Options for deep-vein thrombosis (DVT) prevention in medical patients include both unfractionated heparin (UFH) and low-molecular weight heparin (LMWH). Although more expensive than UFH, LMWHs are associated with fewer side-effects. We hypothesized that because of this, LMWH would prove cost-effective for DVT prevention in medical patients.
We modeled the cost and efficacy of UFH vs. LMWH for DVT prevention in a hypothetical cohort of 1,000 medically ill subjects. Model estimates were derived from published trials of DVT prophylaxis in this setting and from various meta-analyses describing rates of bleeding and heparin-induced thrombocytopenia (HIT). Costs of clinical events (e.g., DVT, bleeding, HIT) were taken from reports measuring the economic consequences of these outcomes. We used enoxaparin as the representative LMWH and biased the model against LMWH. We reduced literature based estimates of event rates by 20%, did not include fixed, pharmacy administration costs, and assumed that successful DVT prevention altered neither mortality nor quality of life. We performed multiple sensitivity analyses.
In the base-case, enoxaparin had minimal impact on rates of either DVT or bleeding. However, the rate of HIT was reduced by 90% with LMWH. Despite its higher acquisition costs, use of enoxaparin led to net savings. Total costs (e.g., sum of acquisition costs for medications plus costs for treatment of DVT, bleeding, and HIT) were $405 per patient with UFH vs. $316 with LMWH (net savings $89, 95% CI: $7 to $373 by Monte Carlo simulation). Sensitivity analysis revealed the model to be moderately sensitive to the costs of HIT and the frequency of HIT. In a worst-case scenario (all inputs skewed against enoxaparin), routine use of the LMWH still resulted in savings.
From a health system perspective, LMWHs are economically attractive for DVT prevention in medical patients. The impact of LMWHs on rates of HIT and bleeding more than balance their acquisition costs.
Physicians should consider side-effect profiles and the costs of potential side-effects when selecting among pharmacologic agents for DVT prevention.
Andrew Shorr, None.