514

Shoulder Pain in Hemiplegia: Statistical

Relationship with Five Variables
Richard W. Bohannon, PT, Patricia A. Larkin, PT, Melissa B. Smith, PT, Melissa G. Horton, PT
Southeastern Regional Rehabilitation Center. Fayetteville, NC 28302

ABSTRACT. Bohannon RW, Larkin PA, Smith MB, Horton MG: Shoulder pain in hemiplegia: statistical relation-
ship with five variables. Arch Phys Med Rehabil 67514-516, 1986.
0 The incidence of shoulder pain and the statistical relationship between it and five other variables (patient age,
time since onset of hemiplegia, range of hemiplegic shoulder external rotation, spasticity and weakness) were inves-
tigated retrospectively. Of 50 consecutive hemiplegic patients whose records were reviewed, 36 had shoulder pain.
The variables significantly (~~0.01) correlated with shoulder pain were: time since onset of hemiplegia (r = 0.45)
and ROSER (I = - 0.61). The relationship between shoulder pain and range of shoulder motion remained significant
when other factors were partialled out. The relationship between shoulder pain and time since onset was not significant
when the affect of range of shoulder motion was partialled out. Therefore, range of shoulder external rotation was
considered the factor related most significantly to shoulder pain. This finding suggests that shoulder pain demonstrated
by hemiplegia patients may be, in part, a manifestation of adhesive capsulitis.

KEY WORDS: Hemiplegia; Shoulder; Shoulder-hand syndrome: Shoulder joint

As clinicians are well aware, hemiplegia is often accom- Table 1: Mean and Standard Deviation of Factors
panied by shoulder pain. Although the specific cause of this Whose Relationship with Shoulder Pain was Tested
pain remains unclear, several factors have been associated with Pronounced/
it: glenohumeral malalignment or subluxation,8*‘3*‘9~20 mental No pain Some pain severe pain
(n= 14) (n = 20) (n = 16)
depression, l9 adhesive changes or decreased range of motion
Factor X s X s SI S
(ROM) of the shoulder,3,6*8*9,‘6,18 rotator cuff tears,13 cap-
Age (years) 60.6 16.8 64.6 11.6 64.9 13.5
sulitis,7 swollen hemiplegic hands,16 and spasticity of the
Days since onset 25.0 19.9 28.4 15.6 81.1 73.8
shoulder musculature.5 Most of these associations have been Range of motion 89.6” 1.3” 84.2 10.4” 59.2” 24.3”
implied, rather than statistically established. A clearer under- Tone (Ashworth .7 .9 1.0 .9 1.3 1.1
standing of the relationship between specific variables and rank)
shoulder pain is, therefore, warranted. The study was under- Strength deficit 77.5% 35.1% 82.5% 26.2% 81.7% 30.1%
taken to determine, in patients with hemiplegia, the incidence
of shoulder pain and the statistical relationship between shoul- nonhemiplegic shoulder, in the abduction and external rotation
der pain and five other variables (patient age, time since onset movements (table 1). The Norkin and White technique,15 em-
of hemiplegia, range of hemiplegia shoulder external rotation ploying a standard full-circle goniometer was used to deter-
(ROSER), spasticity of the hemiplegic shoulder internal ro- mine ROM. Weakness was determined with gravity eliminated,
tator muscles, and weakness of the hemiplegic shoulder ab- using a hand-held dynamometer, as described by Bohannon.2
ductor and external rotator muscles). Pain, spasticity, and ROM were measured simultaneously as
the hemiplegic shoulder was externally rotated to the maximal
MATERIALS AND METHODS tolerable range (up to 90”). Relationships between pain and
other variables were determined, employing computer pro-
Consecutive records were analyzed of patients whose hemi-
grams which calculate correlations and analysis of variance
plegia was secondary to cerebrovascular accident, whose un-
(ANOVA). “*l’ All zero-order, multiple, and partial correla-
affected shoulders demonstrated normal and pain-free ROSER
(90‘7, and who were able to adequately follow instructions to tions were determined. The level of significance selected for
allow testing of all variables pertinent to the study. Of the 50 this study was p<O.Ol.
patients whose records were reviewed, 30 were women and
20 were men; 30 were hemiplegic on the left and 20 on the RESULTS
right. The information retrieved from their initial physical ther-
apy evaluation included: shoulder pain, as graded by Fugl- Of the 50 patients whose records were reviewed, 36 (72%)
Meyer and associate& (pronounced during all the movement had shoulder pain. Twenty had some pain, and 16 had severe
or very marked pain at the end of the actual ROM, (some pain. Table 2 shows the correlation matrix for all zero-order
pain, no pain), patient age; time between onset of hemiplegia Pearson product moment correlations. Three zero-order cor-
and initial evaluation; ROSER in degrees; spasticity of shoul- relations were significant: ROSER and shoulder pain (r =
der internal rotator muscles, as graded ordinally (04), by -0.061, p < 0.001); time since onset of hemiplegia and
Ashworth,’ and weakness expressed as the mean static strength
deficit (percent) of the hemiplegic shoulder, compared to the Submitted for publication November 19. 1985. Accepted February 18. 1986.

