Incidence, Prevalence, and Risk Factors of Hemiplegic Shoulder Pain: A Systematic Review

The current systematic review aimed to investigate the incidence, prevalence, and risk factors causing hemiplegic shoulder pain (HSP) after stroke. Two independent authors screened titles and abstracts for the eligibility of the included studies in the electronic databases PubMed and Web of Science. Studies which reported the incidence, prevalence, and risk factors of HSP following stroke were included. The included studies were assessed using the Newcastle–Ottawa Scale for evaluating the quality of nonrandomized studies in meta-analyses. Eighteen studies were included in the final synthesis. In all studies, the number of patients ranged between 58 and 608, with the mean age ranging from 58.7 to 76 years. Seven included studies were rated as “good “quality, while one study rated “fair” and 10 studies rated “poor” quality. Eight studies reported incidence rate while 11 studies reported the prevalence of HSP following a stroke. The incidence of HSP was ranging from 10 to 22% in the metanalysis of the included studies. The prevalence of HSP was ranging from 22 to 47% in the metanalysis of the included studies. The most significant predictors of HSP were age, female gender, increased tone, sensory impairment, left-sided hemiparesis, hemorrhagic stroke, hemispatial neglect, positive past medical history, and poor National Institutes of Health Stroke Scale score. The incidence and prevalence of HSP after stroke vary considerably due to various factors. Knowledge of predictors is important to minimize the risk of developing HSP following a stroke.


Introduction
Hemiplegic shoulder pain (HSP) is a common and disabling complication following a stroke, and it may affect the quality of life [1]. It often occurs following two to three months of stroke [2,3]. Consequently, HSP may result in withdrawal from rehabilitation programs, longer hospital stays, reduced limb movement, and impaired quality of life [1]. Numerous causes have been implicated in developing HSP in stroke. This includes muscle flaccidity around the shoulder joint, shoulder subluxation, shoulder-hand syndrome, increased muscle tone, impingement syndrome, frozen shoulder, brachial plexus injury, and the thalamic syndrome [4,5].
There is an inconsistency in reporting incidence and prevalence of HSP following stroke. Some studies reported that the incidence of HSP ranges from 16% to 84% [6,7], while others reported a range of 65% to 70% [8,9]. In other studies, the incidence rate was ranging from 24% to 64% in those stroke populations who are admitted to an inpatient rehabilitation unit [10][11][12][13] and it varied from 9% to 40% in those who are not admitted [14][15][16].
A reduced arm motor function at the time of stroke or over a period during rehabilitation was found to be one of the risk factors for developing HSP following stroke [6,10]. Additionally, HSP often occurs and tends to be more severe among people with left-sided hemiplegia [15,17,18]. Reduced

Materials and Methods
This systematic review followed the guideline for the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) [26]. This systematic review was prospectively registered in PROSPERO (CRD42017077594) and available at http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID= CRD42017077594.

Search Strategy
The electronic databases PubMed, Web of Science, and Scopus were searched until November 19, 2019. Additionally, potential articles were searched manually from the reference list given in each article. The literature search in PubMed was conducted using the following keywords: ("Stroke" OR "Hemiplegia") AND ("Shoulder" OR "Arm" OR "Shoulder joint") AND ("Shoulder pain" OR "Pain") AND ("Prevalence" OR "Incidence") [ Table 1].

Eligibility Criteria
The current review included all the published literature that qualified the following criteria: studies included adults over 18 years of age with a history of stroke for more than one month; the outcomes of included studies should be pain in and around shoulder excluding ribs and the neck pain; all cross-sectional and longitudinal studies to assess the natural course of events after stroke and; studies describing potential risk factors such as demographic factors or impairments related to post-stroke shoulder pain that was evaluated after the incidence of stroke.
Studies were excluded if they were not published in English. Additionally, case reports and case series were also excluded as these types of studies might have a high potential bias. Furthermore, if the cause of shoulder pain was not secondary to stroke, those studies were also excluded (for instance, shoulder pain due to neck pathology).

