Background/Purpose: Identifying patients with Systemic Lupus Erythematosus related cognitive impairment (SLE-CI) is critical as SLE-CI can negatively affect employment, quality of life, and disease self-management. Evidence-based and resource-efficient SLE-CI screening instruments are needed. The patient-reported 20-item Perceived Deficits Questionnaire (PDQ-20) is a potential instrument. However, our previous analyses questioned the PDQ-20 purported four-factor structure: attention/concentration, retrospective memory, prospective memory, and planning/organization. The purpose of this study is to explore the PDQ-20 factor (subscale) structure in an SLE cohort.

Methods: This study used PDQ-20 data from consecutive patients aged 18-65 years who met the ACR classification criteria for SLE in a single Canadian rheumatology center. Exploratory factor analyses (EFA) included squared multiple correlations and the maximum likelihood method to extract factors, followed by a varimax (orthoganal) rotation. Subsequent confirmatory factor analyses (CFA) was completed on identified factor structures. The validity of the results was tested with the assumption that each item had one factor pathway, with factor correlations considered. The sample size met standards for EFA (>5 x items) and was sufficient to perform CFA (power=0.99).

Results: Participants who did (n=177) and did not return (n=31) the PDQ-20 did not statistically significantly differ in age, age at SLE diagnosis, education or employment status. EFA without any restriction resulted in 5 factors. However, one factor included only one item (Q19. Forget to take your medication). The EFA scree-plot and eigenvalues (factor 1=119.38) provided strong support for a single factor model. Four factors had eigenvalues over 1, with each factor having 4-6 items, which pulled for examination of a new four-factor model. Overall hypothesis test rejected the assumption of no common factor (Tucker and Lewis coefficient = 0.94). CFA results indicated adequate model fit for the new four-factor model (standardized root mean square residual=0.04; root mean square error of approximation=0.08; Bentler comparative fit index=0.93; all item factor loadings statistically significant with standardized factor loading range 0.56-0.88). Lagrange Multiplier statistics in the CFA found alternate item-factor pathways that significantly improved the model by adding or removing pathways. Correlations between the new four factors were all statistically significant with high interrelatedness (range:0.80-0.92).

Conclusion: Results indicate that the PDQ-20 has one factor. Some results support a four-factor model distinct from the original PDQ-20 subscales. However, this new model exhibits high correlations between factors and multiple item pathways, and it does not correspond with a theoretical cognitive conceptual structure. This may be due to the recruitment of multiple cognitive domains in items related to everyday living activities. Evidence from this analysis and our previous CFA of the original PDQ-20 factor structure indicates that the PDQ-20 has one factor and should be interpreted using the total score and not subscales.

« Back to 2018 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/patient-reported-cognitive-screen-does-not-identify-cognitive-sub-domains-exploration-of-the-subscale-structure-of-the-perceived-deficits-questionnaire-in-a-systemic-lupus-erythematosus-cohort/