Ann Rheum Dis 1996;55:907-914

Shoulder capsulitis in type I and II diabeticpatients: association with diabetic complicationsand related diseases

Perttu E T Arkkila, Ilkka M Kantola, Jorma S A Viikari, Tapani Ronnemaa

AbstractObjective-To examine the associationbetween shoulder capsulitis and chronicdiabetic complications and diseasesclosely related to diabetes.Methods-A cross sectional study in 291type I [mean (SD) age 33.2 (9.9) years] and134 type II [61.1 (12.4) years] diabeticpatients. The presence of shoulder capsu-litis, Dupuytren disease, and limited jointmobility was sought. The patients wereassessed for background and proliferativeretinopathy, nephropathy, autonomic neu-ropathy, and peripheral symmetricalsomatic polyneuropathy. Diseases closelyrelated to diabetes (hypertension, historyof myocardial infarction, coronary heartdisease, and peripheral vascular disease)were also recorded.Results-Prevalence of shoulder capsuli-tis was 10.3% in type I and 22.4% in type IIdiabetic subjects. Shoulder capsulitis wasassociated with the age in types I (P < 0.01)and II (P < 0.05) diabetic patients, andwith the duration of diabetes in type Ipatients (P < 0.01). Odds ratios forautonomic neuropathy in type I and typeII diabetic subjects with shoulder capsuli-tis were 4.1 (95% confidence interval, 1.6to 10.9) and 2.7 (95% CI, 1.1 to 7.0),respectively, after controlling for age andduration of diabetes. Odds ratio forhistory of myocardial infarction in type Idiabetic subjects with shoulder capsulitiswas 13.7 (95% CI, 1.3 to 139.5) aftercontrolling for age, duration of diabetes,hypertension, and smoking habits. Otherassociations between shoulder capsulitisand diabetic complications, relateddiseases, and other hand abnormalitieswere fully explained by age and theduration of diabetes.Conclusions-Shoulder capsulitis is com-mon in type I and type H diabetic patients.It is associated with age in type I and IIdiabetic patients and with the duration ofdiabetes in type I patients. It is associatedwith autonomic neuropathy in type I andII diabetic patients and with history ofmyocardial infarction in type I diabeticpatients, independently of time relatedvariables.

(Ann Rheum Dis 1996;55:907-914)

In 1872 Duplay described a condition,"peri-arthrite scapulo-humerale", which dif-

fered from arthritis in its symptoms andclinical course.' The true nature of this condi-tion has been discussed ever since. Duplayhimself believed that a subacromial bursitiswas the basic cause of the pain anddysfunction. By 1916 Klapp and Riedelbelieved that the joint capsule was affected,and later, in 1931, Payr tried to reduce disten-sion of the retracted capsular tissue byintra-articular injections (citation by Lund-berg'). In 1934 Codman showed that stiffnessand pain in the shoulder could occur withoutnoticeable exogenous influences and hetermed the condition "frozen shoulder", whichhas later been used as a synonym for humero-scapular periarthritis and adhesive capsulitis ofshoulder (shoulder capsulitis).' Codmanbelieved that shoulder capsulitis was caused bytendinitis in the short rotators, but morerecently the thickening of the joint capsule andits adherence to the head of the humerusresulting in marked reduction in the volume ofthe glenohumeral joint have been consideredas the basic pathological change in shouldercapsulitis.4The aetiology of shoulder capsulitis is not

clearly understood. Attempts have been madeto relate it to various circumstances such asinactivity, strain, and pre-existing shoulder dis-orders, for example trauma.5 Several condi-tions have been associated with shouldercapsulitis; these include cervical spondylosis,6coronary heart disease,78 hemiplegia,9 pulmo-nary tuberculosis,'0 bronchial carcinoma,"hyperthyroidism,," cerebral tumour, andepilepsy. 'MThe relation between diabetes mel-litus and shoulder capsulitis has been shown infew previous studies. Bridgman reviewed themedical records of 800 diabetic subjects andfound evidence of periarthritis in 10.8%, com-pared with 2.3% in a control group of 600non-diabetic subjects.'4 Pal et al foundshoulder capsulitis in 20.4% of insulin-dependent and in 18.3% of non-insulin-dependent diabetes patients and in 5.3% ofnormal subjects.'5 In a study of 824 type II dia-betic and 320 control subjects shoulder capsu-litis was observed in 31.8% and 10.3% of sub-jects, respectively.'6

Bilateral involvement of shoulder capsulitiswas more common in diabetic (10%) than inthe control subjects (3%).16 Diabetic shouldercapsulitis seems to appear at a younger age,may be less painful, responds less well to treat-ment, and lasts longer than non-diabeticshoulder capsulitis."'7- A high frequency ofother hand syndromes, such as limited joint

Department ofMedicine, TurkuUniversity CentralHospital, Turku,FinlandP E T ArkkilaI M KantolaJ S A ViikariT Ronnemaa

Correspondence to:Perttu E T Arkkila MD,Department of Medicine,Turku University, CentralHospital, Turku, Finland.

Accepted for publication4 September 1996

907

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Arkkila, Kantola, Viikari, Ronnemaa

mobility, has been found among diabeticpatients with shoulder capsulitis."8l9 Theassociation between limited joint mobility andmicrovascular complications of diabetes hasbeen well documented, but there are also twostudies showing an association between shoul-der capsulitis and retinopathy.'9-22 However, noassociation between shoulder capsulitis anddiabetic neuropathy has been found.'4The purpose of this study was to investigate

the prevalence of shoulder capsulitis and itsassociation to the diabetic complications intype I and II diabetic subjects.

