Objectives This study assessed the psychometric properties of the fatigue numeric rating scale (NRS) and sought to establish values for clinically meaningful change (responder definition).
Methods Using disease-specific clinician-reported and patient-reported data from two randomised clinical trials of patients with psoriatic arthritis (PsA), the fatigue NRS was evaluated for test–retest reliability, construct validity and responsiveness. A responder definition was also explored using anchor-based and distribution-based methods.
Results Test–retest reliability analyses supported the reproducibility of the fatigue NRS in patients with PsA (intraclass correlation coefficient=0.829). Mean (SD) values at baseline and week 2 were 5.7 (2.2) and 5.7 (2.4), respectively. Supporting construct validity of the fatigue NRS, moderate-to-large correlations with other assessments measuring similar concepts as measured by Sackett’s conventions were demonstrated. Fatigue severity was reduced when the underlying disease activity was improved and reductions remained consistent at week 12 and 24. A 3-point improvement was identified as being optimal for demonstrating a level of clinically meaningful improvement in fatigue NRS after 12–24 weeks of treatment.
Conclusions Fatigue NRS is a valid and responsive patient-reported outcome instrument for use in patients with PsA. The established psychometric properties from this study support the use of fatigue NRS in clinical trials and in routine clinical practice. Robust validation of reliability for use in routine clinical practice in treating patients with active PsA in less active disease states and other more diverse ethnic groups is needed.
- psoriatic arthritis
- patient perspective
- outcomes research
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
What is already known about this subject?
Fatigue is an important domain of disease assessment in patients with psoriatic arthritis (PsA).
What does this study add?
Using disease-specific clinician-reported and patient-reported data from two phase III randomised, double-blind, parallel-group, placebo-controlled, multicentre clinical trials of patients with PsA, Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT)-P1 (NCT01695239) and SPIRIT-P2 (NCT02349295), this analysis showed that the fatigue numeric rating scale (NRS) is a well-defined, valid and responsive patient-reported outcomes instrument for evaluating fatigue over time in a clinical trial setting.
How might this impact on clinical practice?
The established psychometric properties from this study support the use of fatigue NRS in clinical trials to evaluate treatment efficacy at group level and potentially in routine clinical practice to assess and manage PsA-related fatigue.
Psoriatic arthritis (PsA) is a chronic musculoskeletal disease affecting ~30% of patients with psoriasis in the USA,1 with an estimated 30–100 cases per 10 000 adults.2 Manifestations of PsA are heterogeneous, and assessing domains associated with disease activity drives treatment choices.2
Fatigue is a relevant and important symptom to patients with PsA. Studies have shown that up to 50% of patients with PsA experience moderate-to-severe fatigue.3–5 Fatigue is multifactorial and related to physical disability, pain, psychological distress and poor sleep quality.6 7 Patients with PsA experience inflammation, chronic pain and reduced physical fitness.8 These symptoms, coupled with decreased self-esteem and depression, manifest as fatigue and sleep disorders, ultimately affecting a patient’s ability to work,9 as well as their social relationships and quality of life.10 11 Assessing fatigue is paramount because it is considered an important domain for both clinical practice and clinical trials in patients with PsA, second only to pain.6
The area of fatigue assessment in PsA is evolving and requires further consideration.12–14 Although up to 78% of patients with PsA consider the fatigue domain a priority, fatigue is rarely reported as a core outcome.5 7 14–16 The PsA core domain set includes peripheral joint assessment, skin assessment, pain, patient global assessment, physical function and health-related quality of life,5 17 18 and, since 2016, musculoskeletal disease activity (arthritis, enthesitis, dactylitis, spondylitis), fatigue and systemic inflammation have been included.15
Several patient-reported outcome (PRO) scales have been used to assess fatigue in patients with PsA.12 Until 2016, 10 different instruments have been used to assess fatigue in PsA randomised, controlled trials, observational studies or registries.16 Few have been validated in PsA,8 18 and no single measure is favoured to evaluate symptoms of fatigue in PsA patients.12
The fatigue numeric rating scale (NRS) is a single-item PRO measure assessing severity of fatigue. The fatigue NRS is validated for use in rheumatoid arthritis,19 and it is currently used in a PsA-specific composite score (psoriatic arthritis impact of disease, PsAID). The objective of this study was to assess the psychometric properties of the fatigue NRS, including (1) test–retest reliability, (2) construct validity and (3) responsiveness and to establish an appropriate clinically meaningful responder definition for the fatigue NRS using disease-specific clinician-reported and patient-reported data from two randomised clinical trials of patients with PsA.
