Objectives The Psoriatic Arthritis Impact of Disease (PsAID) Questionnaire is a recently developed patient-reported outcome measure (PROM) of disease impact in psoriatic arthritis (PsA). We set out to assess the validity in an independent cohort of patients, estimate the minimally important difference for improvement and explore the potential of individual components of the PsAID in clinical practice.
Methods Data were collected prospectively for a single-centre cohort of patients with PsA. Construct validity was assessed by Spearman correlation with other PROMs and reliability by intraclass correlation coefficient (ICC) at 1 week. Sensitivity to change at 3 months was determined by the standardised response mean (SRM) in those patients with active disease requiring a change in treatment.
Results A total of 129 patients (mean ±SD age 52.1±13.3, 57% women, disease duration 10.2±8 years) completed the baseline questionnaires and assessments. The mean baseline PsAID12 score was 3.92±2.26 with an ICC of 0.91 (95%CI 0.87 to 0.94). The SE of measurement was 0.51 and the minimal detectable change was 1.41. There was strong correlation (r≥0.70) with most of the PROMs studied and moderate correlation with clinical outcomes (r=0.40–0.57). The SRM of the PsAID12 was 0.74 (95%CI 0.45 to 0.97). There was strong correlation with individual PsAID items and their corresponding PROM questionnaires (r≥0.67).
Conclusion The PsAID is a reliable, feasible and discriminative measure in patients with PsA. The good responsiveness of the PsAID and strong correlation of individual items with other PROMS represent an opportunity to reduce questionnaire burden for patients in studies and clinical practice.
- psoriatic arthritis
- outcomes research
- disease activity
- patient perspective
- dmards (biologic)
Statistics from Altmetric.com
Psoriatic arthritis (PsA) is a chronic inflammatory arthropathy affecting up to 30% of patients with skin or nail psoriasis.1 Psoriasis is a complex disease and has a multitude of clinical manifestations including skin and nail disease, dactylitis, enthesitis, peripheral arthritis and axial disease. This complexity has resulted in the development of a number of disease activity indices to best capture the clinical manifestations of disease, all of which perform similarly and use a ‘biomedical model’ for determining disease activity.2 Disease activity and damage contribute to reduced physical and psychosocial health-related quality of life, with significant social and economic impact.3
Patient-reported outcome measures (PROMs) have been found to be a reliable indicator of baseline status, change during treatment and are predictive of long-term outcome.4–6 The updated core domain set for PsA includes patient global assessment (PGA), physical function (such as the Health Assessment Questionnaire (HAQ)), fatigue and health-related quality of life in the inner core.7 A recent literature review indicated that pain, PGA and HAQ were frequently measured in clinical trials, however other measures of how the patient feels or functions, such as fatigue and sleep, were rarely reported.8 The Short-Form 36, a recognised measure of the economic impact of disease, is long and the interpretation of the score is complex. The HAQ has also been shown to change with disease duration and less reliably reflects disease activity in well-established disease.9
A European League Against Rheumatism (EULAR) taskforce developed a questionnaire to calculate a score reflecting the impact of PsA from the patients’ perspective, termed the Psoriatic Arthritis Impact of Disease (PsAID) Questionnaire.10 Development of this questionnaire was driven by the patient perspective and involved 12 patient research partners from 12 countries all with personal experience of PsA. The taskforce determined 12 domains important to health and well-being, including fatigue, functional capacity, sleep and coping. The PsAID Questionnaire was validated with >470 patients, and had good face validity and good generalisability and also correlated strongly with PGA. The questionnaire has been divided into two scores; the PSAID12 uses all 12 items and was designed to be used in clinical practice, and the PsAID9 uses the first 9 items and was designed to be used in clinical trials. The authors commented that further validation in independent cohorts would be necessary, in particular sensitivity to change.
The Outcome Measures in Rheumatology (OMERACT) filter11 assesses the applicability of a measure and looks at three aspects: truth, discrimination and feasibility. A measure is useful if it measures what is intended, is reliable and sensitive, and can be applied easily. The PsAID has been designed to conform to these principals, but requires further validation, in particular regarding sensitivity to change in comparison with other outcome measures, both clinical and patient reported. The PsAID includes items related to work, fatigue and physical function, all components of the inner core in the PsA core domain set. A single questionnaire could therefore be used in routine clinical care to assess these domains, reducing patient burden. The aim of the study was to further investigate the PsAID in a new cohort to determine reliability, construct validity, sensitivity to change, to validate the minimal clinically important improvement and investigate the correlation of individual questions of the PsAID with established PROMS.