MethodsPATIENTSWe evaluated 425 Finnish diabetic patients(200 men and 225 women) attending thedepartments of medicine in Turku UniversityCentral Hospital (n = 111) and Paijat-HameCentral Hospital (n = 57) in Lahti, in a privatediabetes outpatient clinic in Turku (n = 217),and in the municipal health centre ofTurku (n= 40). Of these, 291 had type I and 134 type IIdiabetes according to the classification of theNational Diabetes Data Group.23 Patients wereconsidered to have type II diabetes if there wasno previous history of ketoacidosis and theyhad not suffered from severe weight loss. Allthe insulin treated patients in the type IIdiabetes group were over 50 years at the time ofthe diagnosis of diabetes, or their fastingplasma C peptide value was over 0.3 nmollitre-', or the stimulated plasma C peptide valuewas over 0.7 nmol litre'1.

CLINICAL EXAMINATIONThe criteria for shoulder capsulitis, previouslydescribed by Pal et al, were shoulder pain for atleast one month, an inability to lie on theaffected shoulder, and restricted active andpassive shoulder joint movements in at leastthree planes.'5 A reliably documented record ofthe patient having had these symptoms andsigns was accepted as evidence of a shouldercapsulitis in the past. The diagnosis ofDupuytren disease was made by the methodearlier used by Noble et al.'4 Limited jointmobility was assessed by the method of Rosen-bloom et al."0

All subjects gave a detailed history. Specialemphasis was placed on age and the durationof diabetes. Weight and height were measuredand the body mass index (BMI) was calculatedfrom the formula BMI = weight (kg)/[height(m)]'. The occupation of subjects (manual orintellectual work) was recorded, as was the his-tory of injuries, operations, or infections in theupper extremities.

Retinal examination was performed by expe-rienced ophthalmologists, with fundoscopyafter dilatation of the pupils with a mydriaticagent. Retinopathy was classified as back-ground or proliferative. Background retin-opathy was characterised by microaneurysms,hard exudates, cotton wool or soft exudates,and retinal haemorrhages. Proliferative retin-opathy was characterised by neovascularisa-

tion, fibrous proliferations, vitreous haemor-rhage obscuring the retina, phthisis bulbi, orenucleation secondary to a complication ofdiabetic retinopathy.The symptoms of somatic peripheral

symmetrical polyneuropathy included pain,paraesthesiae, and muscle weakness. Knee andankle jerks and vibration sense over the lateralmalleoli and over the styloid processes of radii(using a 256 Hz tuning fork) were tested. Thepatients were considered to have peripheralsymmetrical somatic polyneuropathy if theyhad typical symptoms and impaired vibrationsense, or if both ankle jerks were absent. Thediagnosis of autonomic neuropathy was basedon structured questions regarding anhidrosisor hyperhidrosis of the extremities or body, or agustatory sweating disorder. Heart ratevariation was assessed in 121 subjects by meas-uring the standard deviation ofR-R intervals inthe supine position'5 and by calculating the dif-ference between the maximum and minimumheart rates during deep breathing, as describedelsewhere.'6 The diagnosis of posturalhypotension was based on a structuredquestion regarding symptoms of hypotensionon stand up.Blood pressure was measured sitting after a

period of rest of at least 10 minutes using astandard mercury sphygmomanometer. Thereading when the fifth Korotkoff sound disap-peared was taken as the diastolic bloodpressure. Subjects were defined as havinghypertension if they used antihypertensivedrugs for hypertension or had a systolic bloodpressure of . 160 mm Hg or a diastolic bloodpressure of 2 95 mm Hg.The subjects' smoking history was recorded.

An ex-smoker was defined as a person havingsmoked regularly at any time in the past.A history of myocardial infarction was based

on symptoms of chest pain, electrocardiogram(ECG) abnormalities, and enzyme determina-tions. The criteria for coronary heart diseasewere a history of definite myocardial infarctionor ischaemic ECG abnormalities duringangina pectoris type chest pain at rest orduring an exercise test. Peripheral vascular dis-ease was present if one or two tibialis posteriorpulses were absent.

BIOCHEMICAL METHODSThe metabolic control of diabetes wasevaluated by the annual average glycatedhaemoglobin Alc (GHbA1I) over the previousfive years. GHbAlc was determined by fastprotein liquid chromatography (Pharmacia,Uppsala, Sweden). The endogenous insulinsecretion capacity was assessed by plasma Cpeptide measurement (fasting or stimulated).Plasma C peptide was assessed either by radio-immunoassay [antisera MNo 299-029P, Cam-bridge Medical Diagnostics, Billerica, MA,USA), the tracer was human Tyr-C-peptidelabelled with "'I (Novo, Bagsvaerd, Denmark),with C peptide standard (Novo, catalogue No820)] or by the Byk-Sangtec radio-immunoassay-coat C peptide method (stand-ardised to WHO 84/150). Twelve hour or sixhour overnight urine samples were collected

908

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Shoulder capsulitis in type I and II diabetic patients

Table 1 Clinical characteristics of type I diabetic patients with and without shouldercapsulitis

Shoulder capsulitus

Present (n=30; Absent (n=261; Total (n=291;Study group 10.3%) 89.7%) 100%)

Age (years)Mean (SD) 46.7 (8.2)* 31.7 (8.9) 33.2 (9.9)Range 33-61 16-67 16-67

SexMen (% of total) 40.0 48.7 47.8

OccupationManual work (% of total) 51.9 45.2 45.9

BMI (kg m2)Mean (SD) 24.7 (3.2) 24.0 (3.0) 24.1 (3.1)Range 20.0-31.7 17.3-35.3 17.3-35.3

Duration of diabetes (years)Mean (SD) 28.7 (8.2)* 17.1 (9.1) 18.3 (9.7)Range 11-48 1-47 1-48

GHbA1, (%) (5 years)Mean (SD) 7.9 (1.5) 8.4 (1.5) 8.4 (1.5)Range 3.1-10.2 5.0-12.9 3.1-12.9

Non-smokers(% of total) 63.3 66.5 66.2

Ex-smokers(% of total) 20.0* 5.4 6.9

Smokers(% oftotal) 16.7 28.1 26.9

BMI, body mass index; GHbA,C, glycated haemoglobin. * P < 0.01 compared to the patientswithout shoulder capsulitis (ANOVA).

