | |
Yes | 0.91 (0.80-1.05) |
No | Reference |
Variable∗ . | HR (95% CI) . |
---|---|
Active bleeding† (no interaction between IVCF and brain metastasis) | |
Brain metastases | |
Yes | 1.27 (0.98-1.65) |
No | Reference |
IVCF | |
Yes | 0.53 (0.42-0.68) |
No | Reference |
No active bleeding‡ (interaction between IVCF & brain metastasis) | |
Brain metastases | |
IVCF | |
Yes | 0.72 (0.60-0.85) |
No | Reference |
No brain metastases | |
IVCF | |
Yes | 0.91 (0.80-1.05) |
No | Reference |
Inverse propensity-weighted Cox proportional hazard regression models were used. IVCF was included as time-dependent covariate. All models were also adjusted for age, year, and location of VTE; prior history of VTE; sex; race/ethnicity; health insurance; neighborhood socioeconomic status; cancer site; stage at diagnosis; comorbidities; and each contraindication for anticoagulation (in the active bleeding model, aortic dissection and brain, orthopedic, and spine surgeries were consolidated into 1 variable).
There was an interaction between IVCF and active bleeding (P = .0006) warranting stratified models.
Active bleeding: there was no interaction between IVCF and brain metastases (P = .7046).
No active bleeding: there was an interaction between IVCF and brain metastases (P = .0318).