Dekker, Jos H. M.2; Wagenaar, Robert C.; Lankhorst, Gustaaf J.; de Jong, Bareld A.
Effects of intra-articular triamcinolone acetonide on pain and passive range of motion (ROM) in the painful hemiplegic shoulder were studied. A Multiple baseline (or AB) design across seven subjects was used. The length of the baseline condition (or A phase) was either 2 or 3 wk, and randomized across subjects. Subsequently, a treatment condition (or B phase) of 4 wk was applied during which three intra-articular injections of triamcinolone acetonide were administered at day 1, 8, and 22. Pain and ROM were the primary outcome parameters and were measured three times each week by means of a visual analogue scale (VAS) and a fluid-filled goniometer, respectively. In addition, a number of secondary outcome parameters were assessed, i.e., spastic muscle activity (Ashworth scale), motor function (Fugl-Meyer index), upper limb function (action research arm test) and signs and symptoms of a shoulder hand syndrome (clinical scoring list). Statistical analysis of the combined time series showed significant effects on pain (P = 0.025). Analysis of the individual time series revealed that five out of seven patients had significant reduction of pain. ROM improved significantly in four out of seven patients. However, improvement of ROM did not reach significance at the group level (P = 0.13). None of the secondary parameters showed significant changes. The correlation coefficient between upper limb function (ARA) at intake and size of treatment effect approached a level of significance (P = 0.09). The results indicate that intra-articular triamcinolone may be of benefit in reducing hemiplegic shoulder pain.
Objectives: Upon completion of this article, the reader should be able to (1) describe the present knowledge about causes and treatment of hemiplegic shoulder pain; (2) discuss the strengths and weaknesses of the multiple baseline across subjects design; (3) discuss the outcome of pain, range of motion, and arm function parameters after repeated intra-articular steroid injections for hemiplegic shoulder pain. Level: Advanced.
The Association of Academic Physiatrists is accredited by the Accreditation Council for Continuing Medical Education to sponsor continuing medical education for physicians.
The Association of Academic Physiatrists designates this continuing medical education activity for one credit hour in Category 1 of the Physician's Recognition Award of the American Medical Association.
*Disclosure statements have been obtained regarding the author's relationships with financial supporters of this activity. There is no apparent conflict of interest related to the context of participation of the authors of this article.
The painful hemiplegic shoulder often complicates and prolongs rehabilitation in hemiplegia. Progress in performing activities of daily life, walking, and functional recovery of the upper limb is often hampered. A recent study identified that hemiplegic shoulder pain affected stroke outcome in a negative way.1 Estimates of incidence vary from 40 to 80% of newly developed cerebrovascular incidents. For the Netherlands this implies development of 8,000-16,000 new cases of shoulder pain yearly.2
A number of disorders have been proposed in the literature as being major causes of pain in the hemiplegic shoulder, e.g., glenohumeral (sub)luxation,3, 4 spasticity of shoulder muscles,5-7 impingement,8, 9 soft tissue trauma,9 rotator cuff tears,10-12 and shoulder-hand syndrome.5 All of these mechanisms have been suggested rather than established.13 Most authors consider the pathogenesis of hemiplegic shoulder pain to be multifactorial.13
The current treatment of shoulder pain in hemiplegia varies according to the assumptions made concerning the cause of pain. Examples are spasmolytic drugs,5 relaxation techniques,6 prevention of impingement,9 and slings.14 None of these treatment strategies has been shown to give completely satisfactory results, probably because of the lack of prospective, controlled research.15 Considerable agreement, however, exists with respect to preventive measures such as adequate positioning and handling.13
Restriction of external rotation relates most significantly to shoulder pain,7, 9, 13, 16-19 whereas glenohumeral abduction is less impaired. This restriction of glenohumeral range of motion (ROM) is similar to observations in idiopathic and traumatic capsulitis. Pathologic findings at autopsy 20 and arthrography 21 confirm the resemblance between painful hemiplegic shoulders and nonhemiplegic glenohumeral capsulitis. In several studies 7, 21, 22 the occurrence of glenohumeral capsulitis is postulated to play an important role in hemiplegic shoulder pain.
Recently good results were reported concerning the administration of local steroids in idiopathic and traumatic capsulitis of the glenohumeral joint.22 About 75% of the patients became free of pain at the end of treatment, whereas 90% of patients were free from pain at night. Improvements of ROM were less impressive; 80% of patients still had a mild restriction of ROM at the end of the study.
