Table 2

Assessment of methodological quality in diagnostic studies: summary of major biases identified

Study	Index test: imaging modality	Who performed index test, were they blinded to clinical data, was inter/intra-rater reliability reported?	Prospective study?	Does PMR spectrum appear realistic according to information given? Did any also have GCAs?	Consecutive selection of participants?	Comparator condition(s): realistic?	Reference standard; who performed it, when?	Did all participants receive all tests?	Free from incorporation bias?	Free from diagnostic review bias?	Did participants have index test before receiving glucocorticoid treatment?
Musculoskeletal ultrasound (MSK USS)
Dasgupta et al15*†	MSK USS shoulders, hips	Rheumatologist or radiologist, one per site; reliability reported separately (Scheel et al, 2009); at some sites sonographer was clinical assessor	Yes	Yes; none had GCA	No	Yes: >50 years, <12 weeks’ history of bilateral shoulder pain, not felt to be PMR	Clinical diagnosis; by investigator; after 6 months	5 PMR and 15 controls did not have scans	Yes—diagnosis made before USS, and assessors told not to use USS findings in making diagnosis	Sonographer and clinical assessor were sometimes same person	Yes
Ruta et al39*†	MSK USS shoulders	Single rheumatologist-sonographer blinded to clinical data; reliability not reported	Yes	Maybe: relapsing PMR (new-onset bilateral painful shoulder and prior diagnosis PMR); none had GCA	Yes	Maybe: relapsing RA (new-onset bilateral painful shoulder and prior diagnosis of RA)	PMR: clinical diagnosis+Healey criteria; RA: ACR 2010 criteria; by treating rheumatologist	Yes	Yes	Yes	No; were on ≤10 mg prednisolone; most were on 2–4 mg; treatment did not seem to affect USS findings
Falsetti et al44†	MSK USS at multiple sites	Single rheumatologist-sonographer, not blinded to clinical data; reliability not reported	Yes	Yes: all participants referred from primary care with polymyalgic syndrome fulfilling Bird criteria; one developed GCA later. All participants drawn from this same population (single-gate study design). 29/61 (47.5%) had final diagnosis PMR. Many of those with RA were seropositive			Clinical diagnosis; by 2 rheumatologists, after 1 year	Yes	No	No	Yes
Cantini et al34	MSK USS hips and MRI pelvic girdle	Two radiologists for each test (unclear whether these were same people), unclear whether blinded to clinical data (note alternating recruitment of cases/2 controls); reliability not reported	Yes	A subset: PMR with pelvic girdle involvement; 3 also had biopsy-proven GCA; none developed RA (1987 ACR criteria) after average follow-up 26 months	Yes	Maybe: next 2 consecutive outpatients >50 years with active rheumatic disease (RA/PsA/OA) and bilateral hip ache	Clinical diagnosis+Healey criteria PMR, followed up to ensure no evolution to RA	Only 10 of 40 controls had MRI (unclear how these were selected)	Yes	Unclear	Yes for PMR; unclear for controls
Frediani 2002†45	MSK USS at multiple sites	Two rheumatologist-sonographers, blinded to diagnosis; “medium rates concordance [agreement]” reported but no test statistics quoted	Yes	Yes: “PMR patients with a relatively certain diagnosis”—Healey criteria; 2 also had GCA	Yes	No: RA (ARA 1987 criteria); SpA (ESSG criteria)	Clinical diagnosis+Healey criteria PMR; 2-year follow-up to confirm diagnosis	Yes	Yes	No, but diagnosis not changed after USS	Yes
