Patterns of Outpatient Phecodes Predating the Diagnosis of Systemic Lupus Erythematosus in Taiwanese Women

J Clin Med. 2022 Sep 14;11(18):5406. doi: 10.3390/jcm11185406.

Abstract

Shortening the time to diagnosis and initiating early treatment are imperative to improve outcomes in patients with systemic lupus erythematosus (SLE). The aim of this case-control study, based on the data from the Taiwan's National Health Insurance Research Database (NHIRD), was to investigate the patterns of diagnoses of disease phenotypes in female patients with SLE up to eight years prior to its definitive diagnosis. The 547 cases were selected from the 2000-2012 NHIRD catastrophic illness datafile and frequency-matched with 2188 controls. The primary diagnosis based on the first ICD-9-CM code for each outpatient visit was converted to Phecodes. Separate regression models, based on least absolute shrinkage and selection operator (lasso) regularization, with seven different lag periods from 1-2 to 7-8 years, were conducted. Results showed that SLE was associated with 46 disease phenotypes in a lag period of 2-3 years, but fewer in other lag periods. A number of SLE-associated disease phenotypes, such as primary thrombocytopenia, thyroid diseases, Raynaud's syndrome, renal disease, and several infectious diseases, occurred mainly in the first few years prior to SLE diagnosis. In conclusion, SLE should be suspected when the disease phenotypes identified in the present study occurred concomitantly.

Keywords: National Health Insurance Research Database; Phecodes; disease phenotypes; lasso regression; systemic lupus erythematosus.