The PMR Cohort Study is an inception cohort of people diagnosed with PMR in general practice. The methods of the study have been described in detail elsewhere . Briefly, participants were referred from 382 general practices across England between June 2012 and June 2014. When seeing a potentially eligible patient, an electronic prompt alerted the GP to the study as specific disease codes were entered into the patient record, reminding them of the patients’ eligibility and the British Society for Rheumatology (BSR) guidelines for diagnosing PMR. In practices without electronic prompts in place, laminated cards with these details were placed, as an aide memoire, in consulting rooms.
Study packs, including written copies of the BSR PMR clinical guidelines were provided for each practice. General practitioners (GPs) referred patients who provided verbal consent to receive more information into the study via secure fax or email when they made a new diagnosis of PMR (no previous Read-coded consultation in the last three years). Participants were then sent a baseline questionnaire. Participants who did not respond within three weeks were sent a reminder questionnaire. Response to the baseline questionnaire implied consent to be followed up via postal survey at six further time points over two years (1, 4, 8, 12, 18 and 24 months). Participants were also asked whether they consented to the linkage of their survey data to their primary care medical records and nationally held registers. This study describes data from the baseline phase of the study only.
Baseline data collection
The baseline questionnaire collected information on PMR symptoms at the time of diagnosis, treatments received for PMR, general health, lifestyle, function, sociodemographics and participants’ views on the causes of their PMR.
The date of birth (used to calculate age), gender and postcode (used to calculate indices of multiple deprivation  of all patients referred into the study by their GP were available from the referral form. The baseline questionnaire also collected information on current employment status (employed, retired, unemployed, housewife/husband, sick or other), ethnicity (white, Asian/Asian British, black/African/Caribbean/African/black British, mixed or other), socioeconomic class based on current or last job title (“High managerial, administrative and professional”, “Intermediate” or “Routine and manual”, according to the Office for National Statistics criteria , whether the participant lived alone or not and marital status (married, separated, divorced, widowed, cohabiting or single).
General health, lifestyle and function measures
General health-related quality of life was assessed using the descriptive component of the Euroqol-5D (EQ-5D) . The five items of this questionnaire are combined to create an index where 1 is considered perfect health and 0 is considered to be representative of death or unconsciousness. The worst health state on this index is valued at -0.59 (i.e. worse than death).
Function in activities of daily living in the past week was assessed using the modified Health Assessment Questionnaire (mHAQ) [14, 15]. Scores on this 8-item instrument range from 0 to 3, with a score of less than 0.3 considered to represent normal functioning .
Fatigue in the past week was assessed using the 13-item Functional Assessment of Chronic Illness Therapy (FACIT)-fatigue questionnaire that has been shown to be valid and reliable for use with patients with systemic lupus erythematosus. Scores on the questionnaire range from 0 to 52 and higher scores indicate less fatigue. In the general US population aged 18 years and over, the mean score on this instrument has been shown to be 40.1 (SD 10.4) .
Insomnia in the previous two weeks was assessed using the 7-item Insomnia Severity Index (ISI), with individuals classified as having no clinically significant insomnia, subthreshold insomnia, clinical insomnia (moderate severity) or clinical insomnia (severe).
Anxiety and depression in the previous two weeks were assessed using the 7-item Generalised Anxiety Disorder (GAD) questionnaire  and the 8-item Patient Health Questionnaire , respectively. Anxiety was classified as none (score 0–4), mild (5–9), moderate (10–14) or severe (15–21). Depression was classified as none (score 0–4), mild (5–9), moderate (10–14), moderately severe (15–19) or severe (20–24). Participants were also asked whether they had someone they could rely on for emotional and practical support.
Smoking status (never smoked, ex-smoker or current smoker) and frequency of alcohol consumption (daily or almost daily, 3–4 times a week, 1–2 times a week, 1–3 times a month, special occasions or never) were reported by responders, along with height and weight, which were used to calculate body mass index (BMI).
Symptoms of and treatments for PMR
Participants reported the levels of pain and stiffness they were experiencing when their GP diagnosed them with PMR on a 0 to 10 scale where 0 was no pain/stiffness and 10 was “pain as bad as can be”/“very severe” stiffness, by circling the number corresponding to the level of pain/stiffness as appropriate. They were also asked to check boxes indicating which times of day they experienced stiffness (none, morning, lunchtime, afternoon, early evening, late evening or during the night), the duration of any morning stiffness (none, 1–15 minutes, 16–45 minutes, 46–60 minutes or >1 hour) and whether they could raise their arms above their head when they were diagnosed with PMR. Participants were also asked to report whether they used prednisolone or a range of other medications for their PMR.
Responders were compared to non-responders by age, gender and tertile of deprivation score using basic descriptive statistics. The sociodemographic data, general health and lifestyle and PMR-specific symptoms and treatments of the sample were described using means and standard deviations (SD), medians and interquartile ranges (IQR) and frequencies and percentages, as appropriate. The characteristics of the sample were compared between men and women using the chi-square, rank sum or t test as applicable. Levels of pain and stiffness were compared graphically and by calculating correlation coefficients.