Table 2 Clinical characteristics of type II diabetic patients with and without shoulder.capsulitis

Shoulder capsulitis

Present (n=30; Absent (n=104; Total (n=134;Study group 22.4%) 77.6%) 100%)

Age (years)Mean (SD) 65.3 (10.0)* 59.9 (12.8) 61.1 (12.4)Range 44-81 38-84 38-84

SexMen (% of total) 50.0 44.2 45.5

OccupationManual work (% of total) 69.2 58.0 60.5

BMI (kg m2)Mean (SD) 27.3 (4.7) 27.3 (4.6) 27.3 (4.6)Range 19.1-39.8 15.6-43.3 15.6-43.3

Duration of diabetes (years)Mean (SD) 12.6 (6.1) 10.7 (7.2) 11.2 (7.0)Range 2-26 1-32 1-32

GHbAlC (%) (5 years)Mean (SD) 9.1 (2.0) 8.5 (1.7) 8.6 (1.8)Range 5.7-13.9 5.1-13.9 5.1-13.9

Non-smokers(% of total) 53.3 73.8 69.2

Ex-smokers(% oftotal) 10.0 13.6 12.8

Smokers(% of total) 36.7t 12.6 18.0

BMI, body mass index; GHbAlc, glycated haemoglobin. * P < 0.05 and tP < 0.01 compared tothe patients without shoulder capsulitis (ANOVA).

from 256 subjects, each of whom wasinstructed in writing to avoid exercise duringthe collection-only light indoor walking wasallowed. The urine volumes were measuredand the albumin concentrations were analysedafter centrifugation from dipstick-negativesamples by nephelometry (Behring nephelom-eter analyser; Behring Marburg, Germany).25Microalbuminuria and macroalbuminuria wereconsidered to be present when the urinary albu-min excretion rate (UAE) exceeded 20 but wasless than 201 jig minm' in at least two ofthree con-secutive samples and when the samples weredipstick-positive, respectively. Before establishingthe presence ofnephropathy associated micro- ormacroalbuminuria, other causes of albuminuriasuch as physical exercise and concurrent urinarytract infection were carefully excluded.

STATISTICAL METHODSFor statistics, Pearson's X2 test, one way analysisof variance (ANOVA), multinominal poly-chotomous, and stepwise logistic regressionanalyses were used.29

ResultsPREVALENCE OF SHOULDER CAPSULMSThe overall cumulative prevalence of shouldercapsulitis was 14% (13% in men and 15% inwomen). The prevalence was 10% in type Iand 22% in type II diabetic patients (P < 0.01).The mean age was 46.7 (SD 8.2) years in typeI diabetic subjects with shoulder capsulitis and31.7 (8.9) years in subjects without shouldercapsulitis (P < 0.01) (table 1). A significantassociation between shoulder capsulitis and theage was also found in type II diabetic patients[65.3 (10.0) v 59.9 (12.8) years] (P < 0.05)(table 2). Type I diabetic subjects with andwithout shoulder capsulitis had had diabetesfor 28.7 (8.2) and 17.1 (9.1) years, respectively(P < 0.01) (table 1). The duration of diabeteswas not significantly different in type IIdiabetic subjects with and without shouldercapsulitis, at 12.6 (6.1) v 10.7 (7.2) years(table 2). Shoulder capsulitis was notassociated with subjects' height, weight, BMI,or occupation (tables 1 and 2). A surprisingassociation between ex-smoking and shouldercapsulitis was found in type I diabetic subjects(P < 0.01), whereas shoulder capsulitis wasassociated with present smoking in type II dia-betic subjects (P < 0.01) (tables 1 and 2). Theassociation between shoulder capsulitis andex-smoking in type I diabetic subjects was fullyexplained by the fact that ex-smokers wereolder [40.3 (11.6) years] than non-smokers[32.5 (9.7) years] or present smokers [33.4(9.7) years]. Present smoking was notassociated with patients' age, duration ofdiabetes, or control of diabetes in type IIdiabetic subjects.

GLYCAEMIA CONTROLThere was no significant difference in meanlevels ofGHbAlc over the previous five years intype I or II diabetic patients with or withoutshoulder capsulitis (tables 1 and 2). Type IIdiabetic patients with poor control of diabetes(GHbAlc > 9%) seemed to have moreshoulder capsulitis than patients with bettercontrol of diabetes (figure), but this associationwas not significant.

