Patient-reported outcomes in trials of patients with polymyalgia rheumatica: a systematic literature review

Polymyalgia rheumatica (PMR) is a common inflammatory disease of elderly patients, affecting from 0.1 to 0.5 % of over 50- year-olds [1]. It is characterized by proximal pain, especially in the shoulder and pelvic girdle, and morning stiffness with high acute-phase reactants, measured by erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP). A “gold standard” test to make the diagnosis is not available. Physicians mainly rely on the clinical picture, and the diagnosis is further supported by a rapid response to glucocorticoids. The initial presentation may mimic other conditions or present with the absence of acute-phase reactants. This heterogeneity may explain the difficulty of diagnosis and a lack of agreement between physicians [2]. In addition, a lower acute-phase response does not necessary indicate lesser severity or better prognosis [3].

Glucocorticoids are the preferred treatment, leading to a rapid and dramatic improvement, but they may be required for several years in some patients [4]. For this reason, PMR is a common indication for long-term steroid use in the community, and it has been associated with serious adverse effects such as diabetes, osteoporosis, and infections [5].

Although a rapid resolution of symptoms after treatment is a diagnostic hallmark, there is no consensus on what constitutes an appropriate response and which outcomes should be monitored. The lack of reliable and sensitive measures to evaluate disease activity and the lack of standardized classification criteria to identify patients with PMR may explain why the evidence for efficacy of any treatment different to glucocorticoids remains limited [6].

Current clinical guidelines recommend monitoring patients treated for PMR on the basis of symptoms since, as we previously noted, conventional inflammatory markers can be misleading [6]. Reliable and comparable outcomes are required not only to balance the benefits and adverse events of a long-term steroid therapy, but also to investigate the potential use of other drugs such as corticosteroid-sparing agents.

Patient-reported outcomes (PROs), defined as outcomes that are completed by patients, have been increasingly recognized as important measures over the past few years by rheumatologists. They incorporate the patient’s perspective of the disease capturing the impact of the disease in patient’s lives, and they perform well in assessing disease activity in patients with PMR [7]. Different PROs such as pain, morning stiffness, or physical function have been proposed as recommended outcome measures to be used in practice and clinical trials [8]. In addition, most remission or flare definitions include at least one self-reported variable from the medical history [9].

The goal of this study was to review the frequency of use of PROs in published studies including randomized controlled trials and cohorts of PMR through a systematic literature review.

To obtain all published articles reporting any type of PROs in PMR, the literature search was performed in PubMed database on April 13, 2015. Publications were identified using the following MeSH term: “polymyalgia rheumatica” with a limitation to “English” and “clinical trials.” Inclusion criteria comprised articles reporting any type of clinical design, patients with PMR, and including at least the evaluation of any PRO measure. Articles were excluded if they did not concern PMR or if they did not focus on PROs. Editorials and letters were also excluded to provide an overview of the use of PROs in original research articles. A first screening was performed based on titles and abstracts of the articles by AH and IC. The articles fulfilling the inclusion criteria were retrieved for a full paper review.

Data were obtained on year of publication, study design (RCT, prospective cohort), follow-up, number and characteristics of patients, treatment under evaluation, and PROs included. A description of the articles reporting PROs is presented as a table.

All information about PROs was collected, including composite indices, which included at least one PRO, and then classified into domains. Results are presented as frequency of reported domains and frequency of tools to evaluate each domain in RCTs and cohorts as a descriptive analysis.

Of the 118 publications identified by the literature search, 28 were selected for full review and 20 finally included in the analysis. Ninety publications were excluded after screening of title and abstract (Fig. 1).

The characteristics of the included publications are given in Table 1 in chronological order. Of the 20 publications, 10 (50 %) were RCTs, 8 (40 %) were prospective cohorts, 1 (5 %) was a case control, and 1 (5 %) was a pilot observational study. Patients’ characteristics were typical of PMR populations, with a mean age ranging from 62.5 to 76.6 and female gender ranging from 50 to 90.7 %. Number of patients included ranged from 6 to 213. In 13 (65 %) of these articles, different glucocorticoid regimes or other drugs such as corticosteroid-sparing agents were evaluated. The remaining 7 (35 %) studies evaluated predictors of vertebral fractures, the impact of PMR on clinical outcomes, gender differences in PMR patients, laboratory measures to identify relapse, diagnostic, and response criteria.

Seven domains were reported in the 20 articles: pain, morning stiffness, function, patient global assessment, fatigue, quality of life, and anxiety and depression (Table 2).

Only eight domains reflecting patient-reported outcomes were found in this review including trials of patients with PMR. The two domains most frequently reported were pain and morning stiffness.

