Risk Factors for Bleeding after Recent Medical Hospitalization: The Medical Inpatient Thrombosis and Hemostasis Study (MITH)

Introduction: Bleeding after discharge from a medical hospitalization is associated with increased morbidity and mortality. There is limited knowledge of the incidence of and risk factors for post-discharge bleeding

Objective: To determine the incidence, timing and risk factors for bleeding leading to rehospitalization after discharge.

Methods: The study includes primary care patients from a health network who were discharged alive from one or more medical hospitalizations between October 2010 and September 2019 After each index hospitalization discharge, patients were followed for 90 days to capture post discharge (PD) bleeding requiring rehospitalization. Validated computable phenotypes for electronic health record data were used to define PD bleeding. Codes were used to identify baseline comorbidities and admission diagnosis. Laboratory data were collected within 24 hours of discharge from index hospitalization. Medication reconciliation lists were used to ascertain exposure to anticoagulation. Inpatient use and intensity of anticoagulation were determined on the day of and the day after admission. Age, sex, and length of stay-adjusted hazard ratios (HR) for candidate bleeding risk factors were estimated using Cox proportional hazards models.

Results: Over 9 years, there were 15,630 medical hospitalization discharges and 414 (2.6%) PD bleeding events requiring re-admission. Mean age was 68 years (standard deviation=15.9) and 55.5% (8,675) were female. The median time to PD bleeding was 24 days (IQR 8-50). 50% (208) of the PD bleeding events were gastrointestinal, followed by hematomas and cavity bleeding (11.3%) and genitourinary bleeding (10.4%). By contrast to other types of bleeding, central nervous system bleeding had a median time to bleeding longer than one month: 43 days (IQR 17-61), with other sites having a median time to bleeding of less than one month (median, 24 days IQR 8, 24).The strongest risk factors for bleeding were an index hospitalization for bleeding (HR 4.3; 95% CI 1.3-2.7), platelet count <50 x 10³/mm³ (HR 3.7; 95% CI 2.4-5.8), and anemia (HR 3.07; 95% CI 2.21-4.26) at the time of discharge . Other risk factors included index hospital admissions for gastrointestinal diseases (HR 2.23; 95% CI 1.76-2.82) and new cancer (HR 1.77; 95% CI 1.07-2.92). Each 1-standard deviation increase in the Elixhauser comorbidity index for mortality and rehospitalization was associated with a 28% (HR 1.28 95% CI 1.17, 1.4) and 41% (HR 1.41 95% CI 1.3,1.54) increase in the risk of PD bleeding, respectively. Anticoagulation use during the index hospitalization was not associated with increased risk of PD bleeding regardless of intensity (HR 0.66; 95% CI 0.51, 0.84 for full and HR 0.51 95% CI 0.41,0.64 for prophylactic), but discharge on anticoagulation was associated with an increased risk (HR 1.36; 95% CI 1.02-1.82).

Conclusion: PD bleeding requiring re-hospitalization complicated 2.6% of medical hospitalizations in this cohort and had objective and readily identifiable risk factors. Thus, PD bleeding may be predictable and quantifiable with an appropriately developed and validated risk assessment model