• Rheumatoid Arthritis
• Psoriatic Arthritis
• Spondyloarthritis
• Epidemiology

We thank Dr Sykes and colleagues1 for their interest in our publication2 on changes in diagnostic delay in patients with inflammatory arthritis from year 2000 to year 2011 and for their reflections regarding our investigations. Based on data from the nationwide Danish DANBIO registry, we studied diagnostic delay defined as the time between onset of symptoms and the time of diagnosis in 13 721 patients with rheumatoid arthritis (RA), psoriatic arthritis (PSA) and ankylosing spondylitis (AS). We found that for RA, PSA and AS, the delay decreased from 30 months, 53 months and 66 months, respectively in year 2000 to 3–4 months for all three diagnoses in year 2011.

The prognosis for patients with RA is better with early diagnosis, which is reflected in national and international guidelines, and the principle of early referral has spread to PSA and AS. Little is, however, known about the extent to which these recommendations have transformed into earlier establishment of diagnosis in routine care.

Data for our study were collected as part of clinical practice when patients were first registered in DANBIO. The large number of patients is an important strength. As mentioned in the discussion, however, potential sources of bias in study design should be considered, and among those are left and right censorship and channelling bias. Consistent with this, Sykes and colleagues suggest a potential methodological problem with a sample of patients defined by diagnosis, rather than by symptoms. Patients with few symptoms may not seek specialist medical care and have not been captured in DANBIO. This is perhaps most important for patients with AS symptoms, where the majority of patients has mild disease. This relevant worry was addressed in the paper. To estimate the impact of channelling bias and the robustness of our results, we performed a sensitivity analysis in which we only included patients treated with biological drugs. Coverage of this group of patients has consistently been more than 90% in DANBIO, thereby practically eliminating channelling bias. Reassuringly, the key findings of significant reduction in diagnostic delay were confirmed for all three diagnoses in this subgroup.2 Our findings of shorter diagnostic delay were also reported by other research groups. In one study of 135 patients with AS, the diagnostic delay was reported to decrease from 13 years in 1970 to 2 years in year 2000.3 Our paper showed that among patients who seek medical assistance (ie, not those individuals with mild symptoms that require no treatment) the diagnostic delay has dropped substantially during the first decade of the new millennium. The decrease probably reflects a stronger awareness of the importance of early diagnosis.