Challenging the Assumptions About Chemical Thromboprophylaxis in Knee Replacement

Commentary In this study, the authors performed a database-driven, population-based survey of the frequency of hospitalization with (or rehospitalization for) symptomatic deep-vein thrombosis (DVT) within ninety days of total knee arthroplasty. The setting was Denmark whose nationalized health system has little emigration and so little loss to follow-up. Denmark also boasts excellent administrative registries but permits lengths of stay after arthroplasty that are longer than those in the U.S.; patients in this cohort were hospitalized for an average of seven days after total knee arthroplasty. The major findings were that just over 1% of the patients were hospitalized for symptomatic thromboembolic disease within ninety days of surgery (three-quarters of those hospitalizations were for DVT and one-quarter, for pulmonary embolism) and that the risk of thromboembolic disease increased, rather than decreased, over the ten-year study period. The role of a written editorial or commentary in this setting generally is to offer a critique of the source material and to place that material into a broader context. My intention here is to sacrifice the former on the altar of the latter, at least to some degree. My reason for doing so will become clear in a moment. By way of brief critique, the article was well done. We can surmise that it was limited by the usual suspects known to beset registry studies, as the literature on this is fairly consistent (namely, a low, but non-zero, likelihood of administrative error and misclassification1, and an underestimation of the frequency of comorbid conditions2). A few important questions, which I thought could easily have been covered, were not answered and perhaps should have been; for example, did the frequency of the use of chemical thromboprophylactic agents change over the ten-year period of the study? This matters because the authors claimed a change in the frequency of symptomatic DVT over that time. Also, why did they attribute all study deaths to thromboembolism, when nearly every other paper that has looked at mortality has found most deaths in this context to be from other causes? Just the same, the major conclusions drawn were well supported by the data and the methods from which those data were derived. The context here is much more interesting. I could not help thinking, as I was reading this paper, how much fun it is to be a consumer of good-quality clinical research. This is not a sales pitch for JBJS. It is, rather, a sales pitch for being open to letting a well-done paper call into question the assumptions we carry with us as we read and as we practice. We may, after reading, conclude that our up-front beliefs were, in fact, right. Much less commonly, a paper may convince us that what we thought to be true was not correct. Most often, a paper will answer a few small questions but not do much to our major assumptions. Yet something about this paper tickled a number of commonly cited assumptions of mine and others so, as I mentioned earlier, it was fun to let the paper do some work on them. I provide three examples in the paragraphs that follow. By way of full intellectual disclosure, my general opinion on blood-clot prevention is that the chemical thromboprophylaxis recommended by the widely cited American College of Chest Physicians (ACCP) guideline is more aggressive than is necessary for most patients and results in more drug-related harm than it prevents in terms of serious thromboembolic events; therefore, I am supportive of the American Academy of Orthopaedic Surgeons (AAOS) guidelines on the topic, which allow for the use of aspirin as an alternative to stronger anticoagulants for many patients. My three examples are as follows. First: “Thromboembolic disease has become less of a clinical problem, as contemporary clinical pathways that include early mobilization have replaced earlier, less aggressive approaches to mobilizing after total knee arthroplasty.” We hear this frequently, particularly from the aspirin-for-DVT-prophylaxis crowd, of which I am a member. The findings of Pedersen et al. are in direct conflict with this idea; their study demonstrated that the relative risk of having a symptomatic episode of venous thromboembolism was substantially higher in 2007 than it was in 1997. There are possible explanations for, and limitations to, their finding, but it is pretty evident from their data (and from others’, as it turns out) that thromboembolic disease is not a smaller problem now than it was ten years ago. Second: “Aggressive chemical thromboprophylaxis is a ‘U.S. thing,’ driven by a medical malpractice system run amok; large swaths of the globe do fine with aspirin, or less.” This is another comment heard often in the aspirin crowds in which I run, crowds that now are marginally more populous since the AAOS guidelines gave an alternative to the ACCP rubric that leaves no room for the use of aspirin. The fact that, in this study from Denmark, a remarkable 83% of the patients surveyed received low-molecular-weight heparin of one form or another caused me to try to get to the bottom of this assumption. It turns out that it is much more nuanced than I thought. While good-quality international practice-pattern data are sparse, and sometimes dubious since they are mostly survey-derived, ACCP-compliant chemical thromboprophylaxis in patients having total knee arthroplasty actually is used slightly more often outside the U.S. than inside—69% outside compared with 61% inside the U.S.3—and even in England, where the use of aspirin is common, only 25% of patients are treated with aspirin rather than something stronger4. It is also interesting that, in Denmark, individual physicians do not face anything like the medicolegal exposure experienced by U.S. physicians, so it is unlikely that medicolegal concerns are driving the practice pattern seen in the Pedersen study, in which well over 80% of patients received aggressive chemical thromboprophylaxis. These surgeons are not prescribing these drugs for the lawyers. They are prescribing these drugs because they believe the drugs are right for patients having total knee arthroplasty, and there is literature to support that point of view. If those are two arguments against the aspirin crowd, here is one against the other crowd: “Aggressive chemical thromboprophylaxis will decrease the clinical impact of thromboembolic disease.” So says the ACCP, as do others. But in Pedersen's study, the patients had an increasing incidence of thromboembolism over the ten-year period, despite consistent (and probably increasing) use of low-molecular-weight heparin over that same time. More to the point, there are data—from nationalized health-care systems4, large clinical series5, as well as meta-analyses of thousands of patients6—showing that aspirin is as effective as the aggressive anticoagulation recommended by the ACCP guidelines at preventing life-threatening thromboembolic disease after total knee arthroplasty. Furthermore, the data may even be better if the end point studied is death (which takes into account not just clot prevention but also bleeding complications from the use of low-molecular-weight heparin), rather than simply the development of DVT. Support for this can be drawn from Pedersen's work in that symptomatic thromboembolic disease was not discernibly less frequent in this large group of patients having arthroplasty and receiving low-molecular-weight heparin than it has been in other large populations that were treated instead with aspirin4-6. While the ACCP guidelines are derived from randomized controlled trials, that same study design lacks an important component of external validity that registry data bring to the table. Data from high-quality, large, nonrandomized sources such as the study by Pedersen and other similar investigations4-6 are not considered in guidelines like those from the ACCP. Perhaps they should be.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.