Arch Phys Med Rehabil Vol67, August 1996

 SHOULDER PAIN IN HEMIPLEGIA, Bohannon 515

shoulder pain (r = 0.45, p < 0.01); and time since onset of Table 3: ANOVA Summary Table for Patients Grouped Ac-
hemiplegia and ROSER (r = 0.37, p < 0.01). cording to Severity of Shoulder Pain
All multiple correlations with pain that included ROSER For ROSER
were significant @ < 0.001). The multiple correlations with Source ss df MS F P
pain that included ROSER (maximum multiple r = 0.67) were Shoulder pain 8578.37 2 4289.19 18.44 0.001
only slightly greater than the zero-order correlation between Error 10931.96 47 232.59
pain and ROSER. Multiple correlations with pain that did not For time since onset of hemiplegia
include ROSER were significant only if they included time Source ss df MS F P
since onset of hemiplegia. All correlations between pain and Shoulder pain 32256.18 2 16128.09 8.28 0.001
ROSER remained significant, regardless of the combination Error 91585.49 47 1948.63
of variables partialled out (range of partial r = -0.53 to
-0.59). Correlations between pain and time since onset were
was only marginally higher than the zero-order correlation be-
not significant if ROSER was partialled out.
tween shoulder pain and ROSER, factors investigated in this
One-way ANOVA (table 3) was used to clarify the signif-
study, other than ROM, cannot be considered particularly im-
icant relationships with pain. The ANOVAs demonstrated that
portant. Spasticity, which Caldwell and associates5 suggested
the time since onset of hemiplegia and the ROSER were sig-
to be important, was unrelated to shoulder pain. Weakness of
nificantly different in patients with no pain, some pain, and
the muscles that externally rotate and abduct the humerus was
pronounce&severe pain. Post-hoc analyses revealed no differ-
also unrelated to shoulder pain. This finding was unexpected,
ences in ROSER or in time since onset in the no-pain and the
given that patients with complete or severe paralysis of the
some-pain groups. The analyses did, however, demonstrate
shoulder muscles may more often demonstrate malalignment
both the time since onset and the ROSER in the no-pain group
than patients with paresis or partial paralysis,12 and that a
to differ significantly from that in the pronounced/severe-pain
statistically significant relationship has been reported between
group, and that in the some-pain group to differ significantly
shoulder malalignment and pain.20 Subluxation and other fac-
from that in the pronounced/severe-pain group @ < 0.01).
tors (including weakness and spasticity of other muscle groups)
which may further explain the incidence of shoulder pain merit
DISCUSSION further investigation.
Although shoulder pain is a frequent complication of hemi- Only the ROSER was tested to the extremes of range in all
plegia, the incidence (72%) among patients in this study was patients. Given that this movement is typically quite limited
higher than that reported by others.9*20 Perhaps this is because in adhesive capsulitis, 14*” and that both capsulitis and adhe-
we determined the presence of shoulder pain during passive sive changes have been identified in hemiplegic shoul-
shoulder external rotation, rather than relying on the patient’s ders,7+9,L8 this study may suggest that hemiplegic shoulder
recollection. The incidence in this study was similar to that pain is, at least in part, a manifestation of adhesive capsulitis.
reported by Grossen-Sills and associates,8 who also reported Of course, contracture of the shoulder internal rotator muscles
the pain that was experienced during ROM. could cause a similar limitation in range.
Of the factors associated with shoulder pain, ROSER ap- This study, which retrospectively investigated relationships,
pears most clearly related. The relationship between pain and does not identify the cause of shoulder pain. It is, nonetheless,
ROM remained significant when other factors were partialled interesting that only patients with full ROSER were pain free.
out, whereas, other relationships with pain were no longer Perhaps, capsulitis without an adhesive component leads ini-
significant when ROSER was partialled out. The ANOVA and tially to pain. If the capsulitis worsens over time and adhesions
post-hoc tests suggest that the significance of this relationship occur, pain may also worsen and be accompanied by a de-
may depend on the presence or absence of severe pain and crease in ROM. If so, treatment might best be directed toward
that the association between shoulder ROM and pain, implied both the elimination of inflammation and the maintenance of
but not established statistically by others,3,9,16,18 is valid. ROM. Reports of efforts to maintain ROM have not, to date,
Although the relationship between shoulder pain and ROM been encouraging. 4,21 A prospective study, investigating the
was significant (p < O.OOl), the coefficient of determination effects of specific interventions on ROSER and pain, might,
(? = 0.37) is small. Thus, the variance in the incidence of nonetheless, prove useful. Such a study, unlike this one, could
shoulder pain can only be explained minimally by the variance investigate pain and other variables over time.
in ROM. Because the multiple correlations between shoulder
pain and any combination of the other five factors investigated ADDRESS REPRINT REQUESTS TO:
Richard W. Bohannon, PT
Table 2: Correlations Between Shoulder Pain and Other Department of Physical Therapy
Factors in 50 Hemiplegic Patients Southeastern Regional Rehabilitation Center
Cape Fear Valley Medical Center
Range of Time since
Tone Weakness onset PO Box 2000
Pain motion
Fayetteville, NC 28302
Pain
Range of motion - .61
Tone .21 -.25 References
Weakness .05 .Ol .22 1. Ashworth B: Preliminary trial of carisoprodol in multiple scle-
Days since onset .45 - .37 .29 - .07 rosis. Practitioner 192:54&542, 1964
Age .I2 .Oo - .I9 - .30 .06 2. Bohatmon RW: Test-retest reliability of hand-held dynamometty

