Prevalence of frailty in patients with polymyalgia rheumatica and association with health-related quality of life, cognition and sarcopenia

Abstract

Objective

To describe the prevalence of frailty in a single-centre cohort of patients with PMR and describe its association with health-related quality of life (HRQoL), cognition and sarcopenia.

Methods

This was a cross-sectional study of patients with PMR, according to 2012 EULAR/ACR Classification Criteria, presenting within 12 months of diagnosis and on treatment with glucocorticoids. Frailty was defined according to the Fried frailty criteria. HRQoL was assessed using Patient-Reported Outcomes Measurement Information System Computerized Adaptive Test (PROMIS-CAT) and cognition was assessed using the Mini-Mental State Examination. Sarcopenia was measured by DXA.

Results

Forty-one patients were enrolled. Prevalence of frailty and pre-frailty was 17% and 59%, respectively. Frail patients had higher inflammatory markers at diagnosis compared with pre-frail and robust patients. Of 27 patients with DXA results, 26% were sarcopenic. Frail patients had worse physical function, and more pain behaviour and interference compared with pre-frail and robust patients. In univariable analyses, frail patients were more likely to have worse physical function, and more pain behaviour and pain interference, which remained significant after adjusting for age. There were no significant associations between cognition or sarcopenia and frailty.

Conclusions

In this cohort of PMR patients, there was a higher prevalence of frailty and pre-frailty compared with that reported in community-dwelling elderly. Frailty was associated with worse physical function, and increased pain behaviour and pain interference, differences that were also clinically meaningful. Larger prospective studies are needed to confirm these findings and analyse the association of frailty with other PMR disease outcomes.

Rheumatology key messages.

• We report a high prevalence of frailty and pre-frailty in patients with PMR.

• Frail PMR patients had higher inflammatory markers at time of diagnosis.

• Frail PMR patients reported increased pain behaviour and interference, and worse physical function.

Introduction

Frailty is a syndrome characterized by decline in physiologic reserve and homeostasis, leading to increased vulnerability to stressors and poor outcomes, including disability, falls, hospitalization and mortality [1]. Higher prevalence of frailty has been described in individuals with chronic conditions in whom frailty also has been shown to be an independent predictor of poor outcomes [2, 3]. Frailty has been associated with chronic low-grade elevation of inflammatory markers such as CRP and IL-6, a state referred to as ‘inflammageing’ [4, 5]. Furthermore, IL-6 and other markers have shown a stronger association with frailty than age, as well as association with functional dependence and mortality [6, 7]. Frail patients are at a higher risk for sarcopenia and cognitive impairment, and both sarcopenia and cognitive impairment have been reported in association with chronic low levels of inflammation [8, 9].

PMR is one of the most common systemic rheumatic diseases, occurring most commonly in patients aged 50 years and older, with a peak incidence between 70 and 80 years of age [10]. PMR is usually regarded as a self-limiting condition requiring 12–18 months of treatment with glucocorticoids; however, some studies have reported prolonged symptoms such as fatigue, pain and stiffness [11, 12]. Furthermore, poor physical and mental quality of life has been reported during the first year of PMR [13].

Although patients with PMR would be expected to be at higher risk of frailty due to their age, cumulative inflammation and glucocorticoids exposure, the prevalence of frailty and associations of frailty with health-related quality of life (HRQoL) in PMR patients have not been previously described. The objective of this study was to describe the prevalence of frailty in a single-centre cohort of patients with PMR, and describe its association with HRQoL, cognition and sarcopenia.

