AUTHORS
Lauren E Merz, MD, MSc, Aliza Fink2, Krista Latta3, Maureen Okam Achebe Jr, MD, MPH4, Ted Wun, MD5, Sarah Reeves6
AFFILIATIONS
1 University of Michigan, Ann Arbor, Michigan, United States
2 American Society of Hematology, Washington, District of Columbia, United States
3 University of Michigan, Ann Arbor, Michigan, United States
4 Brigham & Women's Hospital, Boston, Massachusetts, United States
5 UC Davis Health, Sacramento, California, United States
6 University of Michigan, Ann Arbor, Michigan, United States
ABSTRACT
Introduction: People with SCD have a higher incidence of VTE than the general population, but there is a lack of precision around rate and risk factors. Existing health data-based VTE phenotypes, developed in the general population, use a combination of diagnosis and procedure codes and/or anticoagulation prescriptions. However, these definitions may not generalize to SCD, where frequent acute care and disease-related complications may affect VTE identification. We aimed to develop a reliable phenotype definition to identify VTE among adults with SCD using two complementary data sources: the ASH Research Collaborative (ASH RC) Data Hub and the Michigan Sickle Cell Data Collection Program (MiSCDC). Evaluating both datasets allowed assessment of whether VTE definitions were robust across complementary data environments: population-based SCD surveillance with linked administrative data and detailed site-derived clinical data.
Methods: The ASH RC Data Hub is a high-quality, real-world clinical data resource that aggregates data elements extracted from electronic health records of 19 participating institutions. MiSCDC is a public health surveillance system that links data from multiple data sources (e.g., Medicaid, hospital discharge, newborn screening) to identify and follow everyone living with SCD in Michigan.
In both datasets, the study population included patients with SCD (Data Hub: physician attestation; MiSCDC: SCD genotype available, or 3+ SCD-related claims within a 5-year period) who were 18+ years old with healthcare utilization during the years of observation (Data Hub: 2015-2025; MiSCDC: 2020-2022).
VTE events were identified through the presence of any ICD-10 (or SNOMED codes in Data Hub) for acute deep vein thrombosis or pulmonary embolism; chronic, superficial, or thrombophlebitis codes were excluded. To evaluate whether diagnosis codes were supported by evidence of acute clinical care or treatment, we calculated the proportion of VTE codes with acute encounters within ±7 days (emergency department visit or hospital admission) or an outpatient anticoagulation prescription within ±30 days (Data Hub only). VTE prevalence and rates per person-year of the proportion of codes with corroborating evidence were compared between datasets.
Results: In the Data Hub, 12,208 people with SCD met inclusion criteria (mean follow-up: 5.4 years over 10 years). There were 26,671 coded events of acute VTE. Of these, 11,225 (42%) had an acute encounter within ±7 days and 17,285 (65%) were associated with an outpatient anticoagulation prescription; 9,894 events (37%) had both acute care encounters and anticoagulation prescriptions. This corresponds to 1,525 individuals, indicating a VTE prevalence of 12.5% with an annual VTE rate of 6.4 per 100 person-years.
In MiSCDC, there were 3,416 people with SCD met inclusion criteria (mean follow-up: 2.7 years over 3 years). There were 933 unique VTE coded events. Of these, 556 (60%) had an acute encounter within ±7 days. This corresponds to 273 individuals and an 8.0% VTE prevalence with an annual VTE rate of 2.83 per 100 person-years.
Conclusion: VTE was frequently identified among adults with SCD in both datasets, though prevalence and annual rate was lower in MiSCDC than in the Data Hub. This difference likely reflects the tradeoff between data depth and population capture: the Data Hub contains detailed site-derived clinical information for individuals with SCD that receive care at 19 centers, whereas MiSCDC provides broader population-based surveillance with less granular clinical data. These findings support a VTE phenotype incorporating acute care and/or anticoagulation evidence rather than diagnosis codes alone, with data-source-specific adaptations to account for differences in encounter and medication capture.
As with prior VTE phenotypes proposed, it is more challenging to identify true recurrent events from repeat documentation of, or suspicion for, acute VTE. Future strategies include determining the feasibility of using diagnostic imaging codes for VTE and laboratory results to improve the definition. After further refinement, VTE rates by decade of life, genotype, sex, disease-modifying therapy, anticoagulation use, and pregnancy status will be determined to understand risk factors for VTE in SCD. This will help identify the highest-risk population for VTE prevention measures.
DISCLOSURES
Lauren E Merz, MD, MSc: 23andMe - Consultancy (Includes expert testimony); Johnson&Johnson - Consultancy (Includes expert testimony)