Amh Phys Mod Rehabll Voi 67, August 1986

516 SHOULDERPAIN IN HEMIPLEGIA, Bohannon

during single session of strength assessment. Phys Ther f&:206- 11. Madigan S, Lawrence V: Regress 11. A Multiple Regression
209, 1986 Program for the Apple II/II e/IIc. Northridge, CA, Human Sys-
3. Braun RM, West F, Mooney V, Nickel VL, Roper B, Caldwell tems Dynamics, 1983
C: Surgical treatment of painful shoulder contracture in stroke 12. Najenson T, Pikielny SS: Malaligmnent of gleno-humeral joint
patient. J Bone Joint Surg [Am] 53:1307-1312, 1971 following hemiplegia: review of 500 cases. Ann Phys Med 8:96-
4. Brocklehurst JC, Andtews K, Richards B, Leycock PJ: How 99, 1965
much physical therapy for patients with stroke? Br Med J 1: 1307- 13. Najenson T, Yacubovich E, Pikielni SS: Rotator cuff injury in
1309, 1978 shoulder joint of hemiplegia patients. Stand J Rehabil Med 3: 13 I-
5. Caldwell CB, Wilson DJ, Braun, RM: Evaluation and treatment 137, 1971
of upper extremity in hemiplegic stroke patient. Clin Orthop 14. Neviaser JS: Adhesive capsulitis of shoulder: study of patholog-
63:69-93, 1969 ical findings in periarthritis of shoulder. J Bone Joint Surg [Am]
6. Fugl-Meyer AR, Jtiskii L, Leyman I, Olsson S, Steglind S: Post- 27:21 l-222, 1945
stroke hemiplegic patient: I. method for evaluation of physical 15. Norkin CC, White DJ: Measurement of Joint Motion: Guide to
performance. Stand J Rehabil Med 7: 13-3 1, 1975 Goniometry. Philadelphia, Davis, 1985, pp 36-37
7. Fukui K: Findings and course of hemiplegic shoulders in view 16. Peszczynski M, Rardin TE: Incidence of painful shoulder in
of roentgenogram with contrast medium. Jpn J Rehabil Med hemiplegia. Bull Polish Med Sci Hist 8:21-23, 1965
9182, 1972 17. Rizk TE, Christopher RP, Pinals RS, Higgins AC, Frix R: Ad-
8. Grossen-Sills J, Schenkman M: Analysis of shoulder pain, range hesive capsulitis (frozen shoulder): new approach to its manage-
of motion, and subluxation in patients with hemiplegia. Phys ment. Arch Phys Med Rehabil64:29-33, 1983
Ther 65:73 1, 1985 18. Rizk TE, Christopher RP, Pinals RS, Salazar JE, Higgins C:
9. Hakuno A, Sash&a H, Ohkawa T, Itoh R: Arthographic findings Arthrographic studies in painful hemiplegic shoulders. Arch Phys
in hemiplegic shoulders. Arch Phys Med Rehabil 65:706-711, Med Rehabil65:254-256, 1984
1984 19. Savage R, Robertson L: Relationship between adult hemiplegic
10. Madigan S, Lawrence V: Anova II. A General Analysis of Var- shoulder pain and depression. Physiotherapy Can 34:86-90, 1982
iance Program for the Apple II. Northridge, CA, Human Systems 20. Shai G, Ring H, Costeff H, Solzi P: Glenohumeral malaligmnent
Dynamics, 1983 in hemiplegic shoulder. Stand J Rehabil Med 16:133-136, 1984

PM&R BOARD EXAMINATION DATES

Dates for the 1987 examinations of the American Board of Physical
Medicine and Rehabilitation are:
Part I (written), May 15, 1987
Part II (oral), May 16-l 7, 1987
in Rochester, Minnesota
(Option for Written in Philadelphia)
DEADLINE for receiving completed application (without penalty) is November 15, 1986
Penalty of $100 for late applications (postmarked November 16-December 15, 1986)
No applications, complete or incomplete, will be accepted after December 15.
For information and applications, write to:
Gordon M. Martin, MD, Executive Director
American Board of PM&R
Suite 674, Nor-west Center
21 First Street, Southwest
Rochester, MN 55902
(507) 282-l 776
Requests for applications will not be accepted by telephone. Each candidate must
write for his/her own information and application.

Arch Phys Med Rehabil Vol67, August 1986