Methods

Study participants

This was a single-centre cross-sectional study of patients with PMR. Potential patients were identified between March 2019 and January 2020 via electronic medical record (EMR) review using International Classification of Diseases-10 codes for PMR (M35.3). Inclusion criteria were fulfilment of 2012 EULAR/ACR Classification Criteria, 12 months or fewer since time of diagnosis and current treatment with any dose of glucocorticoids [14]. Exclusion criteria included concomitant diagnosis of GCA, seronegative RA or other inflammatory rheumatologic condition (as deemed by the treating rheumatologist); end-stage renal disease; oxygen-dependent pulmonary disease; active malignancy or malignancy within the past 5 years (other than squamous cell carcinoma, basal cell carcinoma or carcinoma in situ after curative treatment); prior solid organ transplantation; or HIV. The study was approved by the Hospital for Special Surgery Institutional Review Board (#2018-1823), and all participants gave written consent.

Variables

Disease-specific characteristics

Date of diagnosis, initial and current glucocorticoid dose, and steroid-sparing agent use, if any, were obtained from EMR review at the time of eligibility screening. Current glucocorticoid dose was confirmed during the study visit. ESR and CRP, both at time of diagnosis and the most recent measurements (taken during the last visit with treating rheumatologist), were obtained from the EMR. Disease activity was measured using the PMR Activity Score (PMR-AS), as provided by treating physicians [15]. The PMR-AS is a composite score that combines patient-reported pain, physician global, duration of morning stiffness, range of elevation of upper extremities and CRP. <1.5 indicate disease remission, while scores <7, 7–17 and >17 indicate low, medium and high disease activity, respectively.

Other variables

Sociodemographic characteristics (age, race, ethnicity, education level, smoking and alcohol use) were obtained from patients. Comorbid conditions were quantified using the Charlson Comorbidity Index (CCI) [16]. Height, weight and BMI (kg/m²) were recorded from the EMR. Independence in activities of daily living (ADLs) and instrumental activities of daily living (IADLs) were measured with the Katz Independence in Activities of Daily Living and the Lawton-Brody Instrumental Activities of Daily Living scales, respectively [17, 18]. Information on self-reported glucocorticoid-associated adverse effects was also collected.

Outcomes

Frailty

Frailty was measured using the Fried frailty criteria, considered the gold standard phenotypic definition validated in older adults [19, 20]. The Fried frailty definition includes five criteria: (i) low BMI or weight loss (≤18.5 kg/m², or self-report that lost ≥10 pounds in the last year), (ii) exhaustion [measured by two items from the Center for Epidemiological Studies Depression scale (CESD)], (iii) Inactivity/low energy expenditure [measured by the International Physical Activity Questionnaire (IPAQ)], (iv) slow 4-m gait speed (adjusted for age and height) and (v) low grip strength (adjusted for sex and BMI) [21, 22]. Patients with three or more criteria are considered frail; patients fulfiling one or two criteria are considered pre-frail and those fulfiling no criteria are considered robust (supplementary Table S1, available at Rheumatology online).

Cognition

The Mini-Mental State Examination (MMSE) is a test of cognitive function among older adults that includes tests of orientation, attention, memory, language and visual–spatial skills [23]. Scores range from 0 to 30 points: 24 points is considered normal cognition, 19–23 points mild cognitive impairment, 10–18 points moderate cognitive impairment and ≤9 points severe cognitive impairment.

Patient-reported outcomes

The Patient-Reported Outcomes Measurement Information System (PROMIS) is a National Institutes of Health-developed validated framework for capturing patient-reported outcomes [24]. PROMIS measures have been normalized to the US population with an average score of 50 and an s.d. of 10, allowing comparison with the general population and for patients with various chronic diseases. For this study, we used the Computer Adaptive Testing (CAT) version of PROMIS. This instrument reduces responder burden by selecting the most informative questions from the item bank based on the subjects’ previous responses, therefore allowing the use of fewer questions per domain with more precision. The following PROMIS-CAT domains were assessed: anxiety (v1.0), anger (v1.1), depression (v1.0), fatigue (v1.0), pain behaviour (v1.0), pain interference (v1.1), physical function (v2.0), satisfaction with discretionary social activities (v2.0) and cognitive function (v2.0). Based on the reference of 0.5 s.d., minimal important differences (MID) of 5 points in PROMIS scores are considered clinically significant [25, 26].

