• lupus erythematosus
• systemic
• epidemiology
• autoimmune diseases

We thank Huang and Zhang1 for their interest in our article entitled ‘Increased risk of systemic lupus erythematosus (SLE) in patients with autoimmune hemolytic anaemia (AIHA): a nationwide population-based cohort study’.2 In our article, we used nationwide longitudinal population-based data to explore the incidence of systemic lupus erythematosus (SLE) in patients with autoimmune haemolytic anaemia (AIHA) and demonstrated that patients with AIHA are at a very high risk of incidental SLE.2 Huang and Zhang1 have put forward valuable comments and opinions on our article.

First, Huang and Zhang1 proposed that patients with childhood-onset SLE may be more likely to develop AIHA than adulthood-onset patients, and suggested that we describe the incidence of AIHA in patients with SLE of different age groups. Given that it is impossible to provide such information using our original data, we conducted another nationwide cohort study to investigate the risk of AIHA in patients with SLE stratified by age groups (ie, <18 years and ≥18 years). Using the 1997–2013 Taiwanese National Health Insurance Research Database (NHIRD), we identified 6781 newly diagnosed patients with SLE from 2006 to 2013 and randomly selected 130 340 age, sex-matched (1:20) individuals without SLE from one million representative populations. From the cohort, we used propensity score matching (PSM) (1:2) for age, sex, comorbidities and potential confounders (online supplemental table 1). Before PSM, 123 (1.9%) patients with SLE and 1 (0.001%) individual without SLE had a history of AIHA (p<0.001). After excluding those with a history of AIHA, the incidence rates with 95% CIs of AIHA were 26.82 (26.80 to 26.83) cases per 100 000 person-months in patients with SLE and 0.046 (0.045 to 0.046) individuals without SLE (incidence rate ratio 88.68, 95% CI 185.81 to 1865.06) (table 1). After adjusting for potential confounders, including demographic variables, medical use and comorbidities, the risk of AIHA was markedly increased in patients with SLE (adjusted HR 550.63, 95% CI 172.64 to 1756.25) (online supplemental table 2). However, age was not associated with the risk of AIHA. Among patients with SLE before PSM, the incidence of AIHA was not significantly different between patients with childhood-onset SLE and patients with adulthood-onset SLE (table 2).

Second, Huang and Zhang1 also suggest that comparison be made on the incidences of SLE between patients with AIHA with idiopathic thrombocytopenic purpura (ITP), patients with AIHA without ITP and patients with ITP without AIHA. We found that of the 1:20 age-matched and sex-matched patients with AIHA (n=731), 56 (7.66%) had a history of ITP and 675 (92.34%) had no comorbid ITP. The incidences of SLE were 383.15 cases per 100 000 person-months and 404.47 cases per 100 000 person-months in patients with AIHA with and without ITP, respectively (incidence rate ratio 0.95, 95% CI 0.44 to 2.03). Among ITP patients without AIHA, Zhu et al 3 recently reported that the incidence of SLE was 52.45 cases per 100 000 person-months using the NHIRD. Also, we agree with Huang and Zhang1 that both AIHA and ITP are associated with antiphospholipid antibodies.4 Unfortunately, the NHIRD lacks laboratory data. Therefore, future human and animal studies were warranted to elucidate the mechanistic role of antiphospholipid antibodies on the development of AIHA and ITP.

Finally, Huang and Zhang1 were concerned with the validity of AIHA diagnosis by ICD-9 code and suggested checking the autoantibodies to confirm AIHA diagnosis. Given that the NHIRD lacks autoantibodies data, we only considered patients whose AIHA diagnosis was made during hospitalisation but not in outpatient visits only, as patients with AIHA, to improve the validity of AIHA diagnosis. This is also the reason why the case number of patients with AIHA in a nationwide dataset was relatively small, leading to a wide range of 95% CI of HR. We also acknowledge that the extremely high HR of SLE risk in patients with AIHA compared with individuals without AIHA can be explained, at least in part, that some patients with AIHA were actually in the early stage of SLE. We tried to improve the validity of AIHA diagnosis by the exclusion of those (n=36) who did not receive corticosteroid or immunosuppressant therapy withing 6 months after AIHA diagnosis and found a consistent result using the age-matched and sex-matched populations (adjusted HR 151.11, 95% CI 92.46 to 246.95). Furthermore, Huang and Zhang1 suggested stratified analyses based on different time intervals to investigate potential time-varying effects. According to the Kaplan-Meier curves that displayed the cumulative incidence of SLE, we examined the risks of SLE associated with AIHA during the time intervals within the first 2 years and after 2 years of follow-up. We found that in the propensity score-matched population, the HRs for SLE development were 202.97 (95% CI 28.28 to 1456.57) and 47.11 (95% CI 6.25 to 355.25) during the first 2 years and after 2 years, respectively.

In conclusion, patients with AIHA had increased risk of SLE particularly in the first 2 years. The risk did not differ between patients with AIHA with and without a history of ITP. We also found a markedly increased risk of AIHA in patients with SLE, and the risk was not significantly different between patients with childhood-onset SLE and patients with adulthood-onset SLE. We suggested that clinicians conduct comprehensive assessment regarding SLE-related clinical and laboratory characteristics to avoid delayed diagnosis of SLE and to improve patient care.