MICROVASCULAR COMPLICATIONSThe proportions of shoulder capsulitis were4%, 10%, and 21% in type I patients withoutretinopathy, with background and withproliferative retinopathy, respectively (table 3).When a X2 test was used, shoulder capsulitiswas associated with proliferative retinopathy (P< 0.01) in type I diabetic subjects (table 3),whereas no significant association was found intype II diabetic patients (P = 0.55) (table 4).No association was found between shoulder

909

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Arkkila, Kantola, Viikari, Ronnemaa

30 r

25 H

20 H

15

1i0

0

Proportion of dshoulder capsu(mean GHbA

capsulitis ai

type I or IIBecause

ciated withage in typepatients (tathat shouldduration ar

diabetes 1

retinopathythe associastudied usi:No significcapsulitis ,

retinopathy

the confounding effects of the duration ofdiabetes and age were controlled for.Type

Type 11

PERIPHERAL SYMMETRICAL SOMATICPOLYNEUROPATHY AND AUTONOMIC NEUROPATHYThe prevalence of peripheral symmetricalsomatic polyneuropathy (neuropathy) was44% in type I and 56% in type II diabetic sub-jects. Three per cent of type I diabetic subjectswithout neuropathy and 20% with neuropathyhad shoulder capsulitis (P < 0.01) (table 3).No significant association remained when theconfounding effect of the duration of diabetes

<7 7-8 8-9 >9 and age were controlled for by logisticGHbA1 ( regression analysis. No association between

shoulder capsulitis and neuropathy was foundiiabetic patients (%) (type I and II) with in type II diabetic subjects (table 4).litis according to the control of diabetes Nineteen per cent and 25% of type I and IIlc during the last five years) (0/6).

diabetic patients, respectively, had autonomicnd micro- or macroalbuminuria in neuropathy. Twenty five per cent of type I anddiabetic subjects (tables 3 and 4). 35% of type II diabetic subjects withshoulder capsulitis is strongly asso- autonomic neuropathy had shoulder capsulitisthe duration of diabetes and with compared to 7% and 18% without autonomicI and with age in type II diabetic neuropathy, respectively. The odds ratios for

ibles 1 and 2), it is to be expected autonomic neuropathy were 4.1 (95%ler capsulitis is correlated with the confidence interval, 1.6 to 10.9) and 2.7 (95%id age dependent complication of CI, 1.1 to 7.0) in type I and type II diabeticsuch as retinopathy. Because subjects with shoulder capsulitis, respectively,was divided into three categories, compared to subjects without shoulder capsu-

ition with shoulder capsulitis was litis. If the diagnosis of autonomic neuropathyng a multinominal logistic model. was based on hyperhidrosis, anhidrosis, or pos-:ant correlation between shoulder tural hypotension, the odds ratio for autonomicand background or proliferative neuropathy in type I diabetic subjects within type I patients was found when shoulder capsulitis was 3.6 (95% CI, 1.6 to

Table 3 Diabetic complications and related diseases according to age, the duration of diabetes, the control of diabetes(mean GHbAIc during past 5 years) and the presence of shoulder capsulitis (SC) in type I diabetic patients

Mean duration o Prevalence ofSC, %Mean age (years (SD)) diabetes (years (SD GHbA,c (0/6) (number ofprevalent~~years~~~~~~ cases/all cases)

RetinopathyNormal 28.3 (8.5) 10.7 (7.7) 8.1 (1.7) 4.1 (4/98)Background 34.7 (9.5)t 21.0 (7.7)t 8.4 (1.5) 9.9 (13/131)Proliferative 38.9 (9.5)t 26.0 (7.0)t 8.6 (1.4) 21.1 (12/57)*

MicroalbuminuriaNo 32.4 (9.2) 16.9 (9.4) 8.2 (1.4) 8.6 (14/163)Yes 33.2 (11.0) 20.2 (10.1)* 8.8 (1.8)* 11.4 (5/44)

MacroalbuminuriaNo 32.0 (9.5) 16.3 (9.5) 8.2 (1.5) 8.5 (17/201)Yes 35.8 (10.0)t 22.3 (8.5)t 8.8 (1.6)t 14.5 (12/83)

Peripheral symmetrical somaticpolyneuropathyNo 29.0 (7.3) 13.9 (8.1) 8.2 (1.6) 3.1 (5/163)Yes 38.6 (10.3)t 23.9 (8.5)t 8.6 (1.4)* 19.5 (25/128)t

Autonomic neuropathyNo 32.1 (9.3) 17.0 (9.1) 8.3 (1.6) 6.8 (16/234)Yes 37.8 (11.6)t 23.6 (10.2)t 8.5 (1.5) 25.5 (14/55)t

HypertensionNo 31.5 (9.4) 16.4 (9.4) 8.3 (1.6) 6.3 (14/221)Yes 36.0 (8.4)* 23.5 (6.7)t 8.8 (1.3) 21.7 (15/69)t

History of myocardial infarctionNo 32.8 (9.5) 17.9 (9.4) 8.4 (1.6) 8.8 (25/285)Yes 54.5 (7.5)t 34.5 (9.6)t 8.5 (0.6) 83.3 (5/6)t

Coronary heart diseaseNo 32.3 (9.2) 17.6 (9.2) 8.4 (1.6) 8.1 (22/270)Yes 44.9 (12.1)t 28.0 (10.3)t 8.4 (0.8) 35.0 (7/20)t

Peripheral vascular diseaseNo 31.8 (9.7) 17.0 (9.3) 8.4 (1.6) 7.4 (18/244)Yes 43.1 (10.0)t 27.8 (7.8)t 8.3 (1.6) 28.9 (11/38)t

Limited joint mobilityNo 30.7 (8.1) 13.9 (8.9) 8.4 (1.7) 8.5 (10/117)Yes 35.0 (10.7)t 21.2 (9.1)t 8.3 (1.5) 11.5 (20/174)

Dupuytren diseaseNo 31.8 (9.2) 16.8 (9.1) 8.4 (1.6) 7.6 (19/250)Yes 41.9 (10.1)t 27.1 (8.7)t 8.4 (1.5) 25.0 (10/40)4

t P < 0.01, v diabetic patients witiout complication or related disease (X' test).* P < 0.05, v diabetic patients without complication or related disease (ANOVA).t P < 0.01, v diabetic patients without complication or related disease (ANOVA).

