Abstract
Objective. To survey participants with polymyalgia rheumatica (PMR) to evaluate the face validity, acceptability, and domain match of proposed candidate outcome measures.
Methods. A structured, online, anonymous survey was disseminated by patient support groups through their networks and online forums. The candidate outcome measures comprised (1) visual analog scale (VAS) and numerical rating score (NRS) to assess pain; (2) VAS, NRS, and duration to assess stiffness; (3) the modified Health Assessment Questionnaire and Health Assessment Questionnaire Disability Index to assess physical function; and (4) C-reactive protein and erythrocyte sedimentation rate to assess inflammation. Free-text answers were analyzed using descriptive thematic analysis to determine respondents’ views of the candidate instruments.
Results. Seventy-eight people with PMR from 6 countries (UK, France, USA, Canada, Australia, and New Zealand) participated in the survey. Most respondents agreed candidate instruments were acceptable or “good to go.” Free-text analysis identified 5 themes that participants considered inadequately covered by the proposed instruments. These related to (1) the variability, context, and location of pain; (2) the variability of stiffness; (3) fatigue; (4) disability; and (5) the correlation of inflammatory marker levels and severity of symptoms, sometimes reflecting disease activity and other times not.
Conclusion. Participants reported additional aspects of their experience that are not covered by the proposed instruments, particularly for the experience of stiffness and effect of fatigue. New patient-reported outcome measures are required to increase the relevance of results from clinical trials to patients with PMR.
Polymyalgia rheumatica (PMR) is an inflammatory condition common in older people, characterized by pain and stiffness in the muscle groups of the shoulder and pelvic girdle. Prevalence figures vary widely and point estimates are affected by study design and choice of classification criteria, resulting in estimates of 0.1% in Portugal, 0.7% in the United States, and 1.5% in the United Kingdom1,2,3,4. Treatment with glucocorticoids (GC) has been standard practice since the 1960s, but prescribed regimens lack standardization, with wide variation in clinical practice5,6. One barrier to finding newer effective treatment is the lack of robust, validated, and reliable outcome measures to identify disease activity and patient outcome7. Different clinical response measures have been used in PMR but there is no consensus on their use clinically, particularly regarding duration of morning stiffness, which is considered too simplistic to identify the functional effect of stiffness from the patient perspective8,9,10. In addition, many of these measures have yet to be fully validated for PMR. The Outcome Measures in Rheumatology (OMERACT) PMR Working Group is assessing the evidence base and usefulness of existing outcome tools to record disease activity in PMR, using the OMERACT filter within the OMERACT handbook11,12,13.
The purpose of our study was to gain the views of patients with PMR on the face validity, acceptability, feasibility, domain match, and overall acceptability of the identified candidate instruments14.
MATERIALS AND METHODS
Identification of candidate instruments
A Core Domain Set and candidate instruments were identified through Delphi rounds of clinicians and patient research partners with an interest in PMR, with methods as described14. The identified core domains and instruments comprise (1) pain assessed by visual analog scale (VAS) and numerical rating score (NRS); (2) stiffness assessed by VAS, NRS, or duration of morning stiffness; (3) physical function assessed by the modified Health Assessment Questionnaire (mHAQ) and the Health Assessment Questionnaire Disability Index (HAQ-DI); and (4) laboratory markers of inflammation assessed by C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR).
Design
This was an online survey with descriptive qualitative analysis of free-text responses.
Sampling frame
Google Docs links were created for each instrument and included in the Newswire newsletter of the charity PMRGCAuk, which is distributed to 1800 readers, and Web links appeared on the HealthUnlocked forum, which has 6986 members. Participants responding to the advertisements completed the online questionnaires, in keeping with a convenience sampling method.