Patients were males or females at least 18 years old who had an established diagnosis of active PsA for a minimum of 6 months according to the classification criteria for psoriatic arthritis, active PsA defined as the presence of ≥3 tender and ≥3 swollen joints, and the presence of active psoriatic skin lesion or a documented history of plaque psoriasis. Patients in Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT)-P1 were naïve to biological disease-modifying antirheumatic drugs,20 and participants in SPIRIT-P2 had been treated with one or two tumour necrosis factor inhibitors and discontinued due to either an inadequate response (≥12 weeks on therapy) or intolerance.21
The dataset used for these analyses came from two phase III randomised, double-blind, parallel-group, placebo-controlled, multicentre clinical trials; SPIRIT-P1 (NCT01695239)20 and SPIRIT-P2 (NCT02349295),21 conducted to assess the efficacy and safety of ixekizumab (IXE), a high-affinity monoclonal antibody that selectively targets IL-17A, for the treatment of active PsA. Details of these studies are reported elsewhere.20 21 Patients were randomised to IXE, placebo or adalimumab in SPIRIT-P120 and IXE or placebo in SPIRIT-P2.21 Both studies involved an initial 24-week treatment period; the primary safety and efficacy endpoints of the trials focused on this 24-week treatment period. The analyses described in this report do not evaluate treatment effects.
The studies20 21 were conducted in accordance with the consensus ethics principles derived from international ethics guidelines, including the Declaration of Helsinki and Council for International Organizations of Medical Sciences International Ethics Guidelines, the International Conference on Harmonisation Good Clinical Practice Guidelines, and applicable laws and regulations. The protocols were reviewed and approved by the institutional ethical review board, and all participants provided informed consent.
Fatigue numeric rating scale
The fatigue NRS is a patient-administered, single-item, 11-point horizontal scale anchored at 0 and 10, with 0 representing ‘no fatigue’ and 10 representing ‘as bad as you can imagine’. Patients are asked to ‘please rate your fatigue (weariness, tiredness) by selecting the number that describes your worst level of fatigue during the past 24 hours’ (online supplementary figure 1). In the SPIRIT-P1 study, the fatigue NRS was administered at baseline and weeks 4, 12, 16 and 24. The SPIRIT-P2 study had an additional time point: week 2. The following instruments were used in the evaluation of the psychometric properties of the fatigue NRS.
Health assessment questionnaire-disability index
The health assessment questionnaire-disability index (HAQ-DI) is a patient-reported standardised questionnaire that is commonly used in PsA to measure disease-associated disability (assessment of physical function).22 It consists of 24 questions referring to eight domains: dressing/grooming, arising, eating, walking, hygiene, reach, grip and other daily activities. The range of scores is from 0 to 3, with higher scores reflecting higher disability.
Patient global assessment
In the patient global assessment (PGA), the patient’s overall assessment of her or his PsA activity was recorded using the 100 mm horizontal visual analogue scale (VAS), where the left (score=0) anchor represents no disease activity and the right anchor represents extremely active disease.23
Disease activity index for psoriatic arthritis
The disease activity index for psoriatic arthritis (DAPSA) is a composite sum score for joint disease activity, including patient global and pain VAS, numeric swollen and tender joint count, and C reactive protein level.13 The range is from 0 to ~160, with higher scores reflecting higher disease activity.24
Medical outcomes study short-form 36
The short-form 36 (SF-36) is a 36-item patient-administered measure designed to be a generic, multipurpose assessment of health in the areas of physical functioning, role physical, role emotional, bodily pain, vitality, social functioning, mental health and general health.25 Physical component summary (PCS) and mental component summary can be calculated using weighted SF-36 domain scores (scoring manual).26 Higher scores reflect better health status.
Psoriatic arthritis disease activity score
The psoriatic arthritis disease activity score (PASDAS) is a composite outcome measure that includes the variables of patient and physician global VAS scores, dactylitis, enthesitis, C reactive protein (CRP), swollen joint counts (SJC), SF-36 PCS and tender joint counts (TJC). The PASDAS is represented by the equation: PASDAS = (((0.18 × √physician global VAS) + (0.159 × √patient global VAS) – (0.253 × √SF36 – PCS) + (0.101×LN (swollen joint count +1)) + (0.048×LN (tender joint count +1)) + (0.23×LN (Leeds Enthesitis Index+1)) + (0.377 LN (tender dactylitis count +1)) + (0.102×LN (CRP mg/dL+1))+2)*1.5.27
Composite psoriatic disease activity index
The composite psoriatic disease activity index (CPDAI) is a measure in which disease involvement is assessed in up to five domains: peripheral joints, skin, entheseal, dactylitis and spinal manifestations.13 28 Measures used are patient self-administered, physical examination and laboratory tests, recorded on paper or electronically. Higher scores correspond to more severe disease activity. In the SPIRIT studies, a modified CPDAI (mCPDAI) was used by the exclusion of the Ankylosing Spondylitis Quality of Life (ASQoL), which was not measured in SPIRIT-P1.
