Incidence and risk of venous thromboembolism in critical care: a population-based observational cohort study

Rationale — Venous thromboembolism (VTE) remains an important complication of critical illness, but contemporary population-level estimates of clinically recognized VTE and ICU-associated risk are limited.

Objectives — To quantify 90-day VTE incidence in ICU vs non-ICU hospitalizations, estimate the adjusted association between ICU exposure and VTE, and identify factors associated with VTE within the ICU cohort.

Methods — We conducted a population-based cohort study using linked health administrative data in Ontario, Canada (ICES). We included patients aged ≥16 years and randomly selected one hospitalization per patient. ICU exposure was identified using discharge abstract special care unit codes. Clinically recognized VTE within 90 days was identified using ICD-10-CA diagnosis codes with confirmatory imaging/procedure codes in the same encounter or within ±7 days. We estimated cumulative incidence with death as a competing risk and fit cause-specific Cox models to estimate adjusted hazard ratios (aHRs), including an ICU-restricted model with ICU therapies and ICU length-of-stay.

Results — Among 4,821,130 index hospitalizations, 575,155 (11.9%) had ICU exposure. Ninety-day VTE occurred in 43,913 (0.9%) overall and was higher in ICU vs non-ICU hospitalizations (1.6% vs 0.8%; Gray’s test P < 0.001). Hospitalizations involving ICU exposure were associated with higher VTE hazard after adjustment (aHR 1.38, 95% CI 1.34–1.43). In the ICU cohort, higher VTE risk was associated with prior VTE, cancer, recent surgery, prolonged ICU stay, extracorporeal membrane oxygenation, and traumatic brain injury.

Conclusions — Hospitalizations involving ICU exposure were associated with higher 90-day clinically recognized VTE risk, and risk was concentrated in identifiable high-risk ICU subgroups, supporting risk stratification and risk-enriched future studies.