Sarcopenia

Body composition was measured by DXA. Sarcopenia was defined using appendicular skeletal muscle mass (ASM) and height (in m), ASM/m², as recommended by the European Working Group on Sarcopenia in Older People (EWGSOP) 2, using cutoffs of <7.0 kg/m² and <5.5 kg/m² for men and women, respectively [27].

Statistical analysis

Percentages, medians, and interquartile ranges (IQRs) are reported as descriptive statistics. Fisher’s exact test and the χ² test were performed to calculate P-values for categorical variables. T-tests and Kruskal–Wallis tests were performed for continuous variables that were normally and non-normally distributed, respectively. CIs for prevalence estimates were calculated. Univariable and age-adjusted linear regression models were performed to describe the association of PROMIS outcomes, sarcopenia and MMSE with frailty. Statistical significance was determined by using an alpha of 0.05. Stata, version 14 was used to perform all analyses.

Results

Patients were recruited via EMR between 1 April 2019 and 31 January 2020. Disease duration longer than 12 months was the main reason for exclusion during initial screening (Fig. 1). Sixty-nine patients were identified; 41 subjects who were enrolled completed the study visit (Fig. 1). Median age was 73 (IQR 67.1, 77.5) years, and patients were mostly males (51.3%), White (92%) and non-Hispanic (95.1%).

Fig. 1.

Flow diagram of patient recruitment

ICD: International Classification of Diseases; GC: glucocorticoids.

Frailty prevalence and classification

Based on Fried frailty criteria, 7 (17%, 95% CI 7.15, 32.06), 24 (59%, 95% CI 42.11, 73.68) and 10 (24%, 95% CI 12.36, 40.30) patients were classified as frail, pre-frail and robust, respectively (Table 1 and Fig. 2). Frail patients were older, with a median age of 75.8 years (IQR 70.0, 78.5), and more commonly were female (85%). Pre-frail patients were older than robust patients, with a median age 73.5 years (IQR 67.4, 78.0), without any differences in sex. Duration of disease and median glucocorticoid exposure duration was similar among all groups. Among all patients, a high proportion of patients had college education or greater (37/41, 90.2%) and a low burden of comorbidities [median CCI 2 (IQR 1, 3)]. Most patients had either mild disease activity or remission as classified by the PMR-AS score. Only 11 patients (27%) had associated peripheral arthritis symptoms (data not shown).

Table 1.

Baseline characteristics by Fried frailty classification

Characteristic	Robust	Pre-frail	Frail
	N = 10	N = 24	N = 7
Age, years, median (IQR)	68.8 (60.8, 76.4)	73.5 (67.4, 78.0)	75.8 (70.0, 78.5)
Sex, female, n (%)	4 (40)	10 (42)	6 (85)
Race, n (%)
White/Caucasian	10 (100)	23 (96)	5 (71)
Asian	0 (0)	1 (4)	0 (0)
Other	0 (0)	0 (0)	2 (29)
Ethnicity, n (%)
Hispanic	0 (0)	0 (0)	2 (29)
Non-Hispanic	10 (100)	24 (100)	5 (71)
Education, n (%)
High school/GED or less	0 (0)	1 (4)	1 (14)
Some college	0 (0)	2 (8)	0 (0)
College	2 (20)	7 (29)	2 (29)
Graduate/advanced professional degree	8 (80)	14 (58)	4 (57)
Smoking, n (%)
Never	4 (40)	14 (58)	2 (33)
Ever	6 (60)	10 (42)	5 (71)
Alcohol use, n (%)
Never	1 (10)	5 (21)	0 (0)
Ever	9 (90)	19 (79)	7 (100)
Charlson Comorbidity Index, median (IQR)	1 (1, 2)	2 (1, 3)	3 (1, 3)
Number of medications taken, median (IQR)	5 (3, 5)	5 (3, 7.5)	3 (2, 8)
Duration of disease, days, median (IQR)	134.5 (75, 278)	129 (73, 250)	131 (15, 280)
PMR-AS score, median (IQR)	3.45 (0.7, 4.7)	1.2 (0.7, 4.8)	4.0 (2.4, 5.5)
CRP at diagnosis (mg/l), median (IQR)	18.0 (7.0, 30.0)	32.0 (18.0, 53.0)	76.0 (30.0, 87.0)
ESR at diagnosis (mm/h), median (IQR)	35 (30, 42)	52 (34.5, 73)	58 (39, 81)
CRP (mg/l), median (IQR)	7.0 (5.0, 7.0)	7.0 (5.0, 7.0)	7.0 (5.0, 8.0)
ESR, (mm/h), median (IQR)	3.5 (2, 19)	9.5 (5, 17.5)	16 (8, 19)
Prednisolone-equivalent dose (mg/day), median (IQR)	5 (3, 8)	6.5 (3.1, 10.5)	7 (5, 7.5)
Glucocorticoid exposure duration, days, median (IQR)a	127.5 (63, 273)	111.5 (51, 243.5)	131 (15, 280)
Steroid sparing agent use, n (%)	0 (0)	3 (12)	0 (0)
BMI (kg/m2), median (IQR)	26.0 (24.8, 27.3)	26.5 (23.8, 29.9)	24.3 (21.3, 30.1)
Disability, n (%)
ADL disability	3 (30)	3 (13)	0 (0)
IADL dependent	0 (0)	2 (8)	0 (0)