Study Selection
Two independent authors screened titles and abstract for the eligibility of the included studies. Studies, which reported the incidence, prevalence, and risk factors of HPS following stroke were included. Any disagreements between two reviewers in the study selection were discussed and resolved by consensus between them.

Data Extraction and Assessment of the Risk of Bias
Two independent authors (SA and AA) completed the data extraction and assessment of the risk of bias, using structured formats. The important data extraction included the following items: author's name, setting, country of origin, sample size, target population, time since stroke (months), the average age of patients, outcome measurement, and study design. Quality assessment and risk of bias in the selected studies were appraised using the Newcastle-Ottawa Scale for evaluating the quality of nonrandomized studies in meta-analyses [27,28]. This evaluation tool has considered three factors (e.g., selection of exposed and non-exposed cohort, comparability of cohorts based on the design or analysis, and outcome based on reliability and validity of the scale, adequate follow-up and dropout rate) to appraise the quality of each included study [27,28]. The quality of each study was rated as good, fair, and poor by assigning stars in each domain as per given the guidelines of the Newcastle-Ottawa Scale [27,28]. A "good" quality score was given if the included study received 3 or 4 stars in selection domain,1 or 2 stars in comparability domain, and 2 or 3 stars in outcome domain. A "fair" quality score was given if the included study received 2 stars in selection domain,1 or 2 stars in comparability domain, and 2 or 3 stars in outcome domain. A "poor" quality score indicated 0 or 1 star(s) in selection domain, or 0 stars in comparability domain, or 0 or 1 star(s) in outcome domain ( Table 2). Any disagreements between two reviewers were discussed and resolved by consensus between them.

Summary Statistics and Synthesis of Results
Incidence and prevalence of HSP were reported from the included studies. Prevalence of HSP in the defined group was also reported if the information is available in the included studies. Most significant predictors of HSP after stroke were determined from the included studies. The Comprehensive Meta-Analysis software was used to conduct a meta-analysis. Meta-analysis was conducted for the prevalence, incidence, and risk factors of shoulder pain after stroke if at least two or more included studies with the adequate data for the analysis were available. The event rates for the prevalence and incidence estimates and the odd ratios (ORs) for the risk factors of shoulder pain after stroke were calculated. The effect size with 95% confidence intervals (CIs) for the prevalence, incidence, and risk factors of shoulder pain after stroke were calculated.  [14] Three major hospitals in the West of Scotland were participated. Two hospitals provided acute stroke patient care and one hospital provide acute stroke rehabilitation care after one-week discharge.
No non-exposed cohort

Study Selection
Based on the abstract and title search, initially, 390 articles were identified. After excluding duplicates (N = 251) and screening of abstracts, a total of 330 studies were not relevant to the current review, therefore excluded. Sixty full-text articles were included in the final screening, of which 42 articles failed to match the inclusion criteria, and were hence excluded. Henceforth, a total of 18 studies included in the final synthesis [3,6,9,11,14,16,[28][29][30][31][32][33][34][35][36][37][38][39]. Figure 1 shows the details of the study selection process and results of the literature search [26].
Incidence and prevalence of HSP were reported from the included studies. Prevalence of HSP in the defined group was also reported if the information is available in the included studies. Most significant predictors of HSP after stroke were determined from the included studies. The Comprehensive Meta-Analysis software was used to conduct a meta-analysis. Meta-analysis was conducted for the prevalence, incidence, and risk factors of shoulder pain after stroke if at least two or more included studies with the adequate data for the analysis were available. The event rates for the prevalence and incidence estimates and the odd ratios (ORs) for the risk factors of shoulder pain after stroke were calculated. The effect size with 95% confidence intervals (CIs) for the prevalence, incidence, and risk factors of shoulder pain after stroke were calculated.

