Impact of screening versus symptomatic measurement of deep vein thrombosis in a national quality improvement registry

Background: Deep vein thrombosis (DVT) is a quality measure recorded by initiatives such as the National Surgical Quality Improvement Program (NSQIP). However, because surveillance-detected DVT rates may be higher than symptomatic DVT rates, we examined how differences in the method of DVT detection may affect the use of this quality measure. Methods: Using the NSQIP database (2007-2009), we compared DVT rates of vascular (amputation, open aortic procedures, and lower extremity bypass) and nonvascular (prostatectomy, gastric bypass [GBP], and hip arthroplasty) operations. Using a predefined literature search strategy, we compared the incidence of DVT in NSQIP to the incidence of DVT reported in published literature, diagnosed by symptomatic status or by surveillance studies. Results: Within NSQIP, the overall incidence of postoperative DVT was 0.7%. This varied from 0.3% after GBP to 1.8% after open aortic surgery. Across all procedures except amputation, the incidence of DVT in NSQIP was similar to the incidence of DVT reported in our literature survey of “symptomatic” DVTs. The relative rate (RR) of literature-derived symptomatic DVTs to NSQIP ranged from 0.7 for aortic cases (95% confidence interval [CI], 0.3-1.7) to 1.4 (95% CI, .7-3.1) for GBP. Overall, surveillance studies had 11.6 higher RR of DVT compared to NSQIP (95% CI, 10.5-13), ranging from 2.6 for GBP (95% CI, 1.4-5) to 14 .5 for hip arthroplasty (95% CI, 10.5-20). Conclusions: The incidence of DVT reported in NSQIP is similar to the reported incidence of symptomatic DVT for many high-risk procedures but is much lower than rates of DVT reported in surveillance studies. Clear delineation of symptomatic vs surveillance detection of DVT would improve the usefulness of this measurement in quality improvement registries.