^aGlucocorticoid duration calculated from initiation of medication to date of baseline visit.

IQR: interquartile range; PMR-AS: PMR Activity Score; ADL: activities of daily living; IADL: instrumental activities of daily living.

Fig. 2.

Frailty and frailty components in patients with PMR

(A) Frequency of individual frailty components based on Fried frailty phenotype. (B) Patients by frailty classification based on Fried frailty phenotype.

CRP levels at time of diagnosis differed among frail, pre-frail and robust patients [median 76.0 (IQR 30.0, 87.0) vs 32.0 (18.0, 53.0) vs 18.0 (IQR 7.0, 30.0) mg/l, respectively], and a similar, but non-significant difference was noted for ESR at time of diagnosis. No differences were observed for either CRP or ESR at the time of the study visit, and no differences in other baseline characteristics were observed among patients by frailty classification.

Frailty components

Weakness was the most common criterion fulfiled by patients (21/41, 51%), followed by exhaustion (17/41, 41%) (Fig. 2, supplementary Table S1, available at Rheumatology online). Slowness was the least commonly encountered criteria, fulfiled by only one (2%) patient.

Frailty and glucocorticoid-associated adverse effects

All patients reported experiencing at least one glucocorticoid-related side effect, and most patients reported having three or more (Table 2). Except for a difference in the reported frequency of tremors and new-onset high blood sugars, there were no significant differences in the reported frequency of individual glucocorticoid-associated adverse events or the cumulative frequency of them among patients by frailty classification.

Table 2.

Reported glucocorticoid-associated adverse effects by frailty category

	Robust	Pre-frail	Frail
	N (%)	N (%)	N (%)
Face swelling	4 (40)	8 (33)	4 (57)
Muscle cramps	5 (50)	10 (42)	3 (43)
Sleep disorders	6 (60)	10 (42)	5 (71)
Mood disorders	3 (30)	10 (42)	3 (43)
Increased appetite	2 (20)	10 (42)	1 (14)
Hirsutism	1 (10)	4 (17)	0 (0)
Skin thinning	3 (30)	9 (38)	4 (57)
Altered wound healing	2 (20)	2 (8)	1 (14)
Lower extremity swelling	2 (20)	5 (21)	2 (29)
Epigastric pain	0 (0)	6 (25)	1 (14)
Tremors	1 (10)	3 (13)	4 (57)
Muscle weakness	3 (30)	9 (38)	2 (29)
Hyperglycemia	3 (30)	0 (0)	3 (43)
Uncontrolled diabetes	1 (10)	1 (4)	1 (14)
Increase in blood pressure	2 (20)	5 (21)	1 (14)
Thrush	0 (0)	0 (0)	0 (0)
Cataracts	1 (10)	5 (21)	2 (29)
Osteoporosis	2 (20)	4 (17)	2 (29)
Avascular necrosis	0 (0)	1 (4)	0 (0)
Cumulative side effects
1–2	0 (0)	3 (13)	1 (14)
≥3	10 (100)	21 (88)	6 (86)

