Risk factors associated with pain while sleeping on the affected shoulder after primary reverse shoulder arthroplasty
Article information
Abstract
Background
The purpose of this study was to identify risk factors of pain while lying on the operative shoulder following primary reverse total shoulder arthroplasty (rTSA).
Methods
Patients who underwent primary rTSA with available (1-year) follow-up data were retrospectively identified. Demographics, diagnosis, comorbidities, implant configuration, surgical information, and pain scores (including preoperative, postoperative and improvement in pain) were assessed while lying on the operated shoulder. To categorize preoperative pain while lying on the operative shoulder, cohorts were defined based on above or below the average pain level. Postoperative pain and improvement in pain were defined based on the following thresholds: patient acceptable symptomatic state (PASS), minimal clinically important difference (MCID), and substantial clinical benefit (SCB). The PASS was defined as the 75th percentile of pain scores in patients with high satisfaction ratings after rTSA, while MCID and SCB were calculated as the difference in average pain improvement in patients with high versus low satisfaction rates postoperatively. Univariate and multivariate logistic regression analyses were conducted.
Results
A total of 4,235 patients who underwent rTSA were included. Previous shoulder surgery, tobacco use, and preoperative pain lying on the operative shoulder failed to achieve threshold values. Subscapularis repair was associated with an improved ability to achieve the thresholds.
Conclusions
Tobacco use, higher preoperative pain levels, and previous shoulder surgery were negatively associated with satisfactory improvement in pain while lying on the postoperative shoulder. In contrast, subscapularis repair was associated with clinically significant improvements. Given that postoperative pain when lying on the operative side is a frequent preoperative question, understanding these influencing factors is useful when counseling patients on postoperative expectations.
Level of evidence
III.
INTRODUCTION
Pain at night that causes a sleep disturbance is a common finding in patients with shoulder pathology and is reported in as many as 90% of cases [1-4]. Cho et al. [1] demonstrated important correlations of shoulder pain with patient well-being and sleep. The authors evaluated 130 patients with more than 3 months of shoulder pain and found that 74% of them had pathological findings on shoulder magnetic resonance imaging. These patients were compared to a control group without shoulder pain and demonstrated higher prevalence of depression, anxiety, and sleep disturbances [1].
One of the most common postoperative expectations after shoulder arthroplasty is the ability to sleep through the night [5]. Frequently, patients who undergo shoulder surgery for treatment of glenohumeral osteoarthritis or rotator cuff tears expect significant improvements in sleep [6]. Although many patients experience improvement in sleep following shoulder arthroplasty, night pain remains a common complaint in the postoperative period after anatomical total shoulder arthroplasty (aTSA) [7]. Vegas et al. [8] studied 472 patients who had undergone reverse total shoulder arthroplasty (rTSA) more than 2 years prior. While the majority of these patients reported improvement in overall comfort while sleeping, 36.8% were unable to sleep on the side of the operative shoulder [8]. As shown in Vegas et al. [8], a significant number of patients continue to complain of discomfort while sleeping after shoulder arthroplasty. While there is a known correlation between shoulder pain and sleep disturbance [1], there are limited studies regarding the causes of pain while sleeping after rTSA.
The purpose of this study was to determine whether preoperative patient-related risk factors correlate with persistent difficulty lying on the side of the operative shoulder following primary rTSA. The authors hypothesized that there would be demographic factors and/or surgical details that would be associated with the ability to sleep on the side of the operative shoulder after rTSA.
METHODS
After receiving an exemption from the Institutional Review Board (IRB) at Augusta University (No. #1791666), patient data from a prospectively collected multi-center, international database of a single shoulder prosthesis (Equinoxe; Exactech Inc.) were reviewed retrospectively for procedures performed between October 2007 and October 2022. Informed consent was waived by the IRB for this study due to its retrospective nature, but patients provided consent for data use prior to enrollment. The following data were collected via standardized forms according to an IRB-approved protocol: demographics, diagnoses, comorbidities, implant/surgical information, pre- and postoperative outcomes, and patient satisfaction. Patients undergoing primary rTSA with 1 year of available follow-up data, defined as 9–15 months, were eligible for inclusion in the study. The exclusion criteria were revision surgery, surgery performed for fracture, tumor resection, osteomyelitis, and preoperative or postoperative visual analog scale (VAS) pain lying on the operative shoulder.
Relationships of demographics, diagnosis, comorbidities, and surgical parameters with preoperative, postoperative, and improvement in pain lying on the operative shoulder were examined. Preoperative VAS pain lying on the operative side was averaged for the entire cohort to create two groups: above or below average pain. For postoperative pain, cohorts were defined as achieving or not achieving the patient acceptable symptomatic state (PASS) threshold. The PASS threshold was defined as the 75th percentile of the postoperative pain score among patients rating their shoulders “much better” or “better” following surgery. For improvement in pain, cohorts were defined as achieving or not achieving the minimal clinically important difference (MCID) and the substantial clinical benefit (SCB) thresholds. The MCID and SCB were calculated as the difference in the average improvement in pain lying on the operative shoulder for patients rating their shoulders “better” or “much better,” respectively, versus those rating their shoulders “unchanged” or “worse.” Cohort status for each threshold was determined independently of the achievement of other thresholds.
Statistical Analysis
Univariate analysis was conducted on each independent variable in the above average and the below average preoperative pain groups utilizing two-tailed, unpaired Welch’s t-test for continuous variables or Fisher’s exact test for count data. Variables identified as significantly different between the groups were then included in a multivariate logistic regression analysis to determine the odds ratio (OR), calculated with 95% CI, and the P-value of each variable. For postoperative and improvement in pain, the variables identified as significantly different between those achieving or not achieving PASS, MCID, and SCB in univariate analyses were included in the multivariate logistic regression analyses. All analyses were