Incidence and Prevalence of HSP after Stroke
The details of the incidence and prevalence of HSP after stroke is given in Table 4. The overall incidence of HSP was ranging from 1.6 to 40% in the included studies. The overall prevalence of HSP was ranging from 9.41 to 91.9% in the included studies. The incidence of HSP after a stroke at admission was reported in two studies [3,14]. Incidence at admission ranged from 9 to 10%. Three studies reported on the prevalence of HSP after a stroke at admission [9,11,39], ranging from 23 to 63.5%. The incidence of HSP after a stroke at 6 months was reported in three studies [14,16,33], ranging from 15 to 40%. Only one study reported the 6-month prevalence (42%) of HSP after a stroke [30]. Only one study reported the 12-month incidence (21%) of HSP after stroke [3]. Two studies reported the 12-month prevalence of HSP after stroke [30,35], ranging from 8.5 to 47%. Only one study reported the two-year incidence (15.1%) of HSP after a stroke [32]. Weekly point prevalence: 6% (5-7%) Discharge to 6-month incidence: 15% (9-21%) 6-months to 18-months incidence: 11% (6-16%) 18-months to 30-months incidence: 12% (6-17%) Gamble et al. (2000) [29] Incidence at 2-week: 25%   Table 5 presented the risk factors of HSP used in the included studies. A variety of risk factors was assessed in the included studies [3,6,9,11,14,16,29,31,32,34,37,39]. The most significant risk factors of HSP were age, arm weakness, sensory impairment, abnormal shoulder joint examination, average depression score, functional status, self-perceived ill health, subluxation, pathogenesis, left-sided hemiparesis, and prior history of shoulder pain. Six studies did not assess the risk factors of HSP after stroke [30,33,35,36,38,40].

Meta-Analysis
Of the 10 included studies with adequate prevalence data, three estimated event rates for the prevalence at admission, two estimated event rates for the prevalence at 4-month, 12-month, and 16-month, while four estimated event rates for an overall prevalence of shoulder pain after stroke. Figure 2 indicates a forest plot of prevalence rates with effect sizes and 95% CIs. Of the five included studies with adequate incidence data, two estimated event rates of the incidence at admission, while three estimated event rates of incidence at 6-month of shoulder pain after stroke. Figure 3 indicates a forest plot of incidence rates with effect sizes and 95% CIs. Of the eight included studies with adequate risk estimate data, three studies estimated ORs stratified by gender, two stratified by tone, sensation, affected body side, hemispatial neglect, and National Institutes of Health Stroke Scale Score, while three stratified based on type of strokes and seven stratified based on past medical history.        Table 5.

Discussion
Many reviews on HSP have been published in recent decades [41][42][43][44][45][46]. However, these reviews focus on the cause and treatment of HSP after stroke. As per our knowledge, this is the first systematic review, presenting the results of 18 studies on the incidence and prevalence of HSP after stroke. Additionally, the present review assessed the risk factors causing HSP after stroke. In the current review, a total of 5086 patients with a mean age ranging from 58.7 to 76 years were included. Seven included studies were rated as "good" quality, while one study rated "fair", and 10 studies rated "poor" quality. Eleven studies reported the prevalence, while eight studies reported the incidence rate of HSP following a stroke. The incidence of HSP was ranging from 10 to 22% in the metanalysis of the included studies. The prevalence of HSP was ranging from 22 to 47% in the metanalysis of the included studies.

