Letter by Spyropoulos Regarding Article, “Physician Alerts to Prevent Symptomatic Venous Thromboembolism in Hospitalized Patients”
To the Editor:
The article by Piazza et al published recently in Circulation found that the use of physician alerts versus a no-alert system in hospital wards did not lead to a significant reduction in symptomatic venous thromboembolic events (VTE), although it did increase the use of thromboprophylaxis.1 These results contrasted with the authors’ previous study using an electronic alert system, in which statistically significant differences were found for both increased uptake of thromboprophylaxis and decreased VTE events, favoring the intervention group.2
A careful look at the data reveals that a sizeable proportion of deep vein thrombosis (DVT) (8 of 19 events [42%] in the intervention group and 8 of 24 events [33%] in the control group) included upper-extremity disease. Removing upper-extremity events using a 2×2 design does not change the results appreciably (1.94% versus 2.63%, relative risk 0.74 [95% confidence interval 0.44 to 1.24]). However, despite these results, the inclusion of upper-extremity disease in VTE outcome events reveals a potential methodological flaw.
The distinction of upper-extremity versus lower-extremity DVT is an important one, and the decision to include upper-extremity disease in the design of intervention studies on thromboprophylaxis has important methodological consequences. It is well established that the pathophysiology of upper-extremity versus lower-extremity DVT is quite different.3 In addition, although the natural history of lower-extremity DVT is well established in the surgical patient (less so in the medical patient), the natural history and risk of embolization of upper-extremity DVT is a matter of debate.4 Most importantly, whereas the use of mechanical or pharmacological methods of thromboprophylaxis has established efficacy for the prevention of lower-extremity DVT and pulmonary embolism (depending on patient and prophylaxis type), the use of these methods in preventing upper-extremity disease is unknown or uncertain.5
It is possible that the magnitude of the effect of thromboprophylaxis was underestimated in the study design used by Piazza et al. Because upper-extremity DVT constituted a sizeable proportion of the overall VTE events, the use of any thromboprophylactic intervention would not have decreased the risk of upper-extremity disease in this (mostly cancer) patient population because of the fact that the use of thromboprophylaxis has not been conclusively shown to be efficacious in preventing upper-extremity DVT. In addition, close inspection of the events based on the previous weighted risk score by Kucher et al reveals that more than half of the events (32 of 62) in the intervention group and nearly one third of events (33 of 103) in the control group consisted of DVT of the arms, also calling into question the value of that particular intervention and the VTE risk score on which that intervention was based.2
It is the opinion of this author that the design of thromboprophylactic intervention studies and VTE risk scores and analyses derived from them should be based on definitions of VTE outcomes that explicitly exclude upper-extremity DVT, because of the differences in the pathophysiology and natural history of disease between upper- and lower-extremity DVT and the uncertainty of thromboprophylactic measures in the setting of upper-extremity disease. This opinion is also shared by the Joint Commission as it considers the removal of upper-extremity DVT in the definitions of (potentially) preventable hospital-acquired DVT.
Piazza G, Rosenbaum EJ, Pendergast W, Jacobson JO, Pendleton RC, McLaren GD, Elliott CG, Stevens SM, Patton WF, Dabbagh O, Paterno MD, Catapane E, Li Z, Goldhaber SZ. Physician alerts to prevent symptomatic venous thromboembolism in hospitalized patients. Circulation. 2009; 119: 2196–2201.