The reported powerful effect on pain in capsulitis of the nonhemiplegic shoulder in the above mentioned study 22 in combination with the resemblance of traumatic capsulitis and the hemiplegic painful shoulder presented strong arguments for the study of the efficacy of administration of local steroids in hemiplegia after stroke.
The aim of this study is to answer the following questions:
A multiple baseline design across subjects was applied. The length of the baseline condition (A phase) was staggered and randomized (either 2 or 3 wk) after confirmation of inclusion criteria. The treatment condition (B phase) lasted 4 wk. In both A and B phase, patients were assessed three times a week. The treatment consisted of a series of three intra-articular injections of 40 mg of triamcinolone acetonide via the posterior route 22. The patients were seated and the arm of the patient was held somewhat in internal rotation. The index finger of the physician was placed on the coracoid process and the thumb on the angle between the spine of the scapula and the acromion. The needle (5 cm) was introduced 1 cm below the thumb and aimed at the coracoid process.
Injections were given at days 1, 8, and 22 of treatment. During the whole period of study, physiotherapy and occupational therapy regimens and care by the nursing staff were given according to the existing protocol. No changes were made because of the study. Only on the days of injection and one subsequent day no treatment was given.
A multiple baseline design across subjects allows statistical evaluation of group treatment effects as well as individual treatment effects. The randomization of phase lengths controls for selection bias and contamination. A recent review on stroke rehabilitation research shows that interrupted time series or controlled single case experimental designs allow the detection of clinically relevant effects using a small number of patients.23-25
Nine consecutive inpatients in the Rehabilitation Center Amsterdam with hemiplegic shoulder pain entered the study in a 12-mo period. The patients met the following inclusion criteria: (1) presence of shoulder pain in the paretic arm with disturbance of sleep, (2) no previous history of shoulder pain, (3) presence of ROM restriction of external rotation, (4) no changes in pain medication during period of study, and (5) informed consent.
Intake of the patients consisted of recording sex, age, time elapsed between development of the lesion and development of shoulder pain, and location and nature of the neurologic condition. Unilateral neglect was assessed by letter cancellation tasks. All neurologic lesions (cerebral infarctions) had been confirmed by computerized tomograph scan or magnetic resonance imaging (see Table 1).
|TABLE 1 Patient characteristics|
X-rays were made of both the affected and the unaffected shoulder at the time of entering the study to evaluate the presence of subluxation of the glenohumeral joint.26,27 Only a minor subluxation in the affected shoulder of patient 5 was found.
Two patients dropped out. One patient moved to South America without further notice. The baseline attributed to this patient was 2 wk. The other patient showed a major improvement in pain and ROM during the baseline period. Treatment was no longer indicated. This patient was attributed a 3-wk baseline. The removal of these patients did not influence randomization.
Statistics were performed using Statgraphics®2.6. Measurements of ROM were performed with a MIE fluid-filled gravity goniometer by Biokinetics®.
Pain was measured with a vertical visual analogue scale.28 The vertical scale was chosen to control for the influence of possible unilateral neglect. Reliability and validity of the visual analogue scale (VAS) for pain scoring purposes were found best using a VAS without calibration.29
ROM was measured by determining external rotation by means of fluid-filled goniometer. This has been shown to be a reliable and sensitive measurement procedure.30
Spastic activity in the arm muscles was evaluated according to the Ashworth (5-point) scale.31 This scale has recently been reexamined and found reliable.32 Spastic activity was examined during passive motion in the elbow joint of the affected arm.
Motor function of the upper limb was assessed according to the Fugl-Meyer index,33 using the subtest for upper limb function (5-point scale).
Presence of clinical characteristics of a shoulderhand syndrome 34-36 was assessed. Because no scale or index exists for shoulder-hand syndrome we used the following items on which there seems to be consensus in literature. Any present symptom was assigned 1 point: (1) pain in shoulder and hand, (2) capsular pattern of restriction of ROM of the glenohumeral joint, (3) edema of hand and fingers, (4) hyperesthesia of the hand, (5) change in color of the hand, and (6) atrophy. No information on reliability and validity on this item is available.