Data collection
Four separate surveys, 1 relating to each core domain, were developed iteratively among members of the OMERACT PMR Working Group and hosted on the Web using Google Docs (see www.pmrgca.co.uk/content/omeract for the relevant links). The questionnaires included both closed and free-text questions. Each survey introduced the aims and purpose of the surveys, and the aims of OMERACT (Supplementary Material, available with the online version of this article).
The online surveys went live on April 24, 2018, and closed July 3, 2018. Unique identification numbers and time stamps were generated at the time the surveys were completed by participants. All surveys were in English, with respondents answering in English.
Data analysis and rigor
Participants were invited to answer each of the 4 surveys, which (1) asked for their assessment of the domain match and feasibility of each candidate instrument, and (2) provided a free-text section (with no word limit) to provide more in-depth feedback, particularly regarding missing items that were important to them regarding their experience of PMR. Respondents recorded when they were diagnosed with PMR, what dose of GC they were taking, and whether they felt their PMR was active currently. Finally, respondents were asked to give a global overall assessment of each of the identified candidate instruments as either green (I approve), amber (some reservations), or red (not right/should not be used).
Inductive thematic analysis was used to analyze the free-text responses. Individual participant responses were downloaded and the free-text answers broken into operative phrases and grouped by theme. Themes were identified independently by 2 researchers (MY and KG), with review by the executive PMR Working Group (which comprises clinicians and researchers with expertise in PMR), and patient partners.
This study complies with the Declaration of Helsinki. Ethical approval was obtained from the University of East Anglia research ethics committee (2017/18-81). Completion of the survey assumed consent. No identifiable data were collected. All downloaded information was kept confidential and individual responses recorded numerically in order of chronological completion of the survey.
RESULTS
A total of 78 participants with PMR completed at least 1 of the surveys and came from 6 countries (UK, France, USA, Canada, Australia, and New Zealand) representing 3 continents (Europe, North America, and Oceania). The disease duration ranged from new diagnosis to 17 years and doses of prednisolone ranged from 0 mg to 50 mg (9 had ceased GC therapy). Participants also stated whether they thought their PMR was active, with 64% reporting active disease, 23% in remission (of whom 28% were no longer taking GC), and 13% not sure.
There were 45 responses for the survey on pain and its measurement by VAS or NRS, and 45 for stiffness as measured by either VAS, NRS, or duration of morning stiffness. For physical function assessed by HAQ-DI, there were 78 responses, and 30 as assessed by mHAQ. Sixty-four responses were given on laboratory markers of inflammation as identified by either CRP or ESR.
Most respondents agreed the candidate instruments had face validity and were feasible to complete: full approval (green, “good to go”) for pain VAS was 66.7%, for pain NRS 59.3%, for stiffness VAS 51.9%, and for stiffness NRS 48.2%. The proportion with full approval for stiffness duration was 50%; for physical function, 68.5% for HAQ-DI and 53.3% for mHAQ; 56.3% of respondents agreed that CRP or ESR reflected disease activity regarding levels of inflammation. The overall judgment of the candidate instruments and the participants’ self-reported disease activity are shown in Table 1.
Participants’ global assessment of the proposed candidate instruments.
Respondents provided free-text answers (proportion giving free-text responses was 38% for the survey on pain, 53% for stiffness, 50% for HAQ-DI, 53% for mHAQ, and 66% for inflammatory markers), which allowed more nuanced feedback on use of the instruments under test.
There were 5 themes identified, based on respondents’ recollection of symptoms and their comments on the usefulness of the candidate instruments. The themes related to the variation, context, and location of pain; the location and timing of stiffness; fatigue; inability to perform tasks that had been uninhibited prior to their diagnosis of PMR; and the variable correlation of inflammatory markers with severity of symptoms.
Pain
Overall, respondents thought both a VAS or NRS could reflect pain, but gave suggestions on how this could be improved, with the expressed need to record the variation in pain symptoms that occur throughout the day, and record the effect of activity upon its severity.
Theme 1 — Variation in pain, contextualization and association to movement.