Data were collected prospectively from patients attending the Royal National Hospital for Rheumatic Diseases, Bath, UK, from July 2015 to January 2017. The cohort was divided into two groups based on disease activity, with the data from group 2 used to determine sensitivity to change. Patients with stable disease not requiring a change in treatment (patient and physician opinion) were recruited to group 1. Patients with active disease requiring a change in treatment were recruited to group 2. Clinical measures and PROMs including the PsAID were completed at baseline by both groups and the PsAID Questionnaire was repeated at 1 week and returned by post. Patients in group 2 attended for follow-up 3 months after a change in treatment. For each patient, written informed consent was obtained.
To be eligible for the study, patients had to fulfil ClASsification of Psoriatic ARthritis (CASPAR) diagnostic criteria for PsA.12 For patients in group 2, an intensification of therapy was defined as: (i) an increase in the dose of the current conventional synthetic disease-modifying antirheumatic drug(s) (csDMARDs); (ii) initiation of new csDMARD(s); (iii) initiation of a new, or change of the current, biological DMARD (bDMARD). Patients were excluded from the study if they were under the age of 18 years, were unable to speak and read English, had an additional inflammatory disease other than psoriatic disease, or severe, unstable comorbidities that could influence the PROMs.
Demographic data were collected at baseline including age, gender, disease duration, smoking status, body mass index (BMI), and the presence or absence of erosive disease on X-ray. At baseline (both groups) and at 3 months (group 2), patients completed the Health Assessment Questionnaire–Disability Index (HAQ-DI,13), EuroQol-5D (EQ-5D-5L,14 15) Questionnaire, Psoriatic Arthritis Quality of Life (PsAQoL16) Questionnaire, Dermatology Life Quality Index (DLQI,17), Work Productivity and Activity Impairment: Specific Health Problem (WPAI:SHP,18) Questionnaire and visual analogue scale (VAS) scores for PGA and joint global assessment (JtGA). All patients underwent a full clinical assessment at baseline, and a 66/68 swollen joint count and tender joint count, dactylitis score, Leeds Enthesitis Index19, Bath nail score,20 Psoriasis Activity and Severity Index (PASI,21), physician global assessment–VAS and C-reactive protein (where possible) were recorded. The modified Composite Disease Activity Index (mCPDAI,22) was used as a composite measure of disease activity.
Patients in group 2 had a further clinical assessment at follow-up and, in addition, were asked ‘which statement best describes your arthritis today compared with 3 months ago?’: (a) overall my condition is improved; (b) overall, my condition is stable; (c) overall, my condition is worse. If patients indicated that their condition had improved, they were also asked to rate the improvement as somewhat significant, moderately significant or very significant. The PsAID questionnaire was completed at baseline, 1 week and 3 months (follow-up, group 2 only).
Internal consistency of the PsAID scoring tool was evaluated using Cronbach’s alpha coefficient, with a value >0.7 indicating consistency. Relative reliability of the PsAID was tested with the intraclass correlation coefficient (ICC, two-way mixed model absolute agreement), and variability was assessed using the Bland-Altman method.23 Absolute reliability was determined by the SE of measurement (SEM), indicating the variability in scores on repeated testing. The SEM was determined using .24 The minimal detectable change (MDC), representing the magnitude of change required to exceed the measurement error, was calculated for the 95% CI as .25
Construct validity was assessed using Spearman rank correlation coefficient as the PROM scores were not normally distributed (Shapiro-Wilk test). Sensitivity to change was calculated using the standardised response mean (SRM, the mean change between baseline and 3 months divided by the SD of the change) with the analysis done on a per protocol basis. An SRM >0.8 is considered large. CIs were calculated by bootstrap.
The minimal clinically important improvement (MCII) was estimated using receiver–operating characteristic curves, plotted using improvement versus no improvement as the outcome, and the minimal clinically relevant improvement was computed as the change score that had maximal sensitivity while maintaining a specificity of 0.80. This measure indicates the degree of change that ≥80% of patients would indicate as important.
The correlation between individual items of the PsAID questionnaire and specific patient-reported outcome questionnaires was evaluated. The domains studied were fatigue (Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue,26 compared with item 2 of the PsAID questionnaire), skin disease (DLQI, compared with item 3), work (WPAI:SHP, compared with item 4 and item 11), function (HAQ, compared with item 5) and quality of life (PsAQoL, compared with the PsAID12). To enable direct comparison with the PsAID items, the studied PROM questionnaires were transformed to a 0–10 scale. In the case of the FACIT-Fatigue, where higher scores indicate less fatigue, the scale was reversed so that a score of 0 indicated no fatigue and 10 indicated maximum fatigue. Correlation was determined using Spearman rank correlation coefficient. Bland-Altman plots with 95% upper and lower limits of agreement were constructed to examine the differences, using both absolute differences (score A – score B) and relative differences (score A – score B/mean%).27 Statistical analysis was performed using SPSS V.22.0.
