The Lupus Midwest Network (LUMEN ) is a population-based registry of a 27-county region in southeast Minnesota and southwest Wisconsin, nested in the Rochester Epidemiology Project (REP), a medical records-linkage system (7–9). The REP allows ready access to the medical records from health care providers for the local population, including the Mayo Clinic, Olmsted Medical Center, their affiliated health systems and hospitals, local nursing homes, etc. The demographics, distribution of morbidity, and mortality rates in the REP region are similar to those in the upper Midwest (10). The characteristics and strengths of the REP, as well as its generalizability, have been described elsewhere (11–13). This study was approved by the institutional review boards of Mayo Clinic (20-006485) and Olmsted Medical Center (036-OMC-20).

Patient identification and ascertainment of the LUMEN registry have been previously published (7,8,14). Briefly, ascertainment of a potential SLE case was identified via two strategies: 1) through International Classification of Diseases 9 (ICD-9) and ICD-10 codes for SLE, cutaneous lupus erythematosus, and other associated diseases, and 2) through laboratory measures associated with SLE—anti-nuclear antibodies (>1:80), low complement, anti-double stranded DNA, anti-Sm, lupus anticoagulant, anticardiolipin (IgG, IgM, and IgA), and anti-beta 2 glycoprotein 1 (IgG, IgM, and IgA) antibodies. We identified all the potential SLE cases in the REP 27-county region, from 1976 to 2018 for Olmsted County, and from 2010 to 2018 in the other 26 counties. We included as cases those patients with SLE who fulfilled the 2019 European Alliance of Associations for Rheumatology (EULAR)/American College of Rheumatology (ACR) classification criteria (15). We used the 2019 EULAR/ACR criteria since they perform better than other classification criteria at identifying patients in population-based studies (16). To be considered incident, patients needed to be residents of the 27-county region on the date of criteria fulfillment.

We electronically retrieved demographic variables and self-reported race and ethnicity (Hispanic; non-Hispanic White, Asian, and Black). Educational level and disease manifestations based on the definitions of the 2019 EULAR/ACR classification criteria were manually abstracted from medical records. To understand how patients seek medical care, we identified the medical encounters that were related to SLE and abstracted the dates they occurred. We defined an SLE-related encounter as inpatient or outpatient medical visits during which care for SLE manifestations was provided. Encounters wherein clinical findings could be better explained by conditions other than SLE were not counted (e.g., joint swelling related to trauma, a dermatology visit for a mole check). We considered that a patient received care for SLE if disease manifestations were documented in the medical history, physical examination, or clinical interventions. Inpatient encounters were defined as those requiring at least 1 overnight stay, and each hospitalization was considered a single encounter.

We abstracted the medical specialties of the clinicians involved in SLE-related encounters and grouped them as “SLE-specialized” (Rheumatology, Dermatology, and Nephrology), Internal Medicine and subspecialties (excluding SLE-specialized), Family Medicine, Emergency Medicine, surgical specialties, and other specialties.

We abstracted the dates of the medical encounters wherein a physician diagnosis of SLE and SLE treatment were given for the first time. SLE treatment was based on prescriptions for antimalarials (chloroquine or hydroxychloroquine), or conventional (azathioprine, cyclophosphamide, methotrexate, mycophenolate, or tacrolimus) or biologic immunosuppressants (rituximab or belimumab). Patients prescribed glucocorticoids, nonsteroidal anti-inflammatory drugs, or topical medications as monotherapy were not considered to have received SLE treatment. We determined the Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2k) (17) and the Systemic Lupus International Collaborating Clinics (SLICC)/ACR Damage Index (SDI) (18) at the time patients had an encounter with SLE specialties. If patients did not access SLE specialties, SLEDAI-2k and SDI were obtained when the patient fulfilled the classification criteria.

Three physician abstractors (ASR, JAMA, JXY) underwent an initial calibration phase with a sample of 10% of patients. During several rounds of calibration, a >95% of agreement was considered reliable throughout the abstraction process. Variables subjected to judgment calibration were the dates of the first medical encounter, the number of medical encounters, characteristics of specialties seen, and the total number of encounters. The abstraction was completed using the Research Electronic Data Capture (REDCap) online platform (19,20).

To outline the pathways in SLE patients’ journeys, we tracked key events, specifically the first: SLE-related encounter, SLE-specialized encounter, SLE diagnosis encounter, and SLE therapy encounter. Patients were followed from the first SLE-related encounter until we identified these three key events: migration out of the 27-county region, death, or until February 11, 2020. We measured the time in months and summarized all encounters between the first SLE-related encounter and the other key events in care.

We defined a delay in access to SLE-specialized care as a period of 6 months or more between the first SLE-related encounter and the first SLE-specialized encounter. Delays were dichotomized at six months based on prior evidence showing that patients diagnosed within six months from symptom onset have improved prognosis.(5)

Descriptive statistics were used to summarize demographic and clinical variables. SLE manifestations seen up to the first SLE-specialized care encounter were presented overall and according to the type of specialized care received. Among patients who never received specialized care, SLE manifestations through the most recent encounter were reported. Using chi-square and rank sum tests, we compared characteristics identified during the initial encounter and assessed differences in the SLEDAI-2k and SDI between patients with and without delays in accessing specialized care. Fisher’s exact tests were used to compare categorical variables with expected cell counts less than five. We used Kaplan-Meier methods to estimate median time from first SLE-related encounter to reaching each additional key event, to appropriately account for patients who had not yet reached a key event at last follow-up. Log-rank tests were used to compare median times. Logistic regression models were used to estimate associations between characteristics and delays in access to specialized care—adjusted for age, sex, and calendar year of SLE incidence, with results summarized as odds ratios (OR) and 95% confidence intervals (CIs). Cox models were not used because dichotomizing delays at 6 months was considered to be clinically relevant and all patients had at least 6 months of follow-up, so there was no censoring. Complete-case analysis was used in all cases. P values less than 0.05 were considered statistically significant for all analyses. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

