Multicenter retrospective hospital-based registry of adult patients with SLE, with electronic collection of data from medical records review.

The main objectives of RELESSER-T are:

• (1)

  To describe the sociodemographic and clinical and laboratory characteristics of patients with SLE and incomplete SLE (LESI) (< 4 ACR criteria) seen in rheumatology departments in our country, especially those with unusual clinical manifestations (<1%), as well as the status of the disease, severity, treatment, complications and morbidity.

• (2)

  To establish a comprehensive and consistent record of well-characterized patients, from which to select subgroups on which to develop longitudinal studies that allow us to answer different research questions, constituting RELESSER-PROS.

As secondary objectives, the registry is to contribute to improve the level of instruction on systematic and standardized assessment of the disease, its implementation in daily clinical practice and promote cooperative multicenter clinical SLE research among Spanish rheumatologists.

Unselected patients ≥16 years, diagnosed with SLE (regardless of vital status), according to the revised criteria (1997).30,31 ACR, considered “defined SLE” (Lesd), or patients with only 3 criteria but SLE diagnosed according to the clinical judgment of an expert rheumatologist, called Lesi in this record. All patients were from the hospital rheumatology services of the public health system (except one private center) with active members of EAS-SER. We planned to include at least 80% of patients in follow-up (with more than one visit to rheumatology) at some point in each center. We excluded those patients who did not have least 50% of the data, a criteria defined as “minimum essential data” (comprising a total of 151 variables) (Table 1). The recruitment period was set at 10 months. In order to minimize missing data (“missing values”) and optimize the representation, a pre-screening visit was established 3 months before the start of enrollment period, which allowed us to complete the census of patients in each hospital, recovering SLE patients not identified as such and supplementing the local databases, making missing values recoverable.

Data collection in each center was made by rheumatologists with experience in the diagnosis and treatment of patients with SLE. To ensure consistency in the collection, we created an operations manual for the transversal registry phase (procedures, instructions for the proper use of indexes, glossaries with precise definitions of each variable, the instruction and explanatory remarks for the manual and electronic database, etc.), and carried out a preliminary exercise, specifically designed for recording, via the Internet, the way in which we evaluated the agreement in the use of assessment indexes of disease among different researchers and the gold standard, using clinical cases. The gold standard for correct responses was established by the two principal investigators (PI) (IRF and JMPR), both certified in the use of these indices by the Lupus Foundation of America, following a review of the same by the scientific committee of the registry. Unsolvable discrepancies using the definition of the items were resolved by consensus between the PI and the contributing researcher. We also established a forum, via the Internet, which allows direct interaction with the PI. The source document has been the clinical history of patients.

We created an online electronic application, specific for the registry, without any data collection on notebooks or paper, in order to minimize transcription errors and speed up the analysis process. We have avoided as far as possible free text responses and composite scores, the latter being directly performed by the application. The application contains filters, non-surpassable ranges, help menus and dialogs to maximize the reliability of the data. Prior to the start of data collection under the supervision of the UI-SER, we conducted a pilot study that involved five centers outside those of the PIs, reviewing 4 patients per center, with the purpose of testing the platform and identifying possible errors, inconsistencies or clarifying uncertainties that may arise during use.

The first patient was included on 27-10-2011 and the last on 13-08-2012, with effective inclusion having a duration of 10 months. Once entered, the data were reviewed and edited and we proceeded to block the electronic data collection notebooks to prevent further editing.

This was performed via the Internet, conducted by a qualified, accredited instructor, experienced in the field of rheumatology. For this, and using the statistical program Stata®, inconsistencies were monitored, described as missing (e.g. Age, Gender), logical (e.g. active monitoring state but with a cause of death), reviewed between different modules of the electronic notebook, inclusion criteria (e.g. <16) and strict adherence to the criteria for the SLE damage index, activity, severity and other definitions in the protocol. Everything has been reviewed by the PI and agreed upon with the monitor after several inconsistencies detection reviews by the panel. The inconsistencies of all centers were sent as discrepancies to the investigators concerned and were followed up until complete resolution.

We included 359 variables per patient, grouped as follows: (1) demographics, (2) chronological, (3) general clinical data, including vital status, (4) cumulative manifestations of SLE, defined by the glossaries of the ACR criteria for classification of SLE30,31 and the activity indexes Systemic Lupus Erythematosus Disease Activity Index (SLEDAI)32 and British Isles Lupus Assessment Group (BILAG) index33,34 and the damage Systemic Lupus International Collaborative Clinics/American College of Rheumatology Damage Index (SLICC/ACR DI),35 (5) the general SLE activity index, Systemic Safety of Estrogens in Lupus Erythematosus National Assessment (SELENA)-SLEDAI,36 damage index (SLICC/ACR DI)35 and severity index (Katz index) (IGK),37 (6) rare events (<1%), (7) comorbidity, including infections requiring hospitalization or cause of death (causal agent, location and treatment at the time of infection) and the Charlson comorbidity index, as amended by Deyo,38 (8) laboratory testing, (9) imaging or histological evidence when they were needed; (10) treatments, cause of discontinuation as per case (clinical problem solved, ineffectiveness, or other adverse reaction) (Table 1).

