In a recent issue of the ARCHIVES, Mohr et al1 presented a meta-analysis of deep venous thrombosis (DVT) prophylaxis in patients undergoing elective hip surgery. The authors included only studies in which DVT was diagnosed by venography because, as pointed out in the text, noninvasive measures of asymptomatic DVT in this population are unreliable. Unfortunately, the authors were not as rigorous in their other inclusion criteria and have pooled data in a way that may be misleading.
The authors claim that no heterogeneity was introduced after grouping the prophylactic methods together. Potentially important variations within a single modality exist, and not all of these variations are equally effective in preventing thromboembolic complications. For example, many different warfarin protocols have been advocated, although few of them have been evaluated in prospective randomized trials. The best studied protocol uses a two-step procedure in which patients receive warfarin at low doses beginning