A total of 159 patients were screened and recruited to the study; 129 patients satisfactorily completed the required baseline questionnaires and assessments. Demographic data for the cohort are presented in table 1. The mean ±SD baseline PsAID12 score was 3.92±2.36 and the mean baseline PsAID9 score was 4.11±2.40. The mean re-test PsAID12 score was 3.55±2.26 and the mean re-test PsAID9 score was 3.72±2.30. Cronbach’s alpha was 0.95 and the ICC was 0.91 (95% CI 0.87 to 0.94) and 0.91 (0.86 to 0.94) for the PsAID12 and PsAID9, respectively, indicating excellent reliability. Corresponding values for group 1 only were Cronbach’s alpha 0.96, ICC 0.92 (0.87 to 0.96) and 0.92 (0.86 to 0.95) for PsAID12 and PsAID9, respectively. There were very few missing items (<0.5%) indicating excellent feasibility. The variability in test–retest scores is shown in figure 1. There was no proportional bias. The SEM was 0.51 and the MDC was 1.41.
There was strong correlation between the PsAID and most PROMs as shown in table 2, with Spearman’s rank correlation coefficient >0.60. The strongest correlation was with the EQ-5D Index score (−0.87), followed by the Bath Ankylosing Spondylitis Disease Activity Index and FACIT-Fatigue. There was moderate correlation between the PsAID and joint counts (Spearman’s rho of 0.57 and 0.40 for tender and swollen joint counts, respectively) and mCPDAI score (0.57). There was no correlation between the PsAID score and the PASI, possibly due to the minimal skin disease in the patient population (mean PASI 0.6 indicating very mild skin disease).
The PsAID score did not change significantly when stratified by disease duration, age, BMI, or the presence or absence of erosive disease (data not shown). Notably, the PsAID score was higher in women with a mean score of 4.44±2.13 vs 3.23±2.50 in women and men respectively, with women scoring significantly higher in 8 of the 12 items. This compared similarly to the other recorded PROMs in this cohort, with higher scores seen in women for all PROMs except DLQI. The correlation with PROMs and clinical outcomes was also stronger in men than in women (table 2).
To determine the minimal clinically important improvement, an ROC curve was constructed using improved (overall, my condition has improved) versus not improved (overall, my condition is stable or worse) as the outcome (see the online supplementary figure 2). The area under the curve (AUC) was 0.821. The change in PsAID score with greatest sensitivity while maintaining a specificity of 80% was 1.25 (sensitivity 61%, specificity 80%). Furthermore, an ROC curve was constructed with improvement defined as a relative improvement in PsAID score at 3 months of 30%. Again, an absolute change score of 1.25 had the greatest sensitivity and specificity (74% and 77%, respectively). In this cohort, using the previously proposed cut-off of 3 to define a MCII,10 specificity was 100% but sensitivity was only 29%.
Supplementary file 1
Sensitivity to change was assessed in those patients who remained on increased therapy at the time of their follow-up assessment. The SRM of the PsAID12 was 0.74 (95% CI 0.45 to 0.97). In comparison, the SRM of the mCPDAI was 0.71 (0.35–1.01).
The individual items of the PsAID questionnaire typically scored higher than the PROM questionnaires (table 3). There was a strong correlation between PsAID item and corresponding PROM questionnaire, with Spearman correlation coefficients >0.70 (see online supplementary table 4), with the strongest correlation seen in the domains of fatigue and skin disease. Variability was assessed using Bland-Altman plots (see the online supplementary figure 3–13). There was significant variability in the scores, with upper limits of agreement ranging from 2.9 to 5.22 and lower limits of agreement ranging from −4.0 to −1.97. Bland-Altman plots of the percentage difference in scores showed that at mean scores >5, the PsAID item score more closely approximated the PROM questionnaire score.
We have demonstrated that the PsAID questionnaire is a reliable measure of the impact of disease in a UK cohort of patients with PsA, with the 9-item and 12-item questionnaires performing similarly. The questionnaire was also feasible with a very low number of missing items. With a mean difference of 0.10 (95% CI −0.10 to 0.31), there was only minor variability in baseline and 1-week scores despite including patients with both stable and active diseases. Unlike similar instruments used in RA (the rheumatoid arthritis impact of disease (RAID)13), there was no obvious floor effect.