Antiphospholipid syndrome was defined according to the Sydney criteria,39 mixed connective tissue disease by the criteria proposed by Alarcon-Segovia40 and Sjögren's syndrome in the presence of compatible sicca syndrome and positive Schirmer test, or biopsy/salivary gland scintigraphy. SLE was considered refractory if one or more of the following criteria was satisfied: inefficiency of cyclophosphamide, ineffectiveness of at least 2 of the following immunosuppressive treatments: methotrexate, leflunomide, azathioprine, mycophenolate, Rituximab or splenectomy, at any time during the course of the disease and whatever manifestation had led to their use. Unusual events were recorded (prevalence <1% in the literature), not included in the aforementioned definition, by variables per device list plus a free text section.

The information was collected as present if it had been at any time of evolution (cumulative incidence). In the case of treatments, we also collected them if present at last available assessment. We collected the presence of the variables that constitute the SELENA-SLEDAI activity index at any time during the course of the disease and in the 10 days prior to the latest assessment of the patient.

We created a codebook with clear and unambiguous operational definitions, to guide the coding process and to facilitate the location of variables and interpretation of the data during analysis.

Based on a prevalence of approximately 1% (the minimum expected between the manifestations of SLE that we attempted to estimate), an accuracy of 2% and a confidence interval of 95% we calculated a minimum sample size of 815 patients. However, we decided to include the largest possible number of patients in each hospital in order to get a record that was as comprehensive as possible. We estimated, based on a previous survey (on possible losses and other variables), a sample of 2500 target patients, sufficient to achieve the main objective of the registry and enable subgroup analyzes.

We conducted a basic descriptive analysis using tables of absolute and relative frequencies of qualitative variables, position measurements means, median and measures of dispersion (SD: standard deviation, IR: interquartile range).

The project complies with the principles of the Declaration of Helsinki41 and the Oviedo Convention.42 Confidentiality has been respected in accordance with Royal Decree 1720/2007 and Law 15/199943 for Data Protection. The project has obtained authorization from the managements of the participating centers and approval by a Research Ethics Committee (CEIC) and the reference CEIC each center, as per required. All researchers involved have signed a document showing the commitment of the researcher.

45 hospitals were involved, with representation from almost all of the regions (Fig. 1) care levels and dimensions. The average number of patients included per center was 89±99, range: 1–534. Most centers (n=41.91%)≤ contributed 5% of the hospital patients and only one exceeded 10% of the total (Table 2).

Figure 1.

Distribution of participants in RELESSER centers, by regions.

(0.25MB).

4024 patients were included: 3679 (91%) with SLE and 345 (9%) with Lesi. 3.222 patients (81%) were actively followed and 591 (14%) were lost to follow-up at the time of inclusion in the registry. The initial average percentage of “do not follow” patients (relative to overall follow-up time in rheumatology), defined as the time from the onset of the first criterion of SLE until the first assessment for rheumatology service, was 27.4% (SD: 29.2%). 50% of patients have been followed for 14.2% of the time their disease with an IR (3.1%–47.5%). We established the coincidence in date of birth and gender as a mechanism to identify possible duplicates.

Only in 7 of the 359 variables collected≥were more than 20% of the values missing (family history of autoimmune disease, menopause, use of contraceptives, anti β 2 glycoprotein I IgG and IgM, lupus anticoagulant and high creatinine). In 92% of the variables, the percentage of losses has been less than 5%.

Demographic data are shown in Table 3. 90% of patients are women, 93% Caucasian (n=3905), median (IR) age at diagnosis of 33 [25–45] years, median (IR) duration of SLE, since diagnosis, 120 [60.201] months. Table 4 shows the percentage of patients who presented each of the ACR classification criteria. Comorbidities are shown in Table 5.

The median (IR) activity level in patients with SLE (No.=3679), according to SELENA-SLEDAI is 2 [0.4], median (IR) of SLICC/ACR DI: 1 [0.2] and median (IR) of the IGK 2 [1,3]. For patients being currently followed (No.=3222), medium (RI) values of the above indices are very similar: 2 [0.4] 0 [0.1] and 2 [1 3] respectively. The median (IR) of the Charlson comorbidity index was 2 (1–3) on a scale of 0–37. A total of 211 patients included (6%) died, 55 (26%) due to SLE, 53 (25%) infections, 53 (25%) cardiovascular events and 32 (15%) due to cancer.

The authors declare that no experiments have been performed on humans or animals.

The authors declare that they have followed the protocols of their workplace on the publication of data from patients and that all patients included in the study have received sufficient information and have given their written informed consent to participate in the study.

The authors state that no patient data appear in this article.