Background Approximately 38% of adults living with Systemic Lupus Erythematosus (SLE) experience cognitive impairment (CI) that can detrimentally affect employment, disease self-management, and quality of life. Identifying those with SLE related CI is critical, but is difficult to do in busy and resource-limited clinics. The patient-reported 20-item Perceived Deficits Questionnaire (PDQ-20), used to screen for SLE related CI, could be less time and cost-burdensome than other objective instruments. However, there is a dearth of published measurement property evidence for using the PDQ-20 with SLE patients. In adults with Multiple Sclerosis the PDQ-20 is purported to have four factors (subscales): attention/concentration, retrospective memory, prospective memory, and planning/organization. This structure has not been examined in adults with SLE. The purpose of this study is to examine the factor structure and the internal consistency of the PDQ-20 in an SLE cohort.

Methods Consecutive SLE patients aged 18–65 years were recruited from a single rheumatology center between July 2016 and March 2018. Patients completed the PDQ-20. Analyses included socio-demographic descriptive analyses and confirmatory factor analyses (CFA) of the purported PDQ-20 four-factor structure. Sample size calculations indicated that a cohort of n=177 was sufficient to perform the CFA (power=0.99). Analysis was completed on returned baseline PDQ-20 data using SAS® software.

Results Patient demographics are presented in table 1. There was no missing PDQ-20 data. CFA model fitting was adequate (standardized root mean square residual=0.05; root mean square error of approximation=0.10; Bentler comparative fit index=0.90). All factor loadings were statistically significant (factor loading range 0.55–0.88; all t-value >9.82). All factors highly correlated with each other (correlation range: 0.87–0.97; all p<0.01). Lagrange Multiplier (LM) tests indicated that multiple alternate item-factor pathways could improve the four-factor model (ten largest significant LM statistics range from 7.92–20.78; new possible pathways for 7 items to other factors). Item 19 (‘forget to take medication’) had low reliability to its purported factor (‘prospective memory’; R²=0.30). The internal consistency (Cronbach’s alpha) for the four factors ranged from 0.82 to 0.91.

Conclusions The CFA analyses indicate that while the fit of the four-factor model for the PDQ fits, the model could be improved. Particularly concerning is the different factor-pathways for seven items, item 19’s current low item-factor reliability, and the increased correlations between factors. In adult SLE patients, researchers and clinicians should be cautious in interpreting PDQ-20 results using the current four factors (subscales). Further validity analyses, including exploratory factor analyses, are needed.