“There is no opportunity here to indicate whether the pain suffered is constant, or intermittent; or, whether it is worse at different times of day, or even if it is affected by weather, or the seasons of the year. Or if particular activities exacerbate the pain” (no. 35).
“I may not have pain while sitting, may have a pain of 50 [mm on a VAS] when turning over in bed and 75 [mm on a VAS] when bending to pick up an object. There needs to be context. Possibly combine with the disability assessment activity items” (no. 36, 37).
“Pain is variable at different times of the day and depends on what activities I try to do” (no. 24).
“Muscles more painful if sudden movement than slow. Hard to quantify” (no. 40).
Subtheme 1 — Site of pain. Respondents noted that there was no chance to state the site of pain.
“It needs breaking down to localized pain, i.e., pain in shoulders, pain [in] thighs, etc.” (no. 12).
Theme 2 — Location and timing of stiffness. Respondents commented on the lack of ability to describe where the stiffness affects them.
“Flexibility, strength, time, e.g., after starting to walk till stiffness starts, time from stiffness, e.g., on standing to being able to move again … how long after standing up are you able to walk?” (no. 52).
“Break down arms, legs, shoulders, neck” (no. 12). “It does not ask the site of the stiffness” (no. 46).
“Time of day, therefore relevance to effect of stiffness” (no. 10).
Subtheme 2 — Anchoring statements. With both the VAS and NRS, participants highlighted that anchoring statements might be helpful in quantification of stiffness.
“Perhaps an explanation as to what each number might mean in terms of stiffness, e.g., 50 = a distracting stiffness affecting range of movement” (no. 2).
“Stiffness with specific activities would provide a clearer picture” (no. 37).
“Examples of stiffness levels” (no. 19).
Physical health
Participants had several comments about the usefulness of HAQ-DI and mHAQ and its ability to record the effect of PMR on physical health; most participants highlighted the absence of questions concerning activities deemed difficult for patients with PMR. Another common theme was that fatigue is a common symptom in PMR that has a big effect on physical function, yet is not measured by either of these instruments.
Theme 3 — Fatigue.
“Fatigue is a major factor with ability to complete activities and should be measured” (no. 7).
“Brain fog, lack of focus, impaired judgement” (no. 34).
“There were no questions related to fatigue/tiredness as related to quality of life, or our inability to enjoy/take part in social events, or how we no longer ‘fit in,’ because our diminished mental capacity (memory, focus, etc.) excludes us from many conversations” (no. 52).
Theme 4 — Impairment of function: completion of tasks that previously were uninhibited. The activities were diverse and described many aspects of daily life. In particular, the tasks mentioned most frequently were driving an unadapted car, shopping, cooking, getting dressed, hanging out washing, and ability to carry out tasks related to personal hygiene. Although the HAQ-DI and mHAQ ask questions relating to activities of daily living, the following quotations reveal aspects that respondents think are missing.
“Cleaning teeth, picking up a pet or small child, making bed/washing bedclothes, applying cream or lotions, wiping yourself after going to the toilet” (no. 68).
“Can you prepare a hot cooked meal by yourself using a normal oven? Can you change the linen on your bed? Can you hang washing on a washing line to dry?” (no. 45).
“Are you able to get food in and out of your oven? Are you able to get food in and out of your refrigerator? Are you able to brush or dry your hair? Have you had to give up any forms of activity since developing your condition?” (no. 24).
“Has your PMR in the last week prevented you from completing any tasks, or activities you would normally be able to do? Have you had to take any or an increased amount of analgesia to be able to complete activities of daily living?” (no. 4).
“Could include fastening a bra — this was impossible for me and is a marker — can do it when on [prednisolone] but not otherwise — but doesn’t apply to men, clearly” (no. 74).
Overall, respondents preferred the HAQ-DI over the mHAQ, noting that a number of questions were missing from the mHAQ:
“Things such as reaching up for items and opening items requiring finer motor movements. [As in the first questionnaire (HAQ-DI) — which was much easier to complete too.]” (no. 28).
