A large Danish registry study shows that the risk of hemorrhagic events is doubled when an NSAID is prescribed to a patient taking oral anticoagulant for venous thromboembolism. An association to avoid!
The bleeding risk associated with the use of nonsteroidal anti-inflammatory drugs (NSAIDs) in patients treated with oral anticoagulants for venous thromboembolism (VTE) is not clearly established.
Current uncertainties and questions
VKAs and direct oral anticoagulants (DOACs) currently available increase the risk of bleeding, most often digestive hemorrhages, intracerebral hemorrhages being much rarer but more often fatal. Their severity varies, but their occurrence is most often problematic. Their association with NSAIDs is generally not recommended, even if there is uncertainty about the risk of bleeding.
It must indeed be recognized that the studies which are convincing are both rare and limited, relating to small numbers of people exposed most often to a particular DOAC, such as rivaroxaban.
Given their mechanism of action, NSAIDs prescribed long-term in the general population are undeniably the cause of hemorrhagic accidents, most often digestive bleeding. These accidents would be four times more frequent than in unexposed subjects, independently of any association with anticoagulants.
Questions remain unanswered: associated with NSAIDs, are AODs less often involved in these hemorrhagic accidents than AVKs, according to certain popular hypotheses? In the case of VTE, what is the magnitude of the bleeding risk? What are the hemorrhagic complications most frequently caused by the combination of an anticoagulant and an NSAID? What are the most “dangerous” NSAIDs?
A Danish retrospective registry study: more than 50,000 cases of VTE
A national retrospective cohort study conducted in Denmark can answer most of them. Were included 51,794 patients suffering from proven VTE and treated with oral anticoagulants between 1is January 2012 and December 31, 2022, all identified from national registers known for their rigor. The identification of the NSAID prescription was done using the Danish prescription register.
During the study period, bleeding episodes requiring hospitalization were counted in the groups exposed or not to NSAIDs. Data were processed using the Cox proportional hazards model, with adjustments accounting for the most potential confounders.
The frequency of bleeding events per 100 subject-years was estimated at 3.5 [IC 95 %, 3,4-3,7] in the unexposed group, versus 6,3 [5,1-7,9] in the exposed group, which leads to an adjusted hazard ratio (aHR) of 2.09 [1,67-2,62] for all pharmacological classes. The figures vary somewhat from one drug to another: respectively 1.79 [1,36-2,36] for ibuprofen, 3.30 [1,82-5,97] for diclofenac and 4.10 [2,13-7,91] for naproxen. In fact, ibuprofen would be the NSAID potentially least involved in the risk of bleeding.
The risk of intracranial hemorrhage appeared particularly high in the NSAID group, the corresponding HRa being estimated at 3.22 [1,69-6,14]in front of digestive hemorrhages (2.24 [1,61–3,11]), hemoptysis and thoracic hemorrhage (1.36 [0,67-2,77]), hematuria (1.57 [0,98-2,51]). The risk of anemia due to bleeding, for its part, appeared almost three times higher in the exposed group (HRa = 2.99 [1]).
The results were found to be independent of the type of anticoagulant (DOAC versusAVK) and manifestations of VTE, whether deep vein thrombosis or pulmonary embolism.
It would be better to do without it!
This study, which covers more than 50,000 participants suffering from VTE, is nonetheless retrospective, but the reliability of medical databases in a country like Denmark allows this limitation to be put into perspective. Unsurprisingly, the combination of anticoagulants/NSAIDs, which is hardly encouraged in current medical practice, would double the bleeding risk. Contrary to all expectations, intracerebral hemorrhages with a poor prognosis come first, narrowly ahead of digestive hemorrhages.
Ibuprofen would be the NSAID which exposes the least to the risk of bleeding, subject to a significant margin of error in the estimates. Naproxen, for its part, would be the worst on this point. Furthermore, no significant difference was highlighted between AVK and DOAC in terms of the bleeding risk incurred when an NSAID was included in the prescription.
To conclude, in the case of VTE treated with any anticoagulants, the wisest attitude is to do without NSAIDs as much as possible.