2C.a)a-

910

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Shoulder capsulitis in type Iand II diabetic patients

Table 4 Diabetic complications and related diseases according to age, the duration of diabetes, the control of diabetes(mean GHbA,C during past 5 years) and the presence ofshoulder capsulitis (SC) in type II diabetic patients

Meanage(years(SD))Mean duration of Prevalence ofSC, %Mean age (years (SD)) diabetes (years (SD)) GHbA c (%) (number ofprevalentcases/all cases)

RetinopathyNormal 61.7 (11.8) 9.9 (6.8) 8.6 (1.9) 16.3 (13/80)Background 64.3 (9.3) 13.5 (6.1)* 8.3 (1.7) 34.4 (11/32)Proliferative 71.6 (8.1)t 17.5 (6.0)t 9.3 (1.4) 35.3 (6/17)

MicroalbuminuriaNo 55.3 (11.3) 10.8 (6.8) 8.4 (1.6) 15.4 (6/39)Yes 50.3 (9.6) 8.9 (7.9) 9.2 (1.8) 25.0 (2/8)

MacroalbuminuriaNo 62.8 (11.7) 11.3 (7.1) 8.4 (1.6) 20.8 (22/106)Yes 67.4 (7.5) 12.7 (7.1) 9.6 (2.1)* 28.6 (6/21)

Peripheral symmetrical somaticpolyneuropathyNo 58.5 (11.8) 8.7 (6.6) 8.3 (1.6) 18.6 (11/59)Yes 66.8 (9.9)t 13.5 (6.9)t 8.8 (1.9) 25.3 (19/75)

Autonomic neuropathyNo 63.7 (12.0) 11.6 (7.4) 8.5 (1.6) 18.3 (17/93)Yes 62.5 (9.3) 11.9 (6.6) 8.9 (2.1) 35.5 (11/31)4

HypertensionNo 61.5 (11.7) 11.7 (7.2) 8.7 (1.4) 25.8 (17/66)Yes 65.1 (11.0) 11.6 (7.2) 8.6 (2.0) 19.1 (13/68)

History of myocardial infarctionNo 61.8 (11.6) 11.2 (6.9) 8.4 (1.6) 18.5 (20/108)Yes 69.1 (8.6)t 13.0 (8.0) 9.5 (2.1)t 38.5 (10/26)4

Coronary heart diseaseNo 58.2 (11.2) 10.0 (7.2) 8.4 (1.5) 20.3 (16/79)Yes 69.3 (8.3)t 12.9 (6.9) 8.9 (2.0) 25.5 (14/55)

Peripheral vascular diseaseNo 62.1 (12.1) 11.2 (7.2) 8.4 (1.6) 20.6 (20/97)Yes 67.1 (8.5)* 12.9 (7.0) 9.2 (2.0)* 26.5 (9/34)

Limited joint mobilityNo 58.1 (12.5) 8.9 (6.3) 8.2 (1.2) 13.2 (7/53)Yes 66.7 (9.4)t 13.2 (7.2)t 8.9 (2.0)* 28.4 (23/81)t

Dupuytren diseaseNo 63.1 (12.0) 11.3 (7.4) 8.6 (1.8) 20.0 (23/115)Yes 65.3 (6.7) 13.1 (5.4) 8.6 (1.4) 36.8 (7/19)

t P < 0.05, v diabetic patients without complication or related disease (X2 test).* P < 0.05, v diabetic patients without complication or related disease (ANOVA).t P < 0.01, v diabetic patients without complication or related disease (ANOVA).

8.1) compared to the subjects withoutshoulder capsulitis, but no association betweenautonomic neuropathy and type II diabeteswas found. The confounding effects of theduration of diabetes and age were controlledfor by logistic regression analysis. In a subsam-ple of 121 type 1 diabetic subjects, shouldercapsulitis was associated with impaired heartrate variation (P < 0.01), but this associationdisappeared when the age and the duration ofdiabetes were controlled for.

HYPERTENSION, HISTORY OF MYOCARDIALINFARCTION, CORONARY HEART DISEASE, ANDPERIPHERAL VASCULAR DISEASEThe prevalence of hypertension was 24% and51% in type I and II diabetic subjects, respec-tively. Twenty two per cent of type I diabeticsubjects with hypertension and 6% of patientswithout hypertension had shoulder capsulitis(P < 0.01) (table 3). No association betweenshoulder capsulitis and hypertension remainedwhen the confounding effect of the age wascontrolled for. No association betweenshoulder capsulitis and hypertension was seenin type II diabetic patients (table 4).The prevalence of a history of myocardial

infarction was 2% and 19% and of coronaryheart disease, 7% and 41% in type I and II dia-betic subjects, respectively. Both type I andtype II diabetic subjects with myocardialinfarction had shoulder capsulitis more often

than those without myocardial infarction,whereas an association between coronary heartdisease and shoulder capsulitis was found onlyin type I diabetic subjects (tables 3 and 4).Because myocardial infarction was stronglyassociated with the duration of diabetes andage, we used stepwise logistic regression analy-sis to control the confounding effect of thesetime related variables. Smoking habits andhypertension were also included in the model.The odds ratio for myocardial infarction intype I diabetic subjects was 13.7 (95% CI, 1.3to 139.5) when the confounding effects of theage, the duration of diabetes, hypertension,and smoking habits were controlled. Thecorrelation remained significant when serumtotal cholesterol was included in the model. Intype II diabetic subjects no correlation wasfound between shoulder capsulitis andmyocardial infarction when the time relatedvariables were controlled for. No associationbetween shoulder capsulitis and coronary heartdisease was found when the confounding effectsof the duration of diabetes or the age werecontrolled for in type I or II diabetic subjects.The prevalence of peripheral vascular

disease was 13% and 26% in type I and II dia-betic patients, respectively. Type I diabeticpatients with peripheral vascular disease hadshoulder capsulitis more often than those with-out peripheral vascular disease (P < 0.01)(table 3), but this association was fullyexplained by the age and the duration of