“Q9 — Climb a flight of 10 (or X number) of stairs. Q10 — Turn over in bed at night. Q11 — Sit down on the toilet. Q12 — Reach your arms vertically towards the ceiling/upwards” (no. 6).
“Housework tasks that involve bending down, kneeling down, or sitting on the floor cause problems. I can struggle down, but getting up needs furniture or another form of support. PMR affects specific muscle groups, e.g., shoulders, hips, knees — should the questions relate more specifically to activities involving these muscle groups?” (no. 19).
Inflammation
Regarding measurement of CRP or ESR, respondents commented that while they were useful for measuring inflammation levels, their doctor did not always monitor them, and that symptoms were a more important indication. Indeed, several participants stated that their inflammation markers did not seem to reflect their symptom flares.
Theme 5 — Symptoms and variable correlation to inflammatory marker readings.
“Mine were never raised so had hard time diagnosing … using them for me is not a good result” (no. 50).
“At the beginning of my illness there was a distinct correlation between ESR rate and my PMR, but recently sometimes I have a lowish reading but my symptoms are severe, and sometimes a higher reading and I don’t have severe symptoms” (no. 58).
“Need to be looked at together with symptoms as lots of things [can] cause raised markers” (no. 47).
Other participants found that measuring the CRP and/or ESR was useful in establishing a diagnosis.
“They let me know that the steroids I was prescribed after I was first diagnosed and 3 months later had effectively reduced the level of inflammation, which corroborated with the reduction in my symptoms. This was reassuring” (no. 15).
DISCUSSION
This study investigated the face validity and feasibility of candidate instruments for use in future studies enrolling individuals with PMR. From participants’ free-text responses, descriptive inductive thematic analysis was used to assess aspects of their lived experience of PMR that they felt were not covered by the current candidate instruments. The findings show that the majority of survey respondents agree that the identified candidate instruments were acceptable in terms of their face validity and feasibility. The responses were particularly helpful in determining which items relevant to the core domains of pain, stiffness, physical health, and inflammation were missing, or could be better measured than with the identified candidate instruments.
Five themes emerged from the descriptive thematic analysis: the variation, context, and location of pain; the location and timing of stiffness; fatigue; inability to perform tasks that had been uninhibited prior to onset of PMR; and the variable correlation of inflammatory markers to severity of symptoms. Opinions varied about the importance of inflammatory markers, with divergent views expressed on the topic. This appeared to be more of a challenge to the inclusion of the domain itself, rather than a preference for one instrument over another. This same issue had occurred previously during the Core Domain Set development work. However, both physicians and patients felt inflammatory markers were the best measure of pathophysiology, but mentioned caveats in interpreting results at the level of the individual patient. It is a limitation of our methodology that we were unable to explore this topic in more depth with the participants. It is important to note the differences between clinician-derived versus patient-reported outcome measures (PROM), with the former trying to identify objective pathological change of disease, and the latter the effects on quality of life. Both viewpoints should be taken into consideration.
There are advantages in using a well-established PROM such as the HAQ-DI, which has been evaluated in many rheumatic diseases but has many items that will not be specific to PMR. A disease-specific PROM may result in higher face validity for patients: feedback from this survey supports this, with requests for more detailed questionnaire items. There is possibly a role for both. However, no fully validated disease-specific PROM for PMR currently exists.
Among our survey responders, feedback about stiffness was particularly common from those who had active disease, with 39.4% identifying some concerns with the use of stiffness duration, compared to 27.3% for a VAS to record stiffness severity. The questions regarding stiffness for this survey simply asked respondents’ views as to whether a VAS or NRS allowed them to communicate their severity of stiffness. There were no lead-in questions such as “over the past week” or “within the last 24 hours”. It is possible that without a temporal frame to the questions, respondents felt unguided, leading to a higher proportion stating they had concerns. However, the free-text responses suggest that patients would prefer to report stiffness in terms of functional impairment or limitation9.
