A recent study calls into question use of VTE prophylaxis rates as a measure of quality of care because of likely surveillance bias. Bilimoria and colleagues () looked at almost a million surgical patients in almost 3000 hospitals and looked at VTE prophylaxis rates, VTE event rates, and use of imaging in VTE screening. They found that greater hospital VTE prophylaxis adherence rates were weakly associated with worse risk-adjusted VTE event rates. When they looked at hospitals with higher “structural” quality scores (based on 8 factors commonly thought to be associated with higher commitment to quality) they found higher VTE prophylaxis adherence rates but worse risk-adjusted VTE rates. Risk-adjusted VTE rates increased significantly with VTE imaging use rates in a stepwise fashion, leading to their conclusion that surveillance bias limits the usefulness of the VTE quality measure for hospitals. They also note that their study is not the first to note that higher VTE prophylaxis rates do not necessarily result in lower VTE rates. They note that these findings and the likely effect of surveillance bias call into question the use of a VTE prophylaxis metric as a quality measure in public reporting or pay-for-perfomance programs.
The accompanying editorial (Livingston 2013) agrees that public reporting of VTE rates should be reconsidered or curtailed. It also raises the question that the high compliance rates with VTE prophylaxis may mean that many patients are receiving treatments they are unlikely to benefit from.
In our February 15, 2011 Patient Safety Tip of the Week “Controversies in VTE Prophylaxis” we highlighted a study by some very respected investigators in the surgical quality improvement field (Qadan 2011) that questioned the current recommendations on venous thrombembolism (VTE) prophylaxis in elective major surgery. The authors collected data on DVT and PE in patients electively undergoing 4 major surgical procedures (colorectal resection, total knee replacement, total hip replacement, and hysterectomy) from a large database from a consortium of academic medical centers for two periods of time (2003-2004 vs. 2007-2008). The study demonstrated a substantial increase in the use of pharmacologic DVT prophylaxis between the two time periods. Yet the rates of DVT and PE were not significantly impacted by this increased use of such prophylaxis. Moreover, the overall rates of DVT and PE were actually quite low and the rates in patients who did not receive pharmacoprophylaxis actually decreased between the two time periods. The authors concluded that this may show that clinical judgment of physicians in choosing which patients need pharmacoprophylaxis is remarkable.
Unfortunately, the study did not have available data on complications of pharmacoprophylaxis (such as hemorrhage, heparin-induced thrombocytopenia, etc.) which might show that the net benefit of prophylaxis may be even lower.
One of the most striking revelations of this paper is that the overall rate of VTE in this elective surgery population is quite low (0.6% to 3.2%). One of the biggest areas of controversy in all prior investigations on VTE has been the means by which DVT is diagnosed. Most randomized controlled trials have used ultrasound techniques to look for evidence of DVT. Many have argued that such techniques artificially increase the incidence and prevalence of DVT by including many cases that will never be clinically relevant. The authors point out industry funding of studies that may overestimate the true VTE rate. They point out that the number needed to treat (NNT) to prevent a single PE in the elective surgical population is about 125 and an even higher NNT would apply to fatal PE. (We, of course, would argue that even one fatal PE is one too many.)
Pertinent to the comments by Qadan et al. about the influence of “industry” in driving the many clinical studies that have led to larger use of pharmacological VTE prophylaxis, a study just published provided interesting insights into potential conflicts of interest amongst panelists involved in development of the widely used ACCP guidelines (Neumann 2013). Those authors estimated the compliance with a conflict of interest policy by attendees voting on controversies for which they were conflicted. Sixty-three panelists voted in at least one controversy at the final conference; the percentage of conflicted panelists varied from 6% to 39% for eight controversies. The compliance with the COI policy varied from 5% to 33% in seven of the controversies voted on. In two of the controversies (“Compression device plus aspirin vs low-molecular-weight heparin in thromboprophylaxis in orthopedic surgery” and “Low-molecular-weight heparin vs vitamin K antagonists for treatment”), the low compliance may have affected the final recommendations.
The new study by Bilimoria et al, combined with the study by Qada et al. and a study that challenged the use of VTE as a marker of quality of care in trauma patients (Huseynova 2009) all raise the question about whether the VTE metrics included in current quality and pay-for-performance programs should be revisited.
Bilimoria KY, Chung J, Ju MH, et al. Evaluation of Surveillance Bias and the Validity of the Venous Thromboembolism Quality Measure. JAMA 2013; 310(14): 1482-1489
Livingston EH. Postoperative Venous Thromboembolic Disease: Prevention, Public Reporting, and Patient Protection. JAMA. 2013; 310(14): 1453-145
Qadan M, Polk HC, Hohmann SF, Fry DE. A reassessment of needs and practice patterns in pharmacologic prophylaxis of venous thromboembolism following elective major surgery. Ann Surg 2011; 253(2): 215-220
Neumann I, Akl EA, Valdes M, et al. Low Anonymous Voting Compliance With the Novel Policy for Managing Conflicts of Interest Implemented in the 9th Version of the American College of Chest Physicians Antithrombotic Guidelines. Chest. 2013; 144(4): 1111-1116
Huseynova K, Xiong W, Ray JG et al. Venous Thromboembolism as a Marker of Quality of Care in Trauma. J Am Coll Surg 2009; 208: 547-552
Print “PDF version”