911

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Arkkila, Kantola, Viikari, Ronnemaa

diabetes. No association was found betweenperipheral vascular disease and shouldercapsulitis in type II diabetic subjects (table 4).

LIMITED JOINT MOBILITY AND DUPUYTRENDISEASEThe prevalence of limited joint mobility was60% in type I and II diabetic patients. Noassociation between limited joint mobility andshoulder capsulitis was found in type I diabeticsubjects (table 3). Type II diabetic subjectswith limited joint mobility had shoulder capsu-litis more often than those without (table 4),but this association disappeared when the con-founding effects of the duration of diabetes andthe age were controlled for.The prevalence of Dupuytren disease was

14% in both type I and II diabetic subjects.Twenty five per cent of type I diabetic subjectswith Dupuytren disease had shouldercapsulitis compared to 8% without Dupuytrendisease (P < 0.01). This associationdisappeared when the confounding effects oftime related variables were controlled for. Nosignificant association between Dupuytren dis-ease and shoulder capsulitis was found in typeII diabetic subjects (table 4).

DiscussionThis report shows that shoulder capsulitis is acommon disorder in both type I and type IIdiabetic subjects, which is in line with previousstudies.'4 18 21 The presence of shoulder capsuli-tis was highly dependent on the age and theduration of diabetes in type I diabetic subjects,whereas age was the most important factorexplaining shoulder capsulitis in type IIdiabetic subjects. The prevalence of shouldercapsulitis increased after the age of 40 and 50years in type I and II diabetic patients, respec-tively. The prevalence of shoulder capsulitis didnot increase until after the duration of diabeteshad exceeded 20 years in type I diabeticpatients. The reason for the high prevalence ofshoulder capsulitis after a short duration ofdiabetes in type II patients may be explained bythe fact that the known duration of diabetes isprobably an inaccurate marker of the trueduration of the disease in many patients withtype II diabetes. Two previous studies haveshown an association between shoulder capsu-litis and the duration of diabetes,"62' but thesestudies showed no association betweenshoulder capsulitis and the age of diabetic sub-jects. Sattar and Luqman have shown nodifference in the prevalence of shoulder capsu-litis in insulin treated and non-insulin-treateddiabetic subjects.2' The higher prevalence ofshoulder capsulitis in type II diabetic subjectsseen in our study was most probably explainedby the higher age of these subjects compared totype I patients. In addition to the different ageand the duration of diabetes, type II diabeticsubjects differed from type I in that they had ahigher BMI, which is, however, unlikely toexplain higher prevalence of shoulder capsulitisbecause no association was found betweenshoulder capsulitis and BMI.

Smoking has been shown to contribute tomicrovascular complications in diabeticsubjects."0 Eadington et al found that smokingassociated with limited joint mobility but notwith Dupuytren disease,3' whereas Gamsted etal found no association between smoking andthese hand abnormalities."2 In the presentstudy, shoulder capsulitis was associated withprevious smoking in type I subjects, but thisassociation was fully explained by the fact thatsubjects who had previously smoked wereolder than non-smokers or present smokers.Shoulder capsulitis was associated with presentsmoking in type II diabetic subjects and thiswas not explained by the age of the patients, bythe duration of diabetes, or by the control ofdiabetes. The association may be explained bythe fact that smoking causes vasoconstriction,which may also underlie the development ofshoulder capsulitis.

Recently a strong relation between long termglycaemic control and diabetic microvascularcomplications has been definitely established.33It would be important to know if other diseasesclosely related to diabetes are also related tothe long term glycaemic control. Bridgmaninvestigated 800 diabetic patients (23% ofthem were insulin dependent) and he statedthat shoulder capsulitis correlated with theseverity of diabetes because 36% of theaffected (shoulder capsulitis) patients wereinsulin dependent.'4 Mavrikakis et al have alsostated that shoulder capsulitis is associatedwith the poor control of diabetes, because typeII diabetic subjects who had shoulder capsulitiswere more often insulin treated than thosewithout shoulder capsulitis.'6 We found no dif-ference in the prevalence of shoulder capsulitisin type II diabetic subjects treated withdifferent regimens. According to presentknowledge the type of regimen is not goodmarker of the control or severity of diabetes. Inthe present study, type II diabetic patients withpoor control of diabetes (GHbA1C > 9%)seemed to have more shoulder capsulitis thanpatients with better control (figure). However,no significant association between shouldercapsulitis and the control of diabetes wasfound in type I or II diabetic patients. The rea-son for not finding a significant associationmight be that there were very few patientswhose control was as good as in the intensivetherapy group of the DCCT study.33 It wouldbe important to know the effect of control ofdiabetes during the first years of the disease onthe development of shoulder capsulitis.The correlation between limited joint mobil-

ity and microvascular complications ofdiabetes has been well documented,2022 butthere are also two studies showing anassociation between shoulder capsulitis andretinopathy.'9 21 Our study also showed anassociation between shoulder capsulitis andproliferative retinopathy in type I patients, butthe correlation was fully explained by age andby the duration of diabetes. This indicates thatdiabetic subjects with shoulder capsulitis donot have a higher risk of microvascular compli-cations, such as limited joint mobility, thanthose without the condition.