Upper limb function was evaluated with the action research arm (ARA) test once a week. Scores range from 0 to 57. It has been shown to have a high validity, high test-retest reliability, and is practically applicable.37, 38
Side effects of the administered drug were recorded each session. The occurrence of flaring, skin rashes, acne, and disturbances in blood glucose levels were monitored.
Single case methodology requires stepwise analysis of the collected data. Before statistical comparison between intervention phases or conditions can take place, serial dependency-for example, as a result of spontaneous recovery-in the individual series needs to be identified and controlled for.23, 39 To evaluate group effects, regression analysis was applied to the time series of the A phase and the B phase separately. Analysis took place by evaluating significance of fit of data to linear, multiplicative, exponential, or reciprocal regression models, respectively. An exponential regression curve explained the largest amount of variance in pain time series, whereas a linear regression function was superior for the ROM time series. Subsequently, the slopes of the A phases and B phases of the seven subjects were statistically compared, using a one-tailed Wilcoxon-Rank comparison of two samples. The significance level was chosen at 0.05.
Subsequently, time series analysis was performed to detect differences in efficacy within each subject. Regression analysis was used to identify serial dependency in the individual baselines. When a significant fit of a regression curve was found, this trend was extrapolated into the B phase. The next step consisted of removing this trend in the complete time series. Subsequently, Kruskal-Wallis one-way analysis of variance was applied to compare the data in the A and the B phases.
For the analysis of effects on the secondary parameters the same methodology was used. Unlike the other assessments, the ARA test could only be applied once a week. This did not produce enough assessment points to obtain reliable statistics. For this reason the data of the ARA were not further analyzed as described above.
To study correlations between initial values of secondary parameters and size of treatment effect, Spearman-Rank correlation coefficients were calculated. The differences of the slopes of the regression curves from the individual A and B phases were taken as a measure for size of treatment effect.
|Figure 1. Pain scores of patients who responded with pain reduction. On the x-axis measurement points are indicated in time. (Because three measurements took place in 1 wk, one unit corresponds to approximately 2 days.) The time series are synchronized around time = 0, when each patient received the first injection. So each point left of zero means baseline (or A phase) and has a negative value. Each point right of zero means treatment phase (or B phase) and has a positive value. The next injections were administered at time = 3 and time = 9. y-axis: pain score as measured by VAS.|
|Figure 2. Pain scores of patients who did not respond with pain reduction. x-axis: measurement points in time (one unit corresponds to approximately 2 days); y-axis: pain score as measured by VAS.|
Visual inspection of the individual time series suggests a large reduction of pain as a result of the intervention. This reduction took place after either the first or the second injection. Statistical analysis of the combined time series (for detecting a group effect) revealed a significant reduction in pain (P = 0.025). Analysis of the individual time series showed no significant effect in patients 1 (P = 0.23) and 5 (P = 0.057), but a highly significant effect in patients 2, 6, 7, 8, and 9 (P < 0.0005). Both patients 1 and 5 had a unilateral neglect and a low score on the ARA test.
|Figure 3. Degrees of external rotation in the glenohumeral joint of patients who responded with pain reduction. x-axis: measurement points in time (one unit corresponds to approximately 2 days). y-axis: degrees of external rotation as measured with a goniometer.|
|Figure 4. Degrees of external rotation in the glenohumeral joint of patients who did not respond with pain reduction. x-axis: measurement points in time (one unit corresponds to approximately 2 days). y-axis: degrees of external rotation as measured with a goniometer.|
Visual inspection of the individual time series suggests a more variable response for ROM than for pain. Analysis of the combined time series did not reveal significant differences (P = 0.13). Statistical analysis of the individual time series showed no significant effect in patients 5 (P = 0.07), 7 (P = 0.14), and 9 (P = 0.29), but significant effects in patients 1, 2, 6, and 8 (P < 0.02). No clinical explanation could be provided to account for the differences between response in pain reduction and response in ROM improvement.
None of the secondary parameters showed a statistically significant difference between the two conditions when combining the time series (P > 0.07). Analysis of the individual time series revealed a significant increase of spasticity for patients 1 and 6, and a significant decrease of spasticity for patient 2. The Fugl-Meyer index showed significant improvements for patients 1, 6, and 9. With regard to the shoulder hand syndrome a significant diminishing of symptoms for patients 1, 2, 6, and 9 was found.