Frailty classification, physical function and health-related quality of life

A difference in the scores for pain behaviour and pain interference (e.g. higher t-score, more pain) was observed among the three different groups (Table 3). A difference was also noted for physical function (e.g. lower t-score, worse physical function) and was statistically significant in the univariable model for frail patients (Table 4). A similar pattern was observed for fatigue (e.g. higher t-score, more fatigue). Frail patients also were more likely to report higher pain behaviour (behaviours that typically indicate to others that the individual is experiencing pain) and pain interference (extent to which pain hinders engagement in daily activities). This association remained significant after adjusting for age among frail patients (Table 4).

Table 3.

Patient-reported outcomes, mini-mental score and sarcopenia by Fried frailty classification

	Robust	Pre-frail	Frail
	N = 10	N = 24	N = 7
	Median (IQR)	Median (IQR)	Median (IQR)
PROMIS t-score
Physical function	53.7 (41.4, 55.2)	48.5 (43.5, 53.7)	39.55 (35.7, 46.8)b
Fatigue	48.5 (38.2, 52.8)	48.5 (45.1, 50.7)	53.3 (48.5, 60.5)
Pain behaviour	51.3 (35.3, 57.5)	35.3 (35.3, 52.6)	57.5 (56.3, 58.8)b
Pain interference	46.6 (38.7, 54.3)	46.6 (38.7, 52.6)	57.6 (55.8, 60.3)b
Anxiety	52.0 (49, 53.5)	50.8 (46.7, 52.1)	55.9 (51.2, 61.5)b
Anger	48.2 (44.3, 52.3)	47.5 (42.1, 52)	50.15 (47.8, 54.7)
Depression	48.2 (46.9, 50.8)	46.1 (42.6, 49.9)	48.95 (48.2, 57.6)
Satisfaction with social role	52.4 (47.4, 64.6)	54.0 (45.8, 60.2)	49.05 (45.8, 58.3)
Cognition	52.3 (49.1, 60.7)	49.3 (45.8, 54.4)	47.15 (41.5, 68)
MMSE score	29 (29, 30)	29 (28, 30)	28 (28, 30)
Sarcopenia (n = 27)a, n (%)	1 (10)	5 (21)	1 (14)

Values are presented as median (IQR), except where indicated.

^aSarcopenia defined as EWSOG-2 criteria.

^bClinically meaningful difference.

PROMIS: Patient-Reported Outcomes Measurement Information System; MMSE: Mini-Mental State Examination; IQR: interquartile range.

Table 4.

Univariate and age-adjusted cross-sectional analysis of patient-reported outcomes, MMSE and sarcopenia by Fried frailty classification