912

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Shoulder capsulitis in type I and II diabetic patients

In some patients shoulder capsulitis mayprecede, accompany, or follow diffuse swelling,coldness, erythema, tenderness, and hyperhid-rosis of the hand.4 After weeks or months, thetenderness, swelling, and vasomotor dysfunc-tion completely resolve, with residual atrophicor dystrophic changes, finger contractures, andoccasionally shoulder capsulitis. This has beencalled shoulder-hand syndrome or reflexsympathetic dystrophy. Because of a possibleassociation between reflex sympathetic dystro-phy and shoulder capsulitis, autonomicneuropathy may help us to understand thepathogenesis of shoulder capsulitis aspreviously discussed by Steinbrocker andThompson."34 Our study supports this,because we observed about a fourfold and athreefold risk of autonomic neuropathy in typeI and II diabetic subjects with shoulder capsu-litis, respectively, compared to those withoutshoulder capsulitis when the confoundingeffects of age and duration of diabetes werecontrolled for. A subsample of 121 type Idiabetic patients also showed that shouldercapsulitis was associated with impaired heartrate variation, but this association was fullyexplained by the time related variables. Theassociation between shoulder capsulitis andautonomic neuropathy seen in the presentstudy must be interpreted with care, becauseour criteria for autonomic neuropathy were"soft" and the heart rate variation test or otherneurophysiological testing were not performedin all subjects.Many investigators have shown a higher

prevalence of shoulder capsulitis in diabeticpatients and in patients with a history of myo-cardial infarction.614 On the other hand,diabetes is associated with higher rates ofcardiovascular disease.'5 Previously Johnsonhas suggested that vasoconstriction of theperipheral arteries of the hand resulting fromcardiac pain, pre-existing arterioscleroticnarrowing of the vessels of the hands, andanoxia ofvarying duration and intensity result-ing from the myocardial infarction wereresponsible for the production of local changesin the hands and fingers.'6 Mavrikakis et al haveshown that diabetic subjects with shouldercapsulitis had higher cholesterol and triglycer-ide levels than diabetic subjects without shoul-der capsulitis or shoulder capsulitis controlgroup.'6 These results suggest that the higherprevalence of shoulder capsulitis in diabeticpatients could be explained by atheroscleroticchanges in vessels, leading to changes in localblood flow and producing altered physiology intendons, with resultant shoulder capsulitis. Westudied clinical features which could help us tounderstand the possible pathogenic associa-tions between shoulder capsulitis, diabetes,and myocardial infarction. Type I diabeticpatients with shoulder capsulitis seemed morefrequently to have had a history of myocardialinfarction than those without shoulder capsuli-tis, independent of the age of the patient, theduration of their diabetes, hypertension, orsmoking habits. The limited number of type Idiabetic subjects with a history of myocardialinfarction in the present study does not allow

us to draw final conclusions, but our data sug-gest an increase in the prevalence of shouldercapsulitis in patients with a history of myocar-dial infarction. An association betweenmyocardial infarction and shoulder capsulitisin type II diabetic patients was fully explainedby the time related factors. Thus thecorrelation between shoulder capsulitis andmyocardial infarction does not explain the highprevalence of shoulder capsulitis in type II dia-betic patients.

Shoulder capsulitis was associated withDupuytren disease in type I and with limitedjoint mobility in type II diabetic patients in thepresent study. Because Dupuytren disease andlimited joint mobility were strongly related tothe age and duration of diabetes, it is importantto control for this confounding effect. Afterusing a stepwise logistic regression analysis, thecorrelations between shoulder capsulitis andDupuytren disease and limited joint mobilitydisappeared. Fisher et al showed an associationbetween shoulder capsulitis and limited jointmobility in 57 diabetic patients.'8 Anotherstudy of 49 type I and 60 type II diabeticpatients showed no correlation between limitedjoint mobility and shoulder capsulitis.'5 Both ofthese studies may be misleading, becauseFisher et al eliminated half of their originalpatient population on the basis that their jointcontractures were obviously due to Dupuytrendisease, osteoarthritis, or flexor tenosynovitis,and Pal et al made no attempt to distinguishthe causes of the limitation in finger jointmobility.'5 18

In conclusion, this study shows that shouldercapsulitis is a common disorder in type I and IIdiabetic patients. It is associated with age in typeI and II diabetic patients and with the duration ofdiabetes in type I patients, which explains most ofits correlations with diabetic complications.Independent associations were found betweenshoulder capsulitis and autonomic neuropathy intype I and II diabetic subjects and with thehistory of myocardial infarction in tpe I only.

1 Duplay S. De la peri-arthrite scapulo-humerale et des raid-eurs de l'epaule qui en sont la consequence. Arch Gen Med1872;20:513-42.

2 Lundberg BJ. The frozen shoulder. Acta Orthop Scand 1969;suppi 1 9.

3 Codman EA. Rupture of the supraspinatus tendon andother lesions in or about the subacromial bursa. In: Theshoulder. Boston: Thomas Todd Co, 1934:216.

4 Rosenbloom AL. Connective tissue disorders in diabetes.In: International textbook of diabetes mellitus, vol 2. NewYork: John Wiley & Sons, 1992:1415-31.