A qualitative study to investigate the effect of PMR on patients’ lives was carried out in the UK among 20 patients from South Yorkshire15. Five themes were identified: pain, stiffness and weakness, disability treatment and disease course, experience of care, and psychological effect. The same research team is now developing a PROM for PMR, and carried out a postal questionnaire of 28 patients with PMR from the community and hospital setting, to test the face validity of a “long-form” of the proposed PROM, assessed with a second questionnaire called QQ-1016. The research found concerns similar to those of our current study regarding the definition of stiffness, with participants stating that duration of stiffness as a measure failed to adequately reflect the effect of this symptom on their lives16.
From the free-text responses and resultant thematic analysis, it is clear that the experience of stiffness is complex and has contextual factors. It may be to the benefit of the medical community that a PROM is developed to better record and measure this important symptom for patients with PMR.
Regarding pain and stiffness identified by a VAS or NRS, the participants tended to modify or improve statements, such as including location of the symptom and anchoring statements. To determine physical function, participants preferred the HAQ-DI to the shorter mHAQ. However, fatigue was often cited and is missing from both the HAQ-DI and mHAQ. Fatigue is not just important to patients with PMR: qualitative research, replicated internationally, shows that patients with rheumatoid arthritis experience fatigue as uncontrollable and overwhelming, yet find it largely ignored by clinicians17. Currently, fatigue is included as an “important” rather than “core” domain for patients with PMR within the proposed OMERACT domain set, mostly owing to lack of consensus between clinicians and patients on whether it represents a fundamental feature of PMR. However, the results from our survey underline its importance for patients and the need for further research. Assessing properties of the candidate instruments is an aim of future work by the OMERACT group and it is interesting to see the composite score, Routine Assessment of Patient Index Data 3, which combines a VAS for pain, score of physical function, and VAS of patient global assessment, was more sensitive to change than ESR or duration of morning stiffness18. However, evaluating composite outcomes, which cross multiple domains of the Core Domain Set, was outside the scope of our study.
The strengths of our study are its size and the different nationalities of respondents; however, there are limitations. Since this was an online survey with participants self-completing the questionnaires, this may have resulted in a less than representative sample. For example, more actively motivated and possibly more affluent people might choose and have the resources to complete the online surveys. We did not ask for the age of participants, so were unable to draw any conclusions as to how this might affect the results or to generalize whether our respondents are typical for a cohort of patients with PMR. In addition, diagnoses could not be validated by applying classification sets designed to select homogeneous groups of patients to recruit to research studies: the reason for this was to survey as heterogeneous a sample as possible to receive the widest expression of views. The drawback of this approach is that it cannot be ascertained whether some of the symptoms described are attributable to a cause other than PMR. In addition, we cannot say which single outcome measure patients preferred over the others; identifying a “top” instrument was not the objective because the Core Domain Set had already been selected. Our research question was how individual instruments matched to each domain. Finally, the survey was in English, which may have limited responses from non-English speakers.
The domain match and feasibility of the proposed candidate instruments for each of the identified core domains are on the whole acceptable to patients with PMR. However, there is concern that several items of particular importance to patients, for example fatigue, remain missing from these outcome measures, suggesting that fatigue should move from “important” to a “core” domain within OMERACT. Further, participants consistently indicated that duration of stiffness, as an instrument, failed to reflect their real-life experience of this symptom. These may reflect a disconnect between traditional medical teaching about the manifestations of PMR and what it is like for patients who experience the disease. Ultimately, outcome measures that reliably identify disease extent and severity in clinical trials will be vital for the success of future novel therapeutic options in PMR.
ONLINE SUPPLEMENT
Supplementary material accompanies the online version of this article.
- Accepted for publication January 24, 2020.
REFERENCES
In this issue