	Robust	Pre-frail		Frail
		Univariate coefficient (95% CI)	Age-adjusted coefficient (95% CI)	Univariate coefficient (95% CI)	Age-adjusted coefficient (95% CI)
PROMIS
Physical function	Ref.	−0.53 (−6.18, 5.12)	0.48 (−4.67, 5.63)	−9.15 (−16.62, −1.68)	−6.75 (−13.70, 0.20)
Fatigue	Ref.	3.06 (−3.12, 9.23)	2.42 (−3.69, 8.54)	7.78 (−0.39, 15.95)	6.28 (−1.97, 14.54)
Pain behaviour	Ref.	−3.49 (−10.92, 3.94)	−2.92 (−10.39, 4.56)	10.1 (0.27, 19.92)	11.4 (1.36, 21.54)
Pain interference	Ref.	−1.06 (−7.06, 4.94)	−0.63 (−6.68, 5.43)	11.6 (3.61, 19.50)	12.6 (4.40, 20.75)
Anxiety	Ref.	−1.85 (−7.12, 3.42)	−2.68 (−7.56, 2.19)	6.18 (−0.75, 13.10)	4.12 (−2.42, 10.66)
Anger	Ref.	−0.37 (−5.53, 4.80)	−0.88 (−6.00, 4.25)	2.47 (−4.36, 9.30)	1.26 (−5.66, 8.17)
Depression	Ref.	−4.08 (−8.95, 0.78)	− 4.45 (−9.35, 0.45)	−0.56 (−7.00, 5.88)	−1.43 (−8.05, 5.18)
Satisfaction with social role	Ref.	−2.75 (−10.31, 4.81)	−2.99 (−10.73, 4.74)	−6.97 (−16.97, 3.04)	−7.55 (−17.99, 2.89)
Cognition	Ref.	−3.68 (−10.56, 3.21)	−3.93 (−10.96, 3.10)	−2.58 (−11.69, 6.52)	−3.18 (−12.68, 6.31)
MMSE score	Ref.	−0.5 (−1.55, 0.55)	−0.43 (−1.49, 0.63)	−0.90 (−2.26, 0.45)	−0.74 (−2.14, 0.66)
Sarcopeniaa	Ref.	2.37 (0.24, 23.36)	1.78 (0.16, 19.84)	1.50 (0.08, 28.89)	0.70 (0.03, 17.12)

Values are presented as β parameter estimate (P-value) form multiple linear regression analysis. Bold values show statistically significant associations.

^aSarcopenia defined as EWSOG-2 criteria.

PROMIS: Patient-Reported Outcomes Measurement Information System; MMSE: Mini-Mental State Examination.

Frailty classification, sarcopenia and cognition

Body composition was assessed in 27 of 41 patients; 7 patients (26%) were found to be sarcopenic (1 frail, 5 pre-frail and 1 robust) (Table 3). No associations between sarcopenia and frailty classification were found in the univariate and multivariable analysis.

Only one patient was found to have an MMSE score <24. No associations were observed between MMSE score and frailty classification in the univariate or multivariable analysis.

Discussion

In this single-centre cross-sectional study of 41 patients with PMR, we report a 17% (95% CI 7.15, 32.06) prevalence of frailty and a 59% (95% CI 42.11, 73.68) prevalence of pre-frailty based on the Fried frailty definition. Frail PMR patients had higher CRP levels at the time of diagnosis and were more likely to have worse self-reported physical function, pain behaviour and pain interference in daily activities. No differences in the prevalence of sarcopenia or cognitive dysfunction, as measured by MMSE, were noted between frail, pre-frail and robust patients. To our knowledge, this is the first study to report prevalence of frailty in patients with PMR and its association with HRQoL measures. This novel focus of frailty in PMR addresses yet another potential gap our knowledge of long-term morbidity in patients with PMR and highlights the need for further longitudinal studies in this patient population [28]. As physical activity has been shown to mitigate and prevent frailty in older individuals, our findings support the efforts into developing and implementing early physical activity interventions to complement treatment of patients with PMR [29].

Reports of frailty prevalence in community-dwelling elderly vary depending on the population analysed (e.g. high vs low/middle income countries) and definition of frailty [20, 30, 31]. A recent systematic review of 15 studies which included 44 894 participants reported a weighted average prevalence of 9.9% using physical frailty definitions (i.e. Fried frailty phenotype). This same study calculated a weighted average prevalence of 44.2% for pre-frailty [30]. Although our sample size was relatively small, we found a higher prevalence for both pre-frailty and frailty than reported in the literature for community-dwelling older adults, despite low comorbidity burden and high educational attainment, both of which have been found to be associated with lower levels of frailty [20, 32, 33].