No significant Spearman-Rank correlation coefficients were found between the initial values of the secondary parameters and the the size of treatment effect, i.e., for Ashworth scale r = 0.67 (P = 0.10), for Fugl-Meyer index r = 0.31 (P = 0.45), and for shoulder-hand syndrome score r = 0.22 (P = 0.59). Only the Spearman-Rank correlation coefficient for the initial value of the ARA test approached significance: r = 0.70 (P = 0.09).
One side effect of triamcinolone administration, although not of serious nature, was reported frequently. Five out of seven patients complained of flaring in the first 5 days after the injections. Although three patients had diabetes, no disturbances in blood glucose levels were detected.
The results of this intervention study show a positive effect of intra-articular triamcinolone on pain in five out of seven patients. It is not clear, however, why three patients experienced reduction of pain after the first injection and two only after the second injection. Improvement of ROM did not reach the level of significance. However, statistical analysis of the individual time series showed a significant improvement for ROM in four out of seven patients. Our findings are consistent with the results of de Jong,22 who also found a better response for pain than for ROM in traumatic and idiopathic capsulitis of the shoulder.
None of the secondary outcome parameters reached the level of significance in the group evaluations. However, in the analysis of the individual series four out of seven patients showed a significant improvement of clinical symptoms of the shoulder-hand syndrome as result of the intervention.
Most authors conclude that careful positioning, adequate support, and proper handling remain the key actions to prevent hemiplegic shoulder pain.13 The administration of triamcinolone offers clear advantages over other measures taken by nursing staff and therapists. In practical terms the results of this study indicate that triamcinolone is of benefit in reducing hemiplegic shoulder pain.
The results would probably have gained more validity if a placebo treatment could have been incorporated in the design to control for the effect of the injection itself. However, this was not feasible for technical reasons, as no proper placebo fluid similar in appearance was available. Another shortcoming of this study is that the follow-up period was short. Possible relapses of shoulder pain after the period of study were not recorded.
The presence or absence of glenohumeral subluxation did not seem to have a relation to response patterns. This has also been reported by others,15 but literature is contradictory 19 in this respect. Possibly the initial values of the ARA test point to a predictive value for the size of pain reduction. These findings might prove interesting enough for further study.
The multiple baseline design across subjects proved to be useful.39 Significant pain reduction as a result of the intervention could be demonstrated by using only seven patients in a limited period of study. Differential effects on ROM were found at the individual level only. The same applies for effects on shoulder-hand syndrome.
If drawbacks of single-subject designs such as carryover effects between phases or reactive effects of repeated measures cannot be avoided, this type of methodology is not advocated.23 Otherwise it seems to be an attractive and elegant tool that deserves its place in rehabilitation research.
We wish to thank H. Grupstra for administration of the intra-articular injections. We are grateful to the Rehabilitation Centre Amsterdam and Rehabilitation Centre de Trappenberg for the support for this study.
1. Roy CW, Sands MR, Hill LD, Harrison A, Marshall S: The effect of shoulder pain on outcome of acute hemiplegia. Clin Rehabil 1995;9:21-27. [Context Link]
2. Cools HJM: Estimates on numbers and and residence of stroke patients in the Netherlands (in Dutch). Reader Basiscursus Revalidatie Diagnostiek en Behandeling van het Cerebrovasculaire Accident. Paog, Amsterdam, 1991. [Context Link]
3. Najenson T, Pikielny S: Malalignment of the glenohumeral joint following hemiplegia: a review of 500 cases. Ann Phys Med 1965;8:96-99. [Context Link]
4. Van Langenberghe HVK, Hogan BM: Degree of pain and grade of subluxation in the painful hemiplegic shoulder. Scand J Rehabil Med 1988;20:161-166. [Context Link]
6. Bobath K: Letter to editor. Phys Ther 1972;52:444-445. [Context Link]