The significant difference between mean scores for men and women has not been previously demonstrated for the PsAID questionnaire and requires further investigation. In this cohort, baseline demographics were similar between men and women, although men had significantly longer disease duration (mean 11.8 vs 9.1 years). Women had higher tender (11.1 vs 6.4) and swollen (2.6 vs 1.8) joint counts as well as higher mCPDAI scores (3.4 vs 2.6), but after adjusting for disease activity, the difference in PSAID score remained significant. A similar difference was noted with the RAID score, as well as other PROMs in PsA including HAQ,28 and may need to be taken into account when interpreting scores. Di Carlo et al 29 assessed construct validity and reliability in an Italian cohort and noted similar results to this study. In addition, PsAID scores were higher in patients with concomitant fibromyalgia. Although not reviewed in our cohort, coexistent fibromyalgia may have contributed to the gender difference found in our study.
The PsAID questionnaire correlated strongly with other PROMs, with the strongest correlation with the EQ-5D index score. This is perhaps not surprising given the EQ-5D index questionnaire includes the domains of mobility, function, usual activities, pain and mental health, which are similar to the domains assessed in the PsAID questionnaire. The correlation with the FACIT-Fatigue is interesting, and ongoing fatigue despite adequate control of psoriatic disease may be one of the main drivers of disease impact. The moderate correlation with clinical outcomes is also expected, as domains such as work, sleep and fatigue are major determinants of disease impact yet are not captured by clinical scores.
The minimal clinically important improvement of 1.25 in the PsAID score in this cohort is significantly less than the MCII of 3.0 in the preliminary validation of the PSAID questionnaire. In this study, the majority of patients had a change of DMARD (increased dose or additional agent), and only 29% had a change of biological DMARD (change or new biologic agent). This may have resulted in a lower mean change in PsAID score in this study compared with the study by Gossec et al 10 in which 50% of patients were commenced on a biological DMARD. In addition, patients were included if they had a change in DMARD dose, which may have also affected the MCII determined in this study.
The PsAID was found to be sensitive to change, with a SRM of 0.74 which was greater than the SRM for the mCPDAI score (0.71), although this was lower than the SRM as determined in the preliminary validation of the PsAID (Gossec et al).10 Although not a measure of disease activity, the PsAID questionnaire could be a useful outcome measure to guide treatment.
The individual items of the PsAID questionnaire correlated strongly with the corresponding PROM questionnaire, particularly between item two (fatigue) and the FACIT-Fatigue and item 3 (skin) and the DLQI. When examining the differences using Bland-Altman plots, we found that an individual PsAID item was typically scored higher than the PROM. The upper and lower limits of agreement were clinically significant, with a mean upper limit of agreement of 4.4 and a mean lower limit of agreement of −2.9. Given the wide 95% levels of agreement seen in this study, the individual PsAID items may not be able to accurately estimate a PROM score. As the PsAID items are disease specific, and questionnaires such as the HAQ and FACIT-Fatigue are not, comorbid conditions in some patients may have contributed to some of this variability. However, the strong linear relationship between individual items and corresponding PROMs does indicate that the PsAID questionnaire is able to highlight domains that are of most concern for a patient, particularly as the percentage difference between the scores was significantly less at higher mean scores. The PsAID questionnaire could therefore be used to tailor treatment at an individual level, targeting the domains that are of most concern for the patient. This would eliminate the need for multiple PROM questionnaires at every visit, reducing questionnaire burden and fatigue especially in routine outpatient care.
The limitations of this study are the low number of patients in the sensitivity to change arm who had a significant response to a change in treatment, and a formal calculation of sample size was not performed. Additional data from interventional clinical trials, particularly of biologic DMARDs, will be useful in confirming the ability of the PsAID to detect change and further validating the MCII.
In summary, the PsAID is a reliable, feasible and valid measure of the impact of disease in patients with PsA. It can be used in patients with both stable and active diseases, correlates well with clinical outcome measures and PROMs and can be used to tailor treatment to the individual.
The authors wish to acknowledge the support of Arthritis Australia and the Australian Rheumatology Association. They also wish to acknowledge the staff and patients of the Royal National Hospital for Rheumatic Disease involved in the study.
Handling editor Tore K Kvien
Contributors RJH, WT, NJM and MB: conception and design of the study. RH, WT, EK, CC, NW and NJM: acquisition and/or analysis of data. RH, WT, EK, CC, NW, MB and NJM: drafting the work or revising it critically for important intellectual content and final approval of the version to be published.
Funding RJH was supported by an award from the Australian Rheumatology Association and Arthritis Australia, during the conduct of the study; grants from Celgene, outside the submitted work. NM reports personal fees from Abbvie, grants and personal fees from Celgene, personal fees from Eli Lilly, outside the submitted work. WT reports grants and personal fees from Abbvie, grants and personal fees from Celgene, personal fees from UCB, outside the submitted work. CC reports grants from Abbvie, grants from Celgene, outside the submitted work.
Competing interests None declared.
Ethics approval The study was conducted according to the principles of the Declaration of Helsinki and ethical approval was obtained from the NRES Committee Yorkshire & The Humber - Bradford Leeds.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.