An association between frailty and inflammatory markers such as CRP and ESR has been described, including independent associations between CRP and outcomes such as mortality [4–6]. The association between frailty and elevated CRP levels has also been reported in patients with inflammatory rheumatic diseases such as RA [34, 35]. In patients with PMR, inflammatory markers tend to respond rapidly to treatment with glucocorticoids, and since active treatment was an inclusion criteria of our study, it is not surprising that no significant differences were observed in either CRP or ESR levels upon enrolment in the study (i.e. after treatment) in patients by frailty category. However, at time of diagnosis (i.e. prior to treatment), currently frail and pre-frail patients had higher median CRP levels compared with robust patients. A similar difference was observed for ESR levels. Elevation of CRP and other inflammatory markers can be observed with ageing, including ageing-related disorders such as frailty [36]. Particularly for CRP, these variations still remain within the normal range of the test values, therefore making older age in frail PMR patients a less likely explanation for the differences in CRP levels. Most importantly, elevated inflammatory markers have shown a stronger association with frailty than that with chronological age. These differences suggest that a subset of patients with PMR presenting with higher levels of inflammatory markers may be at increased risk of frailty. Interestingly, higher levels of inflammatory markers, such as ESR, at diagnosis have been associated with a higher risk of relapse in patients PMR [37]. Whether frailty plays a role in this association cannot be extrapolated given the cross-sectional nature of study, but it certainly represents an important hypothesis. Further study of the extent to which frailty is predicted by increased inflammation (as measured by routinely used inflammatory markers) at PMR presentation and the role of baseline frailty in longitudinal disease outcomes is needed.

In our study, all patients reported experiencing glucocorticoid-associated adverse effects. Among frailty groups, only a difference in the report of tremor and hyperglycemia was reported, but no difference in report of cumulative glucocorticoid-associated adverse effects was observed. Chronic glucocorticoid exposure is associated with an increased risk of cardiometabolic comorbidities, as well as osteoporosis and muscle atrophy [38]. However, the association between glucocorticoid exposure and frailty is still poorly understood. A recent study of older adults showed an increased risk of frailty, using a cumulative deficit (i.e. accumulation of age-related deficits), associated with the use of glucocorticoids utilizing a cumulative deficit definition of frailty [39]. However, no association was reported with frailty as defined by the Fried frailty phenotype. In patients with rheumatic diseases an association between frailty and glucocorticoid exposure has not been reported so far, and features such as disease activity or manifestations can further complicate the assessment of this association. However, given the well-established associated adverse effects, the concern for glucocorticoid exposure and risk of frailty exists and further work is needed to clarify this association.

In the original Fried study, inactivity followed by weakness and slowness were the most common criteria encountered [20]. In our study, weakness and exhaustion were the most frequent criteria, while slowness and inactivity were the least common criteria. Similarly, weakness and exhaustion have been found commonly in frail patients with RA and SLE, while slowness has been found infrequently [40, 41]. Although peripheral arthritis symptoms could affect the measure of handgrip strength, it is important to note that only a minority of patients (27%) in our study had peripheral arthritis symptoms, making hand arthritis symptoms less likely to influence the hand grip assessment. Differences in the frequency of these criteria when comparing patients with underlying inflammatory conditions could potentially highlight the differences in the process of frailty in these populations.

