Background There is compelling evidence from the general population that acute inflammation promotes a hypercoagulable state that may increase the risk of pulmonary embolism (PE). There are few studies on how this theoretical risk of venous thromboembolism translates into risk for clinical events in inflammatory diseases. In a recent study in the Lancet, Zöller et al used Swedish hospitalization data for RA and PE to inform on risks and time-trends for PE in RA. Based on an observed six-fold increase <1 year from first hospitalization listing RA, which declined to near unity after >10 years of followup, the authors and a commentary suggested that thrombo-prophylaxis should be considered in patients with autoimmune disease. However, using hospital discharge data to identify study populations may not only identify a selected subset of patients, but also at time-points when they are at high risk of PE, for reasons that may be unrelated to the inflammatory disease but associated with PE. Such bias is further augmented if comparators are not, as was the case in the Lancet study, required to be hospitalized. The result, an overestimated immediate but then decreasing risk, may lead to false conclusions regarding the underlying biology of PE in RA, and hence to unnecessary medical risks related to inappropriate clinical intervention, especially in populations such as RA where drugs affecting bleeding risks are frequently used.
Objectives To estimate the rate and relative risk (RR) of PE in contemporary cohorts of RA. Particular attention was paid to the development of risks in relation to time of followup and to risks in relation to hospitalization.
Methods We assembled one incident RA-cohort diagnosed between 1997 and 2009 (n=8088) and one nationwide prevalent RA-cohort followed at Swedish rheumatology clinics between 2005 and 2009 (n=39753) neither of which was identified based on hospitalization data. Each cohort member was matched to a general population comparator. Information on first time hospitalizations for PE through 2009 was retrieved from the Swedish Patient register. RR and 95% confidence interval (CI) were estimated using Cox models.
Results In prevalent RA, the RR of PE was 2.17 (95%CI 1.93-2.45). In incident RA, the overall RR for PE was 1.5 (95%CI 1.2-1.8). When stratified by RA-disease duration, increased risk of PE in the incident cohort was detectable already 1 year after RA-diagnosis (RR>1 year: 1.8 (95% CI 1.2-2.6) and did not change thereafter. Within the general population comparator, hospitalization was associated with an increased risk of PE (RR 4.7, 95%CI 3.1-7.3). When comparing hospitalized RA with hospitalized general population comparators, the risk increase (RR=1.6, 95%CI 1.1-2.1) was similar to that observed for the entire RA population.
Conclusions RA-patients are at increased risk of PE. This risk does not vary appreciably with RA-duration. Hospitalization, a strong risk factor for PE in the general population, does not modify the increased risk of PE in RA. The recently reported dramatic time-trends thus reflect a biased design.
Disclosure of Interest None Declared