Evaluating the Psychometric Properties and Determining Clinical Meaningfulness of the Psoriasis Symptom Inventory

Abstract

Plaque psoriasis is a skin condition that affects two to three percent of individuals around the world by causing not only signs (e.g., redness) and symptoms (e.g., flaking), but also, impairments in health-related quality of life (e.g., work function). Evaluation of psoriasis in research studies is important to provide the occurrence and magnitude of treatment benefit, especially from the patient’s perspective. Therefore, the Psoriasis Symptom Inventory (PSI) was developed. The PSI is a patient-reported outcome (PRO) measures that assesses the signs and symptoms of psoriasis. While initial research has been conducted on the assessing the psychometric properties of the measure, more work is still needed to assist in validating the measure. Therefore, this research was conducted to further the quantitative evaluation of the PSI by conducting a secondary analysis that utilized data from two separate studies. The first was a longitudinal, 16-week study that used the 24-hour recall, electronic version of the PSI, while the second study was a prospective, cross-sectional study that used the 7-day recall, paper version of the PSI. Given the differences in the design of the two studies, focused research aims were created to best utilize the given data from each source to evaluate the measurement properties and assess the clinical meaningfulness of the PSI. The aims were targeted more towards research that had not been done on the PSI as of yet. To provide a comprehensive report, eight chapters were developed to detail the research that was conducted. Chapter 1 provided initial background information on psoriasis and the use of clinical measures, including the PSI, to evaluate the skin condition. In addition, the chapter addressed some of the research that has been done on the measurement properties of the PSI, as well as, the unmet research objectives and hypothesis that were conducted for this study. In Chapter 2, a literature review was conducted to primarily provide an in-depth review of the current use of PRO measures in psoriasis, in addition to, previous research that has been conducted when using the PSI. Chapter 3 focused on the data sources for conducting this research including information on data collection and the variables that were available in each dataset. The remaining chapters focused on the results that met the objectives of this research. In Chapter 4, a summary of the psychometric properties (e.g., classical test theory [CTT], item response theory [IRT]) for the PSI (24-hour) were evaluated at Baseline, Week 8, Week 12, and Week 16. The primary findings of this measurement properties analyses indicated good item and measure performance at Baseline. However, due to low sample size, analysis for the remaining timepoints had to be interpreted with caution as treatment benefit affected the choosing of response options by participants. Chapter 5 evaluated the categorization of the PSI responder against other clinical measure responders. Key findings from this analysis showed that the PSI responder had moderate association (kappa= 0.40-0.60) with some of the previously developed clinical measure responders, and therefore, the PSI can provide clinically meaningful interpretation of psoriasis improvement. In Chapter 6, responsiveness was evaluated, as well as, the clinically important difference (CID) and responder (CIR) were determined for the PSI. The PSI total score averaged (mean + SD) at 6.76+5.93 at Week 16, which resulted in a change score of 6.88+7.72 (improvement) from Baseline. The CID for the PSI was 3.39 (95% CI: 3.08, 3.17) while the CIR was 27.16% (95% CI: 20.74, 33.59). In Chapter 7, the psychometric properties of the PSI (7-day) were assessed. With the use of CTT, CFA, Rasch, and IRT, the PSI showed good item and measure performance. Key findings from the analyses showed that the PSI had high internal consistency (CTT; Cronbach’s alpha=0.95), good overall fit (CFA; comparative fit index/Tucker-Lewis index=0.99/0.98), and no misfitting items (Rasch) and overlapping of response options (IRT). Chapter 8 observed the differential item functioning (DIF) of the PSI (7-day) between males and females. From the analysis, one item (Item 2: Redness) was flagged for non-uniform DIF (X223=0.007). All items, including the redness item, showed that the magnitude for DIF was negligible (R2<0.13). When evaluating the test characteristic curve, sex was not associated with difference in response for the PSI. Overall, this research assessed various key quantitative analyses for evaluating the psychometric properties, as well as, clinical meaningfulness of the measure. The findings showed the PSI had good item and measure performance. In addition, the responder status of the PSI associated well with other clinical measure thresholds. Furthermore, the PSI did not show differences in item functioning between sexes. And finally, clinically important difference and responder were established for the PSI to be assessed between treatment groups in future studies. Further research is warranted to assess the PSI in a larger sample and in a real-world setting for both versions for quantitative analyses, such as determining measurement invariance across time and cross-cultural performance of the PSI.