Descriptive statistics (means, percentages) are presented for study participant characteristics at baseline. We assessed test–retest validity, construct validity and responsiveness of the fatigue NRS. Since SPIRIT-P1 and SPIRIT-P2 enrolled different patient populations, all the analyses were conducted separately, and presented side-by-side. Within each study, all treatment groups were pooled. Missing data were low (<3% missing fatigue NRS in both studies at baseline and <10% on or before week 16). Missing postbaseline fatigue NRS scores and other continuous measures were imputed using the last observation carried forward.
Test–retest validity was assessed using SPIRIT-P2 data from baseline and week 2 in patients with stable disease. Stable disease was defined as patients in the placebo group with <20% change in tender joint count. An intraclass correlation coefficient (ICC) was calculated between the initial and retest scores (week 2), and paired t-tests were used for differences in means. An ICC of ≥0.70 was considered acceptable.29
Construct validity assesses the degree to which a measure correlates with other measures that are evaluating a similar construct.30 Construct validity was determined by Pearson correlations at baseline, week 12, and week 24 between scores of the fatigue NRS and HAQ-DI, PGA, DAPSA and SF-36 role emotional, social functioning, role physical, physical functioning and vitality domains. Analyses were conducted on pooled cohort data within each trial and missing data at weeks 12 and 24 were imputed using a last observation carried forward analysis. We hypothesised a high correlation of fatigue NRS with the SF-36 vitality scale which measures a very similar concept, and moderate correlations with other measures of disease activity and impact.
Responsiveness was evaluated by associating calculated changes from baseline to week 12 and 24 in scores on the fatigue NRS with American College of Rheumatology (ACR) 20% (ACR20) and 50% (ACR50) response criteria. Effect size, defined as change from baseline divided by baselineSD, is also provided to help interpret the magnitude. The association was evaluated using the analysis of covariance model, including ACR response status and baseline fatigue NRS score. The anchor was defined as disease activity at endpoint or outcome. Missing data at weeks 12 and 24 were imputed using a last observation carried forward analysis. Data from the trials were not pooled for this analysis.
A responder definition for the interpretation of the fatigue NRS score that corresponds with marked clinical improvement in PsA was identified using both an anchor-based and a distribution-based approach.31 Selected anchors included ACR20, ACR50, HAQ-DI minimum clinically important difference (MCID),32 minimal disease activity (MDA)33 and psoriatic arthritis response criteria (PsARC).13 Receiver operating characteristic method was utilised to identify the cut-off best representing treatment benefit. In addition to the commonly used metrics, like sensitivity and specificity, three more metrics, including positive prediction, negative prediction and phi correlation,34 were also utilised in the method.
A total of 780 patients (SPIRIT-P1 (n=417) and SPIRIT-P2 (n=363)) were included in this analysis. Table 1 describes the demographics and disease characteristics, in addition to baseline fatigue NRS scores. Mean patient age was 49.5 years in SPIRIT-P1 and 51.9 in SPIRIT-P2, ≥91% were white, and mean baseline fatigue NRS scores were 5.5 and 6.0, respectively. In SPIRIT-P1 and SPIRIT-P2, mean baseline tender joint count out of 68 joints (TJC 68) was 20.1 and 23.4, respectively, and swollen joint count out of 66 joints (SJC 66) was 11.0 and 12.3, respectively.
The test–retest reliability supported the reproducibility of the fatigue NRS in patients treated with placebo with stable PsA (n=38; ICC (95% CI 0.829 (0.697 to 0.907)). The mean (SD) values at baseline and week 2 were 5.7 (2.2) and 5.7 (2.4), respectively (p=0.815).