Multiple studies have shown that HRQoL is significantly worse in frail community dwelling older adults [42]. In our study, PROMIS scores for pain behaviour and pain interference were higher in frail patients with PMR while physical function was significantly worse compared with population norms. These scores were also clinically significant; however, these thresholds have not been validated in PMR. It is important to note that MIDs might differ in patient populations, with recent studies showing even lower MIDs in patients with conditions like RA and OA [43, 44]. In our study, frailty was associated with worse PROMIS pain behaviour and pain interference scores compared with robust patients, even after adjusting for age. Although non-significant, frail patients also reported worse physical function compared with pre-frail and robust patients despite lack of differences in disease activity between groups. A significant association with pain PROMIS measures (behaviour and interference) was observed in frail patients, and although we did not adjust for disease activity in our analysis, there were no differences in disease activity between different frailty groups. Similarly, a recent study of patients with RA showed an association between frailty and reduced physical function, independent of disease activity and duration, and also showed that fatigue was a primary contributor to this association [45]. Pain, together with stiffness, is one of the cardinal symptoms of PMR, and titration of therapy is usually guided by the presence of symptoms and therefore can influence treatment decision. Given the potential impact of physical function and pain in PMR, further studies are needed to explore these associations in relation to longitudinal disease outcomes (e.g. glucocorticoids treatment duration).

Increased fatigue, as measured by the Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue questionnaire, and an association with worse pain has been reported at time of diagnosis in patients with PMR [46]. Fatigue, as a reason for consultation, has even been reported more frequently in older patients prior to a diagnosis of PMR [47]. Although patients in our study have some important differences with regards to these previous studies (i.e. established diagnosis of PMR), the frequent report of exhaustion as part of the frailty criteria highlights the importance of this reported symptom among patients with PMR. Worse fatigue, as measured by PROMIS, was observed in frail patients with PMR, but this was not significantly or clinically meaningful. Even despite these discordances between our study and previous literature, which could be potentially explained by the differences in tools capturing exhaustion or limitations in sample size, our observations still highlight the frequency of fatigue in patients with PMR and highlight the need for better studies.

Sarcopenia has been associated with frailty and worse clinical outcomes in older adults [48, 49]. Although limited by a small sample size, we found a 26% prevalence of sarcopenia in our cohort of patients with PMR. Prevalence of sarcopenia in this cohort was higher than that reported in community-dwelling elderly using the same definition [50]. However, no associations between frailty and sarcopenia were observed and future studies should consider the alternative measurements of muscle mass such as skeletal muscle fat, or other measures of muscle function such as muscle power. No associations between MMSE score and frailty were observed in the univariate and multivariate analyses.

This study has several limitations. Recruitment was halted due to cessation of non-urgent in-person visits as a result of the COVID-19 pandemic. Participants were enrolled at a tertiary care rheumatology centre, potentially limiting the generalizability of our results to all patients with PMR. Only 66% of participants completed body composition assessment, limiting our ability to detect an association between frailty and sarcopenia. Further, consensus has not been reached in the sarcopenia community on a gold-standard definition of sarcopenia, and it is possible that use of an alternative definition of sarcopenia would have yielded different results. Finally, enrolment of comparator patients without PMR was beyond the scope of our study. Despite these limitations, our study has several strengths. This study is the first to assess the prevalence of frailty and its association with HRQoL in patients with PMR, for whom granular data were available on PMR diagnosis and treatment, as well as comorbidities and medications. We used a validated definition of PMR, as well as validated definitions of frailty and sarcopenia. Finally, we used validated HRQoL measures that have been developed at a population level and provide age and gender norms.

In conclusion, in this single-centre cross-sectional study of PMR patients we found a higher prevalence of frailty and pre-frailty compared with that reported in community-dwelling older adults. Significant differences in pre-treatment CRP levels were observed in patients based on frailty classification after 1 year of treatment, potentially identifying a subset of patients at higher risk of frailty. We observed an association between frailty and worse patient-reported physical function, pain behaviour and pain interference in daily activities. Although limited by a small sample size, these data suggest that frailty is common in PMR and that it is associated with clinically relevant characteristics. Larger studies are required to confirm these associations and to better understand the longitudinal impact of frailty on HRQoL and disease-specific outcomes such as relapse. Frailty may be an important feature of the PMR phenotype and is a potentially novel outcome measure which could be explored in both observational studies and interventional clinical trials.

Data availability statement

The data underlying this article will be shared on reasonable request to the corresponding author, after Institutional Review Board approval and completion of a data user agreement.

Supplementary data

Supplementary data are available at Rheumatology online.