Cantini et al33	MSK USS shoulders	Two radiologists together, blinded to clinical diagnosis (but note recruitment of 2 controls after each case); reliability not reported	Yes	Yes: >1 month pain neck and shoulder girdle; morning stiffness> 1 h; ESR>40; 5 also had biopsy proven GCA; follow-up for mean 8 months to exclude those fulfilling 1987 ARA RA criteria	Yes	Maybe: next 2 consecutive outpatients >50 years with bilateral shoulder aching, stiffness (RA/PsA/SpA/OA/FM/CTD)	Clinical diagnosis+Healey criteria; by 1 of 4 rheumatologists; follow-up to confirm diagnosis	Yes	Yes	Unclear; but participant selection protocol implies participants did not switch between case/control groups	Yes for PMR, unclear for controls
Coari et al38	MSK USS shoulders	Two rheumatologist-sonographers, unclear whether blinded to clinical data; reliability not reported	Not stated but implied	No: treated PMR; not stated whether any had GCA	Not stated	No: treated; one-third of RA patients erosive	Clinical diagnosis (ARA 1987 for RA); not stated by whom or whether followed up	Only PMR each had both shoulders scanned; unit of analysis was shoulder not patient	Yes	Unclear	No
Lange et al46	MSK USS shoulders	Not stated; reliability not reported	Not stated but implied	Yes: >60 years, pain and several hours’ morning stiffness of shoulders, neck and/or pelvic girdle, limited motion in neck and shoulder, ESR>45, response to prednisolone 30 mg or less); 6 had headache, 2 had biopsy-proven GCA	Not stated	Maybe: “initially had similar complaints (to the PMR cases) … involvement of arthritis in additional joints and bony erosions”	Clinical diagnosis; not stated by whom or whether followed up	Yes	Yes (implied but not stated)	Unclear	Yes (implied but not stated)
Lange et al47	MSK USS shoulders	Not stated; reliability not reported	Not stated but implied	Yes: >60 years, pain and several hours’ morning stiffness of shoulders, neck and/or pelvic girdle, >4 weeks duration symptoms, ESR>45, response to prednisolone 30 mg or less); 5 had headache, 4 had biopsy-proven GCA	Not stated	Maybe: “initially had similar complaints (to the PMR cases) … involvement of arthritis in additional joints and bony erosions”	Clinical diagnosis; not stated by whom or whether followed up	Yes	Yes (implied but not stated)	Unclear	Yes (implied but not stated)
Macchioni et al23	MSK USS shoulders, hips	Single rheumatologist-sonographer; blinding to clinical data not stated; reliability not reported	No	Yes: patients seen with suspected PMR; patients with GCA excluded	Yes	No: patients in early arthritis clinic; no requirement for comparable symptoms	Clinical diagnosis; confirmed at 1 year by 2 lead authors	Yes	Unclear	No	Yes
MRI
Salvarani et al37	1.5 T MRI lumbar spine (bursitis)	Radiologist; blinded to clinical findings and diagnosis; reliability not reported	Yes	A subset: PMR by Chuang criteria+pelvic girdle symptoms; none had GCA	Yes	Maybe: treated patients with lumbar pain (SpA/OA/RA)	Clinical diagnosis+Chuang criteria, followed up for 10–16 months to exclude RA (ARA 1987) or other conditions	Yes	Yes	Yes	Yes for PMR, unclear for controls
Cimmino et al40	0.2 T MRI hands (extremity MRI)—tenosynovitis	Two rheumatologists and one PhD, blinded to diagnosis; reliability not reported but Parodi et al 2006 quoted in support	Yes	Yes: PMR by Chuang criteria; none had GCA	Yes for PMR, not for controls	No: Healthy controls of similar ages, no mention of symptoms	Clinical diagnosis+Chuang criteria, followed for 8–124 months to exclude GCA, RA and other erosive disease	Yes but 4 hands could not be interpreted	Yes	Yes	Yes
Salvarani et al36	1 T MRI cervical spine (bursitis)	One radiologist, blinded to clinical data and diagnosis (but note alternating recruitment of cases, controls); reliability not reported	Yes	Yes: PMR (reference Salvarani review 2002); none had GCA	Yes	No: Next patients with neck pain seen after PMR patients	Clinical diagnosis+criteria; followed for 10–16 months to exclude other conditions	Yes	Yes	Yes	Yes