9. Cailliet R: The Shoulder in Hemiplegia. Philadelphia, FA Davis, 1980. [Context Link]
13. Griffin JW: Hemiplegic shoulder pain. Phys Ther 1986;12:1884-1893. [Context Link]
14. de Vries J: Schouderpijn bij de hemiplegische patient. PhD thesis, RUG Groningen, 1978. [Context Link]
15. Bruton JD: Shoulder pain in stroke patients with hemiplegia or hemiparesis following a cerebrovascular accident. Physiotherapy 1985;71:2-4. [Context Link]
16. Hurd MM, Farell KH, Waylonis GW: Shoulder sling for hemiplegia: friend or foe? Arch Phys Med Rehabil 1974;55:519-522. [Context Link]
17. Kumar R, Metter EJ, Mehta AJ, Chew T: Shoulder pain in hemiplegia; the role of exercise. Am J Phys Med Rehabil 1990;69:205-208. [Context Link]
19. Griffin J, Reddin G: Shoulder pain in patients with hemiplegia. Phys Ther 1981;61:1041-1045. [Context Link]
20. DeSèze S: Etudes sur L'epaule douloureuze. 1. Les lesions anatomique de l'epaule bloquee. Rev Rhum 1959;67:323. [Context Link]
22. de Jong BA: The painful stiff shoulder. PhD thesis, University of Amsterdam, Amsterdam, 1991. [Context Link]
23. Wagenaar RC, Meijer OG: Effects of stroke rehabilitation. I,II. A critical review of the literature. J Rehabil Sci 1991;4:61-73. [Context Link]
24. Kazdin AE: Statistical analysis for single case experimental designs, in Hersen M, Barlow DH (eds): Single Case Experimental Designs: Strategies for Studying Behaviour Change. Pergamon Press, Oxford, 1982, pp 265-316. [Context Link]
25. McCain LJ, McClearly R.: Statistical analysis of the simple interrupted time-series quasi-experiment, in Cook TD, Campbell DT (eds): Quasi-experimentation: Design and Analysis Issues for Field Settings. Boston, Houghton Mifflin, 1979, pp 233-293. [Context Link]
27. Arsenault AB, Bilodeau M, Dutil E, Riley E: Clinical significance of the V-shaped space in the subluxed shoulder of hemiplegics. Stroke 1991;22:867-871. Ovid Full Text Bibliographic Links [Context Link]
28. Rouwet E, van Egmond J, Nijhuis G: Analysis of methods of scoring pain; measurement of and discrimination between subjective feelings (pain and fear) (in Dutch). Ned Tijdschr Geneeskd 1985;129:2356-2359. [Context Link]
30. Andrews AW, Bohannon RW: Decreased shoulder range of motion on paretic side after stroke. Phys Ther 1989;69.9:768-772. [Context Link]
31. Ashworth B: Preliminary trial of carisoprodol in multiple sclerosis. Practitioner 1964;192:540-542. [Context Link]
32. Lee KC, Carson L, Kinnin E, Patterson V: The Ashworth Scale: a reliable and reproducible method of measuring spasticity. J Neuro Rehabil 1989;3:205-209. [Context Link]
33. Fugl-Meyer AR, Jääskö L, Leyman I, Olsson S, Steglind S: The post-stroke hemiplegic patient, a method for evaluation of physical performance. Scand J Rehabil Med 1975;7:13-31. [Context Link]
34. Steinbrocker O, Argyros TG: The shoulder-hand syndrome: present status as a diagnostic and therapeutic entity. Med Clin North Am 1958;42:1533-1553. [Context Link]
35. Moberg E: The shoulder-hand-finger syndrome as a whole. Acta Chir Scand 1955;109:284-293. [Context Link]
38. Wagenaar RC, Meijer OG, van Wieringen PCW, et al: The functional recovery of stroke: a comparison between neuro-developmental treatment and the Brunnstrom method. Scand J Rehabil Med 1990;22:1-8. Bibliographic Links [Context Link]
39. Wagenaar RC, van Wieringen PCW, Netelenbos JB, Meijer OG, Kuik DJ: The transfer of scanning training effects on visual inattention after stroke: five single case studies. Disabil Rehabil 1992;14:51-60. [Context Link]
To obtain CME Category 1 credit, this educational activity must be completed and postmarked by December 31, 1998. Participants may read the articles and take the exams issue by issue or wait to study several issues together. After reading the three CME Articles in this issue, participants may complete the Self-Assessment Exam by answering the questions on the CME Answering Sheet and the Evaluation pages, which appear later in this section. Send the completed forms to: CME Department, Association of Academic Physiatrists, 5987 E. 71st Street, Suite 112, Indianapolis, IN 46220. Documentation can be received at the AAP National Office at any time throughout the year, and accurate records will be maintained for each participant. CME certificates are issued only once a year in January for the total number of credits earned during the prior year.
Key Words: Hemiplegia; Shoulder Pain; Rehabilitation; Triamcinolone