Discussion
Many reviews on HSP have been published in recent decades [41][42][43][44][45][46]. However, these reviews focus on the cause and treatment of HSP after stroke. As per our knowledge, this is the first systematic review, presenting the results of 18 studies on the incidence and prevalence of HSP after stroke. Additionally, the present review assessed the risk factors causing HSP after stroke. In the current review, a total of 5086 patients with a mean age ranging from 58.7 to 76 years were included. Seven included studies were rated as "good" quality, while one study rated "fair", and 10 studies rated "poor" quality. Eleven studies reported the prevalence, while eight studies reported the incidence rate of HSP following a stroke. The incidence of HSP was ranging from 10 to 22% in the metanalysis of the included studies. The prevalence of HSP was ranging from 22 to 47% in the metanalysis of the included studies.
Previous studies reported the prevalence of shoulder pain after stroke between 21% and 84% [6,12,14,16,30], while others reported 5% and 84% [47,48]. The wide variation in the prevalence of HSP reflects the lack of proper definition or an inconsistency in the quality of care of these patients among diverse populations [49,50]. Incidence at admission was approximately 10% in the present review. The incidence of HSP after a stroke at 6 months was 22% in the present review. Wanklyn et al. [51] reported that about 63% of the patients developed HSP in the first six months after stroke. Other studies reported that the HSP usually develops within 2-3 months following the stroke onset [2,6,16]. In the current review, 12 months prevalence of HSP after a stroke was 39%. A previous study reported the prevalence of HSP about 32% within the 12 months after a stroke [52]. Another study reported the prevalence of HSP about 34% at 12 months after a stroke [36].
In the current review, the most significant identified predictors of HSP were age (younger than 70 years), female gender, increased tone, sensory impairment, left-sided hemiparesis, hemorrhagic stroke, hemispatial neglect, positive past medical history, and poor National Institutes of Health Stroke Scale score. In the previous study, age was not directly linked to the development of HSP, but older people could have preexisting problems that affect shoulder pain [53]. Another study reported various risk factors, including impaired motor control, reduced proprioception, sensory impairment, spasticity of the elbow flexor muscles, limited range of motion (ROM) of the shoulder joint, and type 2 diabetes mellitus [54]. Barlak et al. [13] reported a significant correlation between HSP and complex regional pain syndrome and adhesive capsulitis, but there was no correlation found between the HSP and the grade of subluxation, impingement syndrome or spasticity. Similarly, other studies also reported many risk factors for shoulder pain, and the severity of motor impairment is one of the most important risk factors identified [6,15,16,40,51,54].
During the recovery from stroke, muscle spasticity of the upper extremities is thought to cause shoulder subluxation and limited ROM, resulting in the development of shoulder pain [55]. Another important cause of HSP is frozen shoulder (Adhesive capsulitis), which is indicated by a limited shoulder ROM, with a capsular type of restriction [56][57][58]. However, other factors could also cause a reduced shoulder ROM in stroke patients without frozen shoulder. Mao et al. [59] identified a prolonged, shortened position was one of the causes of developing soft tissues contracture around the shoulder such as muscles, tendons, and ligaments. The current review was not able to identify a single pathology causing shoulder pain in patients with stroke. This area of research should be explored in the future investigations.
The major strength of this review was the inclusion of a metanalysis, which identified many risk factors causing shoulder pain after stroke. Additionally, most of the included studies were categorized as a prospective observational study, which further enhances the validity of the results. The present study had some potential limitations as well. Some of the included studies did not use a valid and reliable outcome measure to assess symptoms of shoulder pain, which could affect the validity of the results. Additionally, most of the included studies did not report the exact cause of shoulder pain following stroke. Furthermore, variations in the participants' characteristics in the included studies prevent to make any causal relationships between shoulder pain and predictor measures.

Conclusions
The incidence and prevalence of shoulder pain after stroke vary considerably due to various factors. Knowledge of predictors is important to minimize the risk of developing shoulder pain following a stroke.
Author Contributions: S.A.: the corresponding author; participated in the study design and concept; wrote the review protocol; contributed to the article review, selection and data extraction; contributed to the analysis, interpretation, and presentation of data; and drafted the manuscript. A.A.: participated in the study design and concept; participated in data extraction; contributed to the analysis and provided a critical revision of the manuscript. Both authors approved the final version of the manuscript. All authors have read and agreed to the published version of the manuscript.