Among the 373 patients included in this study, 2% (8/373) had no access to SLE-specialized care during follow-up (median 55, range 8–114 months). Of the remaining 365 patients, 274 (75%) had their first SLE-specialized care with Rheumatology, 59 (16%) Dermatology, and 32 (9%) Nephrology. However, as previously mentioned, 44 (12%) accessed SLE-specialized care during their first SLE-related encounter, while the remaining 329 (88%) required multiple consultations before first accessing SLE-specialized care. The median time to access SLE-specialized care was 1.1 months (95% CI 0.9–1.5), with patients typically requiring a median of 3 encounters (interquartile range [IQR] 2–4) to achieve this access (Supplementary Table 2).

SLE manifestations (by EULAR/ACR domains) seen up to the first SLE-specialized care encounter are outlined in Table 2. Patients first visiting Rheumatology exhibited musculoskeletal manifestations (46%), patients visiting Dermatology more frequently had mucocutaneous manifestations (59%), and patients who first visited Nephrology had more renal manifestations (34%). The rest of the clinical domains were more evenly distributed across the three SLE-specialized care services.

A total of 284 patients received an SLE diagnosis by a physician during follow-up, while the rest received other diagnoses including undifferentiated connective tissue disease or inflammatory arthritis (23%), yet all fulfilled the EULAR/ACR criteria for SLE. The median time from the first SLE-related encounter to receiving an SLE diagnosis was 30.6 (95% CI 18.9–48.1) months; those diagnosed with SLE had a median of 6 (IQR 3–12) encounters between these two times. Moreover, 338 patients received SLE treatment during follow-up. The median duration from the first SLE-related encounter to the initiation of SLE treatment was 4.7 months (95% CI 3.9–8.4), involving 5 (IQR 3–9) encounters (Supplementary Table 2).

When comparing the first SLE-related encounters between inpatient and outpatient settings, patients initially assessed in inpatient care more frequently presented with the following domains: constitutional (11% vs 1%, P < 0.001), hematologic (42% vs 28%, P = 0.035), serosal (22% vs 3%, P < 0.001), and renal (8% vs 1%, P = 0.005) (Supplementary Table 3). Patients initially assessed in outpatient care more frequently presented with musculoskeletal domain manifestations (28% vs 11%, P = 0.004). Additionally, half of the patients (52%) in inpatient care for their first SLE-related encounter were seen by a specialist in Emergency Medicine. Patients hospitalized at first SLE-related encounter also showed higher disease activity at first access to SLE-specialized care compared to those who had their first SLE-related encounter in an outpatient setting (median SLEDAI-2k of 8 [IQR 5–14] vs 4 [IQR 2–7], P < 0.001) and were more likely to present with damage at first access to SLE-specialized care (SDI ≥1 in 20% vs 11%, P = 0.048). Although the time needed to access SLE-specialized care did not differ statistically, the median time for receiving an SLE diagnosis (40.8 [95% CI 28.9–60.2] vs 3.4 [95% CI 1.2–11.1] months, P < 0.001) and starting SLE treatment (6.9 [95% CI 4.1–10.2] vs 2.3 [95% CI 1.0–4.8] months, P = 0.021) was significantly shorter for those who were initially hospitalized (Supplementary Table 2).

A total of 75% (280/373) of the patients with SLE gained access to SLE-specialized care within 6 months, while 25% (93/373) experienced a delay in access or never achieved access during follow-up. As detailed in Table 3, there were no differences in demographics or education levels among those with or without delay in access to SLE-specialized care. More patients with delay had manifestations in the hematologic and antiphospholipid antibody domains and were affected in only one SLE domain at the first SLE-related encounter, compared with patients without delay (92% vs 72%, P < 0.001). More patients without delay had manifestations in the mucocutaneous, musculoskeletal, specific autoantibodies, and complement domains. The delay in access to SLE-specialized care was associated with a 5-fold increased likelihood of presenting with at least one point on the SDI score at first access to SLE-specialized care (30% vs 7%, adjusted OR 5.30, 95% CI 2.71–10.37, data not tabulated). However, delay in access to SLE-specialized care was not associated with increased disease activity at the time the patient first accessed SLE-specialized care.

After adjusting for age, sex, and calendar year, an initial presentation with hematologic domain was associated with a nearly 2-fold increase in odds of experiencing a delay in SLE specialty care (OR 1.71, 95% CI 1.03–2.84), and antiphospholipid antibodies domain was associated with a 6-fold increase in odds for delay (OR 6.05, 95% CI 2.46–14.88). Conversely, those who initially presented with mucocutaneous (OR 0.17, 95% CI 0.06–0.44), musculoskeletal (OR 0.47, 95% CI 0.25–0.88), specific autoantibodies (OR 0.44, 95% CI 0.23–0.82), or ≥2 SLE domains (OR 0.23, 95% CI 0.10–0.52) were associated with a decreased likelihood of delay. We observed a decreased likelihood of delays over time; specifically, each 10-year increase at the initial SLE-related encounter was associated with a lower risk of delays (OR 0.73, 95% CI 0.60–0.88), and the median time to gain access, which was 2.7 months (95% CI 0.7–8.6) during 1989–1998, decreased to 1 month (95% CI 0.8–1.5) during 2009–2018 (Supplementary Table 2). Age, sex, race and ethnicity, educational level, initial care received in an inpatient or outpatient setting for the first SLE-related encounter, or type of medical specialties at the first SLE-related encounter were not associated with delays (Table 4).