Background/Purpose: The Psoriatic Arthritis Impact of Disease (PsAID) questionnaire is a patient reported measure of disease impact. The 12-item questionnaire has many advantages including demonstrated validity, reliability, responsiveness and discrimination.  However, few studies have examined the use of individual items as surrogates for important patient reported outcomes. For example, we were interested in whether the single PsAID fatigue item could be used in place of the 13 item FACIT-F instrument.  Likewise, the pain, function, depression, and anxiety items could be used in place of additional questionnaires if the items correlated well with validated instruments both in cross-sectional and longitudinal settings. The objective of this study was to examine the longitudinal construct validity of the individual PSAID items for pain, fatigue, function, depression and anxiety.

Methods: Patients with PsA were enrolled in the Psoriatic Arthritis Research Consortium (PARC) between 2015-2017.  PARC is a longitudinal observational cohort at four institutions: University of Pennsylvania, Cleveland Clinic, New York University, and University of Utah. Two of these institutions (Utah and Penn) administered the PSAID.  Patient characteristics at the first/baseline visit were descriptively reported. The correlations were calculated among individual PSAID items with similar constructs (e.g., “tired” item with FACIT-F, BASDAI tired item, etc.) at baseline using Spearman’s correlation coefficients.  The change scores (e.g., score at visit 1 – score at visit 0) and the correlation among change scores between the individual items and related constructs were also calculated.

Results: PSAID data were available from 862 visits; 302 patients completed at least one PSAID and 208 patients completed PSAIDs at ≥2 visits.  Most patients were in low disease activity (mean 66/68 swollen and tender joint counts were 2.6 and 5.3, respectively).  At baseline, the mean PsAID9 and PsAID12 scores were 3.39 (SD 2.45) and 3.22 (2.40) respectively. The individual PsAID items were moderately correlated (rho= 0.5-0.8) with similar constructs at baseline.  However, aside from the moderate to strong correlation between the PsAID pain question with the RAPID3 pain question (rho=0.71), change scores were only slightly correlated with like instruments (rho = 0.2-0.45, Table).

Conclusion: The individual PsAID items did not correlate well with change in similar constructs over time. This may be due to the attribution of each symptom to PsA in the PSAID questionnaire.  The assessed PsAID items cannot be used as close substitutes for the validated questionnaires with which they were compared.