This cross-sectional study used data from a multicenter cohort (Lupus Registry of Nationwide Institutions, LUNA) established in 2016 to examine clinical manifestations, social background, and outcomes in patients with SLE from 15 institutions across Japan, ranging from the Tohoku region to the Kyushu region. LUNA provides data on patients aged ≥ 20 years diagnosed with SLE according to the revised 1997 American College of Rheumatology (ACR) classification criteria [8]. Patients deemed ineligible by the investigator and those who found it difficult to complete the questionnaire, such as patients with dementia, were excluded. Approximately 1700 cases (2.5% of Japanese patients with SLE) have been registered in LUNA.

This analysis used data obtained from medical records from July 2019 to March 2021. The data included laboratory tests, medications, activity scores, and comorbidities. Quality of life and comorbidity were collected from self-administered questionnaires. Data were collected annually for each patient through the LUNA. Although multiple surveys were conducted during the course of the study for a patient, only the most recent data were included.

Patients who provided the number of attending physicians enrolled in LUNA were eligible to participate in this study. Patients with a missing Systemic Lupus International Collaborating Clinics/ACR damage index (SDI) were excluded from the study. Patients with a disease duration of less than 6 months were excluded, since the primary outcome, SDI, required a minimum of 6 months for irreversible manifestation. Patients with more than 10 years of disease were also excluded because of recall bias. The disease duration was evaluated using patient questionnaires, which may have differed from the true disease duration.

Exposure was defined as the number of attending physicians. The patients provided this item using a questionnaire. The questionnaire item was “Is your SLE disease duration less than 10 years? If yes, please answer the following questions. How many attending physicians have treated you in an outpatient visit since you were diagnosed with SLE? Please include the attending physician at the hospital prior to transfer.”

The primary outcome was SDI, which measures cumulative damage since the onset of SLE. SDI reflects irreversible damage lasting more than 6 months in various systems [9]. The total score is 47 points. We divided SDI into two categories: 0 and 1 or more points, in accordance with a previous study [10]. The secondary outcomes were corticosteroid-related damage and non-corticosteroid-related damage. Corticosteroid-related damage consists of ocular cataract, osteoporosis with fracture or vertebral collapse, avascular necrosis, or diabetes [9, 11]. Non-corticosteroid-related damage consists of SDI items other than corticosteroid-related damage. We also divided both types of damage into two categories: 0 and 1 or more points, in accordance with a previous study [10].

The following factors were used as potential confounders: age, sex, number of hospitalizations due to SLE, systemic lupus erythematosus disease activity index (SLEDAI) at diagnosis, disease duration, and emotional health. Emotional health is one of the domains of Lupus PRO, a disease-specific quality of life measurement [12, 13]. These variables were selected based on those previously reported for SDI and changes in attending physicians [1, 4, 5, 7]. Furthermore, a directed acyclic graph was constructed to show the relationships between these variables (Supplementary Fig. S1).

Descriptive statistics are presented as median [interquartile range (IQR)] for continuous variables and as absolute numbers (percentage) for categorical variables. Subsequently, we performed multiple logistic regression analysis to assess the associations between exposure and outcome variables (primary and secondary) to adjust for the potential confounders mentioned above. We also performed a sensitivity analysis that excluded patients with a large number of attending physicians (30 or 50). We performed multiple imputations on the assumption of missing values at random to deal with the missing values of potential confounders. The results of 100 imputed datasets were averaged, and the standard error was adjusted to account for variability within and between imputations. The estimates and their standard errors were combined using Rubin’s rules. A two-sided p value of less than 0.05 was considered to indicate a statistically significant difference. All statistical analyses were conducted using STATA 16 software (StataCorp).

The study was approved by the Ethics Committee of the Showa University School of Medicine (authorization number 22–082-A) and the institutional review boards or ethics committees of each participating hospital. Written informed consent was obtained from all patients. Before analysis, patient data were anonymized and deidentified. The procedures for this study were conducted in accordance with the Declaration of Helsinki and the Ethics Guidelines for Medical and Health Research Involving Human Subjects in Japan. Ethics committee names and reference numbers are listed in Supplementary Table S3.