Variability in the use of thromboprophylaxis and outcomes in critically ill medical patients

Purpose To describe practices for preventing venous thromboembolism in critically ill medical patients and to identify associations between prophylactic measures and survival. Methods We reviewed the records of all medical admissions to the intensive care units of a university hospital and an affiliated Veterans Affairs hospital over a 1-year period. We recorded patients’ demographic characteristics, risk factors for venous thromboembolism, methods of prophylaxis, and in-hospital deaths. Results We identified 272 critically ill medical patients who received intensive care for at least 24 hours. Some form of prophylaxis was used in 205 patients (75%), including pharmacologic prophylaxis alone in 55 (20%), mechanical prophylaxis alone in 102 (38%), and both methods in 48 (18%). In-hospital mortality rates were 23% (24/103) for patients who received pharmacologic prophylaxis, and 36% (61/169) for those who received mechanical prophylaxis alone or no prophylaxis (P = .03). After adjusting for demographic characteristics, risk factors for thrombosis and severity of illness, the odds of death were 55% lower in patients who received pharmacologic prophylaxis (odds ratio [OR] = 0.45; 95% confidence interval (CI): 0.22 to 0.93; P = .03). Similar results were obtained in propensity-adjusted and propensity-stratified analyses. Use of mechanical prophylaxis was not associated with survival (OR = 0.88; 95% CI 0.44 to 1.77; P = .73). Conclusion In this cohort of critically ill medical patients, pharmacologic but not mechanical thromboprophylaxis was associated with reduced risk of in-hospital death. This hypothesis must be tested in randomized trials.