5 Charnley J. Periarthritis of the shoulder. Postgrad Med Jf1959;35:384-8.

6 Steinbrocker 0, Argyros TG. The shoulder-hand syndrome:present status as a diagnostic and therapeutic entity. MedClin North Am 1958;42:1533-53.

7 Askey JM. The syndrome of painful disability of the shoul-der and hand complicating coronary occlusion. Am HeartJf 1941;22:1-12.

8 Minter WT. The shoulder-hand syndrome in coronary dis-ease. J7Med Ass Ga 1967;56:45-9.

9 Swan DM. Shoulder-hand syndrome following hemiplegia.Neurology 1954;4:480-2.

10 Johnson JTH. Frozen shoulder syndrome in patients withpulmonary tuberculosis. J Bone Joint Surg Am 1959;41:877.

11 Engleman RM. Shoulder pain as a presenting complaint inupper lobe bronchogenic carcinoma: report of 21 cases.Conn Med 1966;30:273-6.

12 Oldham BE. Periarthrosis of the shoulder associated withthyrotoxicosis. NZMedJ 1959;58:766-8.

13 Thompson M. Shoulder-hand syndrome. Proc R Soc Med1961;54:679-81.

14 Bridgman JF. Periarthritis of the shoulder and diabetes mel-litus. Ann Rheum Dis 1972;31:69-71.

913

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from

Arkkila, Kantola, Viikari, Ronnemaa

15 Pal B, Anderson J, Dick WC, Griffiths ID. Limitation ofjoint mobility and shoulder capsulitis in insulin- and non-insulin-dependent diabetes mellitus. BrJRheumatol 1986;25:147-51.

16 Mavrikakis ME, Sfikakis PP, Kontoyannis SA, AntoniadesLG, Kontoyannis DA, Moulopoulou DS. Clinical andlaboratory parameters in adult diabetics with and withoutcalcific shoulder periarthritis. Calcif Tiss Int 199 1;49:288-91.

17 Crisp AJ. Diabetes mellitus and the rheumatologist. Br JfRheumatol 1986;25:135-40.

18 Fisher L, Kurtz A, Shipley M. Association betweencheiroarthropathy and frozen shoulder in patients withinsulin-dependent diabetes mellitus. BrJ Rheumatol 1986;25:141-6.

19 Moren-Hybbinette I, Moritz U, Schersten B. The clinicalpicture of the painful diabetic shoulder-natural history,social consequences and analysis of concomitant handsyndrome. Acta Med Scand 1987;221:73-82.

20 Rosenbloom AL, Silverstein JH, Lezotte DC, Richardson K,McCallum M. Limited joint mobility in childhooddiabetes indicates increased risk for microvascular disease.NEnglJMed 1981;305:191-4.

21 Sattar MA, Luqman WA. Periarthritis: another duration-related complication of diabetes mellitus. Diabetes Care1985;8:507-10.

22 Arkkila PET, Kantola IM, Viikari JSA. Limited joint mobil-ity in type 1 diabetic patients: correlation to other diabeticcomplications. J Intern Med 1994;236:215-23.

23 National Diabetes Data Group. Classification and diagnosisof diabetes mellitus and other categories of glucoseintolerance. Diabetes 1979;28: 1039-57.

24 Noble J, Heathcote JG, Cohn H. Diabetes mellitus in theaetiology of Dupuytren's disease. Y Bone Joint Surg Br1984;66:322-5.

25 Murray A, Ewing DJ, Campbell IW, Neilson JMM, ClarkeBF. RR interval variations in young male diabetics. Br

Hearty 1975;37:882-5.26 Ewing DJ, Clarke BF. Diagnosis and management of

diabetic autonomic neuropathy. BMJ 1982;285:916-8.27 Koskinen P. Nontransferability of C-peptide measurements

with various commercial radioimmunoassay reagents. ClinChem 1988;34:1575-8.

28 Kouri TT, Viikari JSA, Mattila KS, Irjala KMA. Micro-albuminuria. Invalidity of simple concentration-basedscreening tests for early nephropathy due to urinaryvolumes of diabetic patients. Diabetes Care 1991;14:591-3.

29 Dixon WJ, ed. BMDP statistical software. Los Angeles: Uni-versity of California Press, 1990.

30 Muhlhauser I. Smoking and diabetes. Diabet Med 1990;7:10-15.

31 Eadington DW, Patrick AW, Frier CB. Limited joint mobil-ity, Dupuytren's contracture and retinopathy in type 1diabetes: association with cigarette smoking. Diabet Med1989; 6:152-7.

32 Gamstedt A, Holm-Glad J, Ohlson C-G, Sundstrom M.Hand abnormalities are strongly associated with the dura-tion of diabetes mellitus. J Intern Med 1993;234:189-93.

33 The Diabetes Control and Complications Trial researchgroup. The effect of intensive treatment of diabetes on thedevelopment and progression of long-term complicationsin insulin-dependent diabetes mellitus. N Engl Y Med1993;329:977-86.

34 Steinbrocker 0, Neustadt D, Bosch SJ. Painful shouldersyndromes. Their diagnosis and treatment. Med Clin NorthAm 1955;39:563-85.

35 Wilson PW, Anderson KM, Kannel WB. Epidemiology ofdiabetes mellitus in the elderly: the Framingham study.AmYMed 1986;80:3-9.

36 Johnson AC. Disabling changes in the hand resemblingsclerodactylia foliowing myocardial infarction. Ann InternMed 1943;19:433-56.

914

on March 17, 2021 by guest. P

rotected by copyright.http://ard.bm

j.com/

Ann R

heum D

is: first published as 10.1136/ard.55.12.907 on 1 Decem

ber 1996. Dow

nloaded from