Pain is a very important symptom for PMR patients, and it has been included in each set of diagnosis criteria proposed since the first one by Bird and Wood in 1979 [13‐16]. Pain has also been shown to be important to evaluate response to therapy. In 2003, the European Collaborating PMR group proposed the first response criteria for PMR based on a core set of five variables [17]. Pain on a VAS was selected as the central measure for disease activity being the only one mandatory in this core set. The selection was based on the dominant role of pain in patients’ symptoms and has proved sensitivity to change. Morning stiffness was also included in this response criteria core set with CRP, elevation of upper limbs, and the doctor’s global assessment, but only a change in three of these four is required to reflect a change in disease activity. Having two PROs included in a set of only five variables for response highlights the importance of patient self-evaluation in PMR. In addition, based on this core set and in analogy with a simplified disease activity index (SDAI) for rheumatoid arthritis (RA), a disease activity index for PMR has been proposed [18]. This composite index, the PMR activity score (PMR-AS), includes pain on a VAS and morning stiffness in minutes multiplied by 0.1 to avoid a high weighting of this specific symptom. PMR-AS shows a high correlation not only with patient’s global assessment, but also with patient satisfaction. Using this composite index helps describe the clinical situation and adds feasibility. Having a score as an absolute number also helps comparing patients much more easily.

In the majority of the studies included in this review, pain was recorded on a VAS referring mainly to proximal pain in the shoulder or pelvic girdle. In some studies, pain was evaluated as an outcome to compare treatment groups [19‐24], or to evaluate disease activity differences according to gender [25], or as a potential predictor of vertebral fractures [26].

Morning stiffness was the second most reported PRO in this review mainly evaluated by duration but with no consistency about how this was defined or collected. Morning stiffness is considered an important diagnostic clue in PMR, but it is difficult to measure accurately especially when using duration of morning stiffness that has been reported to show poor test–retest reliability in PMR [7]. A possible explanation for this poor test–retest could be the fluctuation of this symptom during the day. From the patient’s perspective, morning stiffness has been defined as a restriction of movement or better described as what it has prevented them from doing. From patient experience, morning stiffness is less responsive to glucocorticoids in comparison with pain [27]. Morning stiffness has also been included in different diagnostic criteria—lasting for more than 1 h [13, 14], as part of the response criteria previously described [17], and in the PMR-AS [18].

The two PROs included in most of the diagnostic criteria were also the most frequently reported in clinical trials: pain and morning stiffness. The evaluation of pain has been included in each one of these criteria as pain/aching or tenderness in shoulder or pelvic girdle area, in shoulders, upper arm, hips or pelvis, and thighs reflecting how pain is experienced by the patients in the course of the disease. Pain is an overwhelming symptom for patients always part of the clinical picture. Although it is often not well localized, it tends to be more responsive to medication in comparison with other symptoms [27]. This may explain why it is also included as an outcome in most clinical trials. In contrast, morning stiffness was included only in three of these diagnostic criteria and reported less frequently in clinical trials. It is also considered an important diagnostic clue in PMR but more difficult to evaluate and poorly responsive to treatment.

Function through HAQ or MHAQ was only reported in the 25 % of the articles. This is a surprisingly low percentage having taken into account that both are generic instruments that can be used in any rheumatic diagnosis [28]. Function correlates with other measures of disease activity in PMR and is responsive to change [29, 30]. Moreover, function is a strong predictor of mortality not only in patients with RA [31], but also in the general population [32]. For patients with PMR, being able to perform common activities of daily living was described as the most important aspect of their disease that would indirectly reflect their morning stiffness [27].

Other domains such as patient global assessment, fatigue, quality of life, and anxiety and depression were infrequently reported, though they appear important from the patient’s point of view [8].

Pain, function, and patient global are part of an established core set of PROs used across most rheumatic diseases [33]. It is surprising that only pain is a well-established PRO in PMR patients. A possible explanation is that in contrast with other rheumatic diseases, the typical increase in ESR (40 mm/h or more) and CRP plus the prompt response of symptoms to low-dose glucocorticoids seen in PMR patients may facilitate the monitoring of the disease activity. Laboratory data and response to treatment may be considered as more objective outcomes compared to PROs for most rheumatologists explaining why other PROs would not be included in these clinical trials.

A limitation for this study is that only PubMed was consulted during the search; no other databases such as EMBASE or the Cochrane Library were reviewed. This may limit the exhaustiveness of the actual review but give a reasonable overview regarding use of PROs.

Thus, in conclusion, although PROs can be useful for better monitoring of disease activity and evaluating treatment response in PMR, only pain on a VAS is systematically included. While morning stiffness is an important symptom for patients, there is no consistency about how it should be measured. In addition, no PROs have been defined yet for PMR as part of a core set in the rheumatology community. An OMERACT special interest group is working toward the development of a core set of outcomes for PMR. Although several PROs have been identified to be of importance, such as pain, stiffness, fatigue, sleep disturbance, function, and anxiety and depression, further validation work is still in progress [33]. Additional work is also needed to obtain a better insight of which outcomes should be necessary to incorporate the patient’s perspective.