Marzo et al13	1.5 T MRI of most swollen hand	One assessor per MRI feature, blinded to clinical data; reliability not reported	Yes	No: Bird criteria+MCP joint swelling	Yes for RA, not stated for PMR	No: ARA 1987 criteria+MCP joint swelling	Clinical diagnosis+Bird criteria; followed for mean of 6 years	Yes	Yes	Yes	Yes except for one PMR patient
McGonagle et al12	1.5 T MRI shoulder	Two radiologists, blinded to clinical data; reliability not reported	Yes	Yes: untreated PMR and bilateral shoulder disease without peripheral arthropathy	No	No: early RA fulfilling 1987 ARA criteria	Clinical diagnosis; no follow-up reported to exclude other conditions	Only 6/14 PMR patients had both shoulders imaged	Yes	Yes	Yes for PMR; not for 8/14 RA
Salvarani et al48	0.5 T MRI shoulder	One radiologist, blinded to clinical data and diagnosis; reliability not reported	Yes	Yes: Healey criteria PMR; none had GCA	Unclear	No: elderly-onset RA by modified 1987 ARA criteria, with clinical evidence shoulder involvement	Clinical diagnosis+Healey criteria; no follow-up reported to exclude other conditions	The first 4 PMR had both shoulders imaged; after that only one shoulder	Yes	Yes	Yes
¹⁸F-fluorodeoxyglucose—positron emission tomography (FDG-PET)
Yamashita et al35	FDG-PET/CT whole body	Not stated who reported test; unclear whether blinded to clinical info; reliability not reported	No	No: inpatients, having PET/CT to exclude other diseases for example, suspected malignancy; none had clinical evidence GCA	Yes	No (other rheumatic diseases with suspected malignancy; 11/17 RA)	Clinical diagnosis+Chuang+Healey criteria; length of follow-up not specified	Yes	Unclear	No	Yes for PMR, not stated for controls
Camellino et al25†	FDG-PET/CT	Rheumatologist and radiologist, blinded to clinical data (pers comm); reliability not reported	Yes	Little information on how patients were identified	Yes	No (65 matched controls with no inflammatory disease; 10 with treated RA)	Fulfilled Bird and ACR/EULAR criteria; median follow-up 22 months	Yes	Yes	Probably	Yes for PMR/controls, no for RA
Takahasi et al24†	FDG-PET/CT	Radiologists, blinded to clinical data [pers comm]; reliability not reported	No	No: inpatients and outpatients, having PET/CT to exclude other diseases, for example, suspected malignancy; none had clinical evidence of GCA	Yes	Maybe (untreated, elderly-onset RA)	Diagnosed by attending doctors prior to PET/CT (pers comm); diagnosis did not change on follow-up (pers comm). and verified by classification criteria	Yes	Yes	Yes	Yes

The PET or PET/CT studies that did not report data extractable into 2×2 table format are not listed here. Before-after or prognostic studies, if they did not report data extractable into 2×2 table format, are not reported here.
Incorporation bias means where the imaging (index test) informs the diagnosis (reference standard).
Diagnostic review bias means where the diagnosis (reference standard) was carried out or verified with knowledge of the imaging (index test).
*Further data were supplied by corresponding authors on request.
†Methodological details supplied by corresponding authors on request.
ACR, American College of Rheumatology; ARA, American Rheumatism Association; CTD, connective tissue disease; ESR, erythrocyte sedimentation rate; EULAR, the European League Against Rheumatism; FM, fibromyalgia; GCA, giant cell arteritis; MCP, metacarpophalangeal; OA, osteoarthritis; PET/CT, positron emission tomography CT; PMR, polymyalgia rheumatica; PsA, psoriatic arthritis; RA, rheumatoid arthritis; SpA, spondyloarthropathies.