Polymyalgia rheumatica is largely managed in primary care [2, 3], yet to date there is a paucity of research evidence from patients managed in this setting. As such, clinicians rely on data generated in secondary care that can be challenging to generalise to patients in the community, who may have less severe disease [20] and different prognostic trajectories. This study is the first to characterise a large primary care sample of PMR patients at the time of their diagnosis, thereby providing an unselected and more representative PMR population.

Participant demographics seen in this inception cohort are similar to those recruited from hospital populations, with an average age at disease onset of approximately 72 years and around two thirds of participants being female. Also in keeping with existing evidence on the propensity of this condition to affect those of Northern European descent, the cohort self-identifies almost entirely as being of white ethnicity, despite having recruited from a large range of geographical areas with differing levels of ethnic diversity. Similarly to a previous population study, there is a broad spread of occupational class across the sample [21]. This suggests that hospital PMR cohorts have similar demographics to this unselected primary care sample, providing some reassurance as to the accuracy of GP diagnosis for participants in this cohort. However, unlike the classic description of stiffness predominantly in the morning, within this cohort stiffness was severe and diurnal, with more than 60 % also reporting stiffness in the evening and the majority also describing night-time stiffness. In part this may reflect questionnaire design, in that participants were asked specifically about the presence of stiffness at other times of the day, although as qualitative data in both patients with rheumatoid arthritis [22] and PMR [23] highlights that stiffness may be more varied and complex than is usually reported and is present at various times of day, not just in the morning. Given that stiffness is considered key as part of the core domain set for PMR by Outcome Measures in Rheumatology (OMERACT) [24], we felt that timing of stiffness warranted further exploration within the cohort.

The reported high levels of pain and stiffness that persisted throughout the day (especially in women) may explain the high proportion reporting use of pain killers and non-steroidal anti-inflammatory drugs in addition to prednisolone, especially in women. Furthermore, the greater severity of symptoms in women is in concordance with a previous case series from a tertiary referral clinic [25].

Comparison of the current study to those reported in the literature suggests that overall levels of quality of life [26, 27], depression [28, 29] and obesity [28, 29] are similar to those expected in the population of this age, and levels of daily alcohol consumption [29], current smoking [8, 29] and anxiety are lower [28], although the age structure of available comparator populations varies between studies.

Fatigue was greater in the sample than the in general US population [17], and physical functioning was much better in the PMR population (n = 85) used to evaluate change in patient-reported outcomes in PMR [30] where the baseline median (IQR) mHAQ scores were much higher (1.1 (0.8, 1.6)) than in this cohort, although the scores at one week (0.4 (0.1, 0.8)) were similar to those in the current study (0.4 (0.0, 1.0)). In part, this may reflect patient recall bias, with patients asked to recall baseline disability or may reflect patients with a more severe spectrum of disease being referred to specialist services; however, reported duration of morning stiffness and oral glucocorticoid use were similar between studies, although this cohort had higher median (IQR) levels of pain (8 (7, 9) on a numerical rating scale) compared to this American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) sample [30].

Whilst this study has many strengths, including its size and the availability of detailed information, it also has a number of limitations that need to be considered. First, the sample consists of patients with a GP diagnosis of PMR. Diagnosing PMR is notoriously difficult, without a single gold-standard diagnostic test to inform clinicians. The recent ACR/EULAR classification criteria highlight the challenges in diagnosing PMR, with 10 out of 128 people (7.8 %) diagnosed with PMR by specialists subsequently having their diagnosis revised [30]. Although rheumatology specialists may criticise the use of GP diagnosis in this study, the majority of patients with PMR are exclusively managed in the community and as such GPs and rheumatologists are likely to encounter a different spectrum of disease. In this study, we optimised the diagnostic process by providing BSR guidance in the consultation to support decision-making. Furthermore, the rate of referral into the study of around 4.4 patients per 10,000 per year in those aged over 40 years, is similar to or below that previously reported in primary care data (e.g. see [4]) providing further confidence in our cohort.

A further limitation is the reliance on self-reported data, and although questionnaires were mailed to participants as soon as a diagnosis was made, some recall was required. This is necessary for many of the constructs, as they cannot be objectively measured (e.g. pain, stiffness and quality of life), although for others (e.g. medications) this is not necessarily the case. However, additional information on the patient’s attribution of their medication to PMR was possible through this use of self-reported data. At future stages of the study, these data will be linked to primary care consultation and prescription records.

PMR has generally been considered to be a relatively mild condition, associated with few other health complications. However, these patients report high levels of pain and stiffness, which is prolonged after the morning time, along with high levels of oral steroid use and polypharmacy. It is also clear that women report worse scores than men in almost all aspects of disease activity, functioning and mental health. In addition, they are more likely to report taking a range of medications in addition to glucocorticoids for their PMR, notably strong painkillers and anti-depressants.

Although this is the first study to characterise a cohort of patients with incident PMR in primary care, further work is needed to establish whether this group would all be considered to have PMR if they visited a rheumatologist. Further studies will be required to determine whether those with more depression or poorer physical functioning have a worse prognosis over the course of their PMR.