In the present cohort, extRA manifestations had a considerable prevalence in RA patients. Persistently active and severe disease with high RF titres and established sequelae are risk factors for extRA. These findings are similar to previously reported data [6, 11,12,13]. However, unlike previous findings, we did not observe a correlation with ACPA positivity, probably due to the test’s unavailability to almost half of the patients studied. Likewise, there was no significant correlation with smoking status, except for rheumatoid nodules [13, 14]. In this specific group, smoking habit (current and ever, summed up) was more prevalent in comparison to the overall ExtRA patients (66.3 vs. 51.7%).

Although smoking is a known risk factor for extra-articular manifestations, the role of active smoking is more pronounced for this outcome than previous smoking, as demonstrated by Turesson [13], who found a relative risk of 1.52 for the development of ExtRA in active smokers at RA diagnosis. The role of ever-smoking (not at RA diagnosis) was not so well defined for ExtRA. In our cohort, the prevalence of ever-smokers was much higher than active smokers (42.1% vs. 9.6% in ExtRA patients), which can partially explain our findings. We did not specify in our questionnaire if smoking habit was present at RA diagnosis, what could be a bias. Additionally, our cohort mainly consists of older patients presenting long-term, RF-positive, erosive disease, and we believe that these other risk factors overcame smoking risk itself.

Regarding pharmacological treatment, the higher use of glucocorticoids in patients with extRA can be explained both by the greater inflammatory activity and severity of joint disease, as well as by the extRA treatment itself. Additionally, the greater use of NSAIDs is also justified by articular activity, chronic pain and damage.

Interestingly, a trend towards greater use of anti-tumour necrosis factor (TNF) drugs in the extRA group was observed, with statistical significance in favour of etanercept. One hypothesis for this finding is disease severity, which requires biological treatment for inflammatory control. In addition, when the cohort was initiated, anti-TNF drugs had a greater availability in Brazil compared to other biologics. Factors that could favour etanercept could include safety concerns, such as the lower risk for certain infections [18] (herpes zoster [19] and tuberculosis [18]) and the shorter half-life, considering the extRA patient’s profile (older, disabled, with more severe disease and more comorbidities).

When we examined the use of conventional synthetic DMARD treatment, there was a greater proportion of patients using azathioprine (AZA) in the extRA group. Presumably, this drug was chosen for extRA control, because the effectiveness of AZA for inflammatory articular activity is limited [20]. Considering the widespread AZA use for RA-ILD in Brazil, a comparison between the non-ILD ExtRA summed up, versus no-ExtRA, failed to show significant difference regarding AZA use, what may imply a preference for AZA in RA-ILD clinical management (influenced also by the drug wide availability in Brazil). On the other hand, a lower LEF use was detected in the non-ILD extRA group, probably because this csDMARD is usually avoided in the presence of neuropathy and vasculitis.

Further subgroup analysis regarding the most frequent and/or clinically significant manifestations (sicca syndrome, subcutaneous nodules, vasculitis) did not reveal differences between the extRA and control groups. However, patients with ILD presented a relevant difference that should be explored: the lower use of MTX.

ILD is a topic of growing interest because it currently represents the second leading cause of death among patients with RA and because of its novel antifibrotic therapeutic effects [21, 22]. The incidence and morbimortality of RA-ILD have grown over the years [22, 23]. Clinically significant ILD occurs in up to 10% of RA cases [23, 24]. In our cohort, ILD occurrence was lower, reported in 4% of the patients. Considering that access to diagnostic tools (such as high-resolution chest tomography) varies significantly between different regions of the country, underdiagnosis and heterogeneous data may explain this finding. Another contributing factor for this scenario is the lack of specific guidelines for ILD detection and treatment in RA patients [25].

We found that the use of MTX was less frequent in patients with ILD in Brazil. This finding is similar to a recent multicentre study [26]. Avoiding MTX in ILD has been a widespread practice for many years in rheumatology, as it is believed that MTX could exacerbate ILD. Currently, this misconception has been elucidated: MTX probably has a protective role against RA-ILD development, either by a direct immunosuppressive effect targeting the lung or by an MTX-driven decrease in systemic inflammation [26].

The REAL study [17], an observational multicentre cohort, has a limitation in its design. All enrolled sites are tertiary centres, presenting a severe disease profile predominance, and probably not fully representative of the broader management of RA across the country. Additionally, the cross-sectional nature of the study design precludes establishing causal association between variables and outcomes.