Systemic lupus erythematosus (SLE) is a complex autoimmune disease with a wide spectrum of clinical manifestations and variable severity. The clinical presentation ranges from mild to severe life- and organ-threatening illness, and the clinical course often follows a relapsing-remitting pattern. Sometimes SLE may first present as a serious or critical illness that requires hospitalization.
Previous studies have reported significant variations in hospitalization rates and reasons for admission. Around 10-20% of patients with SLE are hospitalized annually, with the most common causes of admission being active disease, infections and cardiovascular events. It should be noted that among patients hospitalized with SLE, 20-30% are new cases diagnosed during hospitalization.
Diagnosing SLE can be challenging, especially in the early stages, when the disease is often insidious, with only a few features present.
In the hospital setting, these features may resemble conditions that mimic lupus, such as other autoimmune, infectious, or hematologic diseases, which may lead to delays in diagnosis. This delay in diagnosis and prompt initiation of treatment has been associated with an increase in flares and organ dysfunction.
To this end, the application of the existing classification criteria can help in the diagnosis of SLE, with the combination of the 3 sets (American College of Rheumatology [ACR] 1997, Systemic Lupus International Collaborating Clinics [SLICC]-2012, European League Against Rheumatism/American College of Rheumatology [EULAR/ACR] 2019) that show the highest sensitivity.
However, the criteria still fail to classify up to 20% of cases at diagnosis, and therefore some patients with potentially severe disease may be “missed.” To facilitate early diagnosis, we have recently developed an easy-to-use algorithm for SLE diagnosis, using machine learning.
The SLE Risk Probability Index (SLERPI) consists of 14 variably weighted SLE clinical and serologic characteristics that can produce individualized risk probabilities for clinical SLE versus competing rheumatologic diseases (i.e. "definite", "probable/probable", "possible", "unlikely"). ), similar to clinical diagnostic reasoning. 30 A threshold >7 can be used as a dichotomous algorithm (i.e., SLE or not) with high accuracy (94.2%) for the diagnosis of SLE, including early and severe disease. 30
The objective of the present study was to 1) determine the clinical phenotype of patients with new-onset SLE requiring hospitalization, 2) estimate the delay between hospitalization and SLE diagnosis, and 3) evaluate the diagnostic performance of the SLERPI in hospitalized patients. with suspected SLE.
Background
Prompt recognition of systemic lupus erythematosus (SLE) in hospitalized patients with severe illness is essential to initiate treatment. We sought to characterize the phenotype of hospitalized patients with new-onset SLE and estimate potential delays in diagnosis.
Methods
An observational study of 855 patients (“Attikon” SLE cohort). The clinical phenotype was categorized according to the main manifestation leading to hospitalization.
Disease characteristics, time to diagnosis, classification criteria, and SLE risk probability index (SLERPI) were recorded for each patient.
Results
There were 191 patients (22.3% of the total cohort) hospitalized for manifestations possibly attributable to SLE. The main causes of admission were neuropsychiatric syndromes (21.4%), cytopenias (17.8%), nephritis (17.2%) and thrombotic events (16.2%).
Although 79.5% of patients were diagnosed within 3 months of hospitalization, in 39 patients the diagnosis was late, especially in those with hematological manifestations.
At the time of hospitalization, a SLERPI > 7 (indicating a high probability of SLE) was found in 87.4% of patients. Patients lost to the SLERPI had fever, thrombotic or neuropsychiatric manifestations not included in the algorithm.
Reducing the SLERPI threshold to 5 in patients with fever or thrombotic events increased the diagnostic rate from 88.8% to 97.9% in this subgroup, whereas inclusion of all neuropsychiatric events did not provide any additional diagnostic value.
Conclusion One in five patients presenting with new-onset overt SLE disease required hospitalization. Although in most cases an early diagnosis was achieved, in approximately 20% the diagnosis was late. A lower SLERPI cut-off point (≥5) in patients with fever or thrombosis could improve early diagnosis. |
