Direct oral anticoagulants have revolutionized pulmonary thromboembolism management, yet their widespread adoption has prompted concerns about generalized application that may not adequately account for individual patient complexity, according to a perspective published in Frontiers in Medicine.
The perspective authors identified a fundamental tension: while direct oral anticoagulants offer advantages through fixed dosing and elimination of routine coagulation monitoring, this convenience has fostered what the researchers characterize as a "diagnose-and-prescribe" approach. The concern centers on extrapolating evidence from highly selected randomized controlled trial populations—which "frequently excluded patients with severe renal impairment, active cancer, frailty, or extreme body weight"—to diverse clinical scenarios, wrote Cheng Meng, MD, and Hao Wang, MD, of the Respiratory and Critical Care Medicine Department at Yan'an People's Hospital.
Special Populations Present Therapeutic Uncertainties
The challenges of applying standard DOAC therapy are especially pronounced in certain patient groups. Cancer patients experience elevated risks for both clot recurrence and hemorrhage, complicated by drug regimens that frequently include potent CYP3A4 and P-glycoprotein modulators that substantially modify DOAC blood levels and undermine treatment effectiveness and safety, according to the researchers.
Those with very high or very low body mass show substantial differences in how these medications are processed; typical doses may result in insufficient drug levels or excessive accumulation, raising the likelihood of repeat clotting events or hemorrhage. In older, frail individuals with kidney impairment, the authors highlighted diminished physiological capacity, variable kidney function measurements, increased fall susceptibility, multiple concurrent medications, and markedly higher hemorrhage risk. Furthermore, those with antiphospholipid syndrome, especially the triple-positive variant, may experience increased risk of repeat clotting with specific DOACs.
Evidence Gaps Complicate Duration Decisions
Current risk prediction models demonstrated substantial limitations for guiding extended anticoagulation. Existing tools such as VTE-BLEED and HERDOO2 were primarily developed using warfarin-treated populations, and their applicability and accuracy in patients receiving DOACs remain inadequately validated. These instruments may not fully account for DOAC-specific risk characteristics—for instance, the lower incidence of intracranial bleeding with DOACs could alter traditional risk-benefit calculations.
Moreover, these assessment tools rely on static evaluation, categorizing patients using baseline data from a single timepoint. This approach cannot accommodate the evolving clinical trajectories patients experience in actual practice. Such one-dimensional, static modeling significantly limits their usefulness for personalized treatment planning, according to Drs. Meng and Wang .
The convenience of fixed dosing without monitoring created an additional problem: critical follow-up elements are sometimes neglected. Dynamic changes in renal function require at least yearly creatinine clearance reassessment, yet this monitoring frequently gets overlooked in high-volume outpatient environments.
Proposed Framework Incorporates Four Dimensions
The researchers proposed a next-generation risk stratification system built on multidimensional, dynamic, and integrative principles. The framework encompasses 4 core dimensions: index event and thrombus biology (including provoked vs unprovoked status and serial D-dimer measurements); demographic and physiological parameters (age, body mass index, renal and hepatic function); comorbidities and complications (active cancer, cardiovascular disease, antiphospholipid syndrome, bleeding history); and treatment-related factors (anticoagulant selection, adherence, drug interactions, shared treatment expectations).
Machine learning technologies may facilitate implementation by rapidly analyzing diverse, complex, and unstructured healthcare information, identifying intricate relationships and patterns that human review might miss. These systems could provide immediate, research-supported risk assessment and treatment guidance during patient encounters.
Implementation Faces Multiple Barriers
Moving these approaches into real-world practice presents significant challenges. The difficulty lies in balancing model sophistication with practical utility—while complex multivariable models may produce better predictions, they can be cumbersome to use in fast-paced outpatient environments. Successful integration demands solutions for standardizing data, ensuring system compatibility, protecting patient information, and overcoming institutional and administrative obstacles.
The researchers recognized that altering established physician practices in diagnosis and treatment is inherently difficult and will necessitate focused educational initiatives. Economic analyses are essential to establish whether the system delivers improved patient outcomes while remaining financially viable.
Key areas for future investigation include conducting large prospective trials in specialized patient populations, creating and testing DOAC-specific prediction tools, and investigating dosing approaches guided by drug metabolism patterns in patients who process medications atypically. The research team intends to launch a prospective study applying their proposed approach in patients with cancer-related blood clots and atypical body weights.
The authors contended that guideline developers should adopt multifaceted, risk-stratified recommendations that gradually move away from standard treatment protocols. Achieving individualized anticoagulation therapy depends on interdisciplinary cooperation among clinicians, pharmacists, informaticists, and epidemiologists.
The researchers declared no conflicts of interest.
Source: Frontiers in Medicine