Introduction Patients with pulmonary embolism (PE) at low or very low risk for 30-day mortality by risk prediction scores such as the Pulmonary Embolism Severity Index (PESI), can be safely managed as outpatients. Outpatient PE management has increased over time, facilitated by higher use of direct oral anticoagulants (DOACs). However, the presence of cancer confers higher mortality risk in the simplified PESI, and whether patients with cancer associated PE (CA-PE) are increasingly managed as outpatients is unknown. Therefore, we sought to describe trends in admission rates and determine factors associated with inpatient admission for CA-PE in a population-based cohort.

Methods Using California Cancer Registry (CCR) data linked with California Department of Health Care Access and Information (HCAI) emergency department (ED) and patient discharge datasets (PDD), we identified individuals ≥18 years with active cancer and PE seen in EDs between 2009-2018. Active cancer was defined hierarchically as 1) primary cancer diagnosis in the CCR < 12 months prior to the PE, 2) HCAI admission with metastatic recurrence coded < 12 months prior, or 3) HCAI admission with a principal or second position cancer coded < 12 months prior. Our analysis was limited to patients with active cancer diagnosed before the PE presentation and healthier patients discharged home for self-care. The primary outcome was admission to the hospital prior to being discharged home rather than direct discharge home from the ED. Descriptive statistics were used to characterize the cohort and multivariable hierarchical regression models to predict inpatient admission. Demographic, clinical, and facility characteristics were used as fixed effects and individual hospital as a random effect.

Results We identified 17798 PEs among 14763 patients with active cancer. Most patients had a single PE (71.6%). Among the 325 facilities analyzed, the median number of emergency department PE encounters was 29 from 2009-2018 (interquartile range [IQR], 8-76) with an overall median admission rate of 78.3% (IQR, 58.5%-88.5%). Median admission rates decreased over time as follows: 88.9% (IQR, 71.9%-100%) in 2009-2011, 80.6% (IQR, 61.7-93.9%) in 2012- 2014, and 73.9% (IQR, 54.5%- 87.5%) in 2015-2019. In the hierarchical models, facilities accounted for approximately 17.5% (intraclass correlation coefficient) of the variability in the admission rate. Urban (odds ratio [OR] 2.64, 95% confidence internal [CI] 2.11-3.30) and non-Kaiser teaching hospitals (OR 2.59, CI 1.87-3.60) were more likely to admit patients compared to Kaiser facilities. Admission rates did not differ significantly between rural facilities and Kaiser hospitals. Metastatic cancer (OR 1.12, CI 1.02-1.22), concurrent proximal deep venous thrombosis (pDVT) present at the time of PE (OR 3.81, CI 3.22-4.52), and major surgery within 90 days prior to PE (OR 1.39, CI 1.25-1.55) were also associated with increased likelihood of admission. Insurance status was available for patients from 2010-2019. When compared with patients with private insurance, those with Medicare (OR 8.52, CI 7.50-9.67) and other public insurance (OR 2.35, CI 2.03-2.71) were more likely to be admitted, whereas uninsured patients were less likely to be admitted (OR 0.68, CI 0.49-0.94). Patients diagnosed with a prior PE within 90 days of the new PE (OR 0.34, CI 0.31-0.38) and age ≥ 80 years (OR 0.27, CI 0.23-0.32) were less likely to be admitted.

Conclusions There was wide variation between hospitals in admission rates for patients with CA-PE, even after adjusting for demographic and clinical variables. A limitation of this study was the inability to calculate a PESI score for individual patients. However, since concurrent cancer is a strong driver of higher PE risk scores, our study results suggest that local practice patterns contribute significantly to variable admission rates. Concurrent pDVT and insurance status were also highly associated with admission. As with non-cancer patients, there was a trend towards outpatient management over time. Ongoing analysis will describe the disposition of all patients that present to the ED with CA-PE, and associations with readmission and mortality.

This content is only available as a PDF.
2025
Sign in via your Institution