Construct validity was supported by correlations between fatigue NRS and other outcomes at weeks 12 and 24 (online supplementary table 1). Fatigue NRS had the lowest correlations with mCPDAI (r=0.28 (0.10, 0.45) for SPIRIT-1 and r=0.51 (0.43, 0.59) for SPIRIT-P2 at week 12; r=0.48 (0.39, 0.55) for SPIRIT-P1 and r=0.53 (0.45, 0.61) for SPIRIT-P2 at week 24) and the highest correlation with the SF-36 vitality domain (r=−0.66 (−0.71, –0.60) for SPIRIT-P1 and r=−0.75 (−0.79, –0.70) for SPIRIT-P2 at week 12; r=−0.71 (−0.75, –0.66) for SPIRIT-P1 and r=−0.76 (−0.80, –0.72) for SPIRIT-P2 at week 24), with the other outcome measures falling in the moderate range for correlation as measured by Sackett’s conventions.35 Responsiveness of fatigue NRS was associated with week 12 and 24 outcomes, ACR20 and ACR50 (figure 1). Fatigue severity was reduced when the underlying disease was improved and reductions remained consistent at week 12 and 24. Consistent results were observed in both clinical trials.
When using ACR20 as the anchor, the clinically meaningful change in fatigue NRS was identified by maximising the selected metrics shown in table 2A,B separately for each randomised controlled trial.
Considering the trade-off between the metrics, a 2-point to 4-point improvement (effect size 0.8–1.7) best balanced the selected metrics. Similar patterns were observed using other anchors, such as ACR50, MDA, PGA MCID and PsARC (online supplementary table 2). A 3-point improvement was chosen because corresponding effect size (1.2) was sufficiently large, and it was also greater than the SE of measurement of 1.3 suggested by distribution-based method. The distribution-based method aims to quantify a lower bound, and any change within that range can be considered random variation and hence not statistically meaningful.
These analyses examined the psychometric properties of the fatigue NRS using data from two randomised clinical trials of ixekizumab in patients with PsA. These trials assessed different patient populations, one biological-naïve and one biological-experienced, thus psychometric properties were evaluated for the two trials separately. The baseline characteristics of the two SPIRIT trials are representative of a typical PsA trial population and demonstrate PsA patients are burdened by a clinically significant level of patient-reported fatigue (score of 6 on a 10 point scale, with 10=worst). This is consistent with the baseline characteristics reported from other biological or targeted synthetic disease-modifying antirheumatic drug clinical trials. While different measures of fatigue were used across studies, a trial using the fatigue assessment scale (range 0–10; higher score=greater fatigue) reported baseline scores of 5.8–6.3 across treatment groups,36 trials reporting scores from the FACIT-fatigue (0–52; higher score=less fatigue) reported scores of 24.5–30.8 across treatment groups,37–40 and a trial using a fatigue 0–100 VAS (higher score=greater fatigue) reported scores of 54.7–55.9 across treatment groups.41
Results from these two studies support the fatigue NRS as a valid and responsive PRO instrument for evaluating fatigue over time in a clinical trial setting. The test–retest reliability analyses supported the reproducibility of the measure (ICC=0.829). However, this value was lower than the ICC value of 0.95 reported by Chandran et al18 in determining the reliability and validity of the functional assessment of chronic illness therapy-fatigue scale in PsA. This was due, in part, to limitations in how the fatigue instrument was administered in the SPIRIT trials: (1) we assessed reliability between two measurements 2 weeks apart rather than 1 week apart in the FACIT study and (2) the two assessment points were before and after randomised treatment was initiated, rather than at both measurement points prior to the start of study treatment. The latter point is the most significant limitation. We tried to circumvent this by looking at patients who had stable TJC scores; however, it did not change the fact that these patients were undergoing a clinical trial potentially affecting this reliability analysis. In addition, reliability in the current analysis was assessed only in the placebo group, and was only based on tender joint count, resulting in a low sample size of 38.
Fatigue levels correlated with disease activity. Notably, a 3-point change is optimal for demonstrating a level of clinically meaningful improvement in severity after 24 weeks of treatment, which corresponds to marked clinical improvements in PsA disease activity. This is consistent with Gudu et al4 who found that fatigue levels were significantly high in PsA patients with more swollen (p=0.002) and tender (p=0.0005) joints. Tender joint count (OR for five extra joints 1.30 (95% CI 1.01 to 1.68)) was also an independent variable associated with high level of fatigue.
A cross-sectional correlation with other patient-reported and physician-reported outcome measures supported the construct validity with the highest correlation with the SF-36 vitality domain and substantial correlation also seen with joint pain, HAQ-DI, physical functioning SF-36 role physical and PGA at week 24. Responsiveness of the fatigue NRS was evident in correlations with week 12 and 24 outcomes. In particular, significant differences in improvement of fatigue NRS scores between ACR20 and ACR50 responders and non-responders were evident. Our results are consistent with Minnock et al42 who confirmed responsiveness in identifying statistically significant differences in fatigue levels using NRS in PsA patients 12 weeks after initiating treatment. Contrary to what we found, data from a longitudinal study by Husted et al5 suggest that a change in a patient’s clinical status is weakly associated with a change in fatigue. Their data suggest that the change in fatigue is more strongly associated with a 6-month change in chronic pain, depression and physical ability. This is related to the fact that fatigue has many facets associated with pain, physical disability and psychological difficulties.6
The SPIRIT trials included the fatigue NRS as the sole measurement of fatigue because at the time the trials were designed, there was no single measure favoured to assess symptoms of fatigue in PsA patients and the fatigue NRS was considered a clinically relevant (useful and easy to apply in a clinical practice setting) and acceptable endpoint. Since that time, additional publications describing the psychometric properties of the FACIT-fatigue in PsA patients are now available and the evolution of the PsAID, which includes a fatigue NRS item, has now been provisionally endorsed by Outcome Measures in Rheumatology (OMERACT) for use in PsA clinical trials, further expanding options for evaluating fatigue in this patient population.43 44
Data regarding patient perceptions of fatigue are essential in the clinic, as well as in clinical trials. Patients’ perceptions of fatigue as measured by NRS yield important information for clinicians to consider and relate to patients’ perceptions of treatment success. Such understandings have the potential to affect decisions on treatment escalation or discontinuation.45 It is recommended that clinical trials for treatment of PsA collect PRO data to supplement other efficacy and safety data.46 Our study supports previous findings37 of the feasibility of a single-item, one-dimensional scale for measuring fatigue and its potential successful use in clinical trials.
One limitation associated with this study is the lack of ethnic diversity within the cohorts, as the majority of patients were white. These findings, therefore, may not be generalised to other ethnicities. This analysis does not take into account comorbidities or psychological and cognitive aspects of fatigue in PsA. Other PROs strongly correlated with fatigue, such as pain, functional disability, sleep quality and depression, were not included in the analysis. This study does not include another multidimensional fatigue measure in the trial to allow for direct comparison but does include the vitality domain in the SF-36. Finally, although our reported ICC supports the reliability of the fatigue NRS within an acceptable range, the definition of the unchanged group using TJC 68 at week 2 and the 2-week interval between the assessments represent limitations.
In conclusion, fatigue NRS is a well-defined PRO instrument that was valid and responsive for measuring fatigue in patients with PsA in a clinical trial setting. The established psychometric properties from this study support the use of fatigue NRS in clinical trials to evaluate treatment efficacy at group level and potentially in routine clinical practice to assess and manage PsA-related fatigue; however, robust validation of reliability for use in routine clinical practice in treating individual patients with active PsA is needed.
The authors would like to thank Shannon E Gardell of Syneos Health for providing writing support.
Contributors All authors have made substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data, have drafted the work or revised it critically for important intellectual content and have provided final approval of the manuscript.
Funding This work was supported by Eli Lilly and Company.
Competing interests DG received research grant support and/or consulting fees from Abbvie, Amgen, BMS, Celgene, Eli Lilly and Company, Janssen, Novartis, Pfizer and UCB. PN has received grants for research and for clinical trials and honoraria for advice and lectures from AbbVie, Amgen, BMS, Celgene, Eli Lilly and Company, Janssen, MSD, Novartis, Pfizer, Roche, Sanofi and UCB. HG has received fees for speaking and/or consulting from Abbie, Asahi Kasei, Astellas, Chugai, Eisai, Eli Lilly and Company, Janssen, Mitsubishi Tanabe, Ono, Pfizer, Santen, Takeda, and received research funding to Osaka City General Hospital from AbbVie, Chugai, Eisai, Eli Lilly and Company, Janssen, Sanofi and Santen. A-MO received research grant support and/or consulting fees from Celgene, Eli Lilly and Company, Horizon, Janssen, Novartis, Pfizer and UCB. TKK has received fees for speaking and/or consulting from AbbVie, Biogen, BMS, Boehringer Ingelheim, Celgene, Celltrion, Eli Lilly and Company, Epirus, Hospira, Merck-Serono, MSD, Mundipharma, Novartis, Oktal, Orion Pharma, Hospira/Pfizer, Roche, Sandoz, and UCB and received research funding to Diakonhjemmet Hospital from AbbVie, BMS, MSD, Pfizer, Roche and UCB. JB and C-YL are full-time employees of and hold stock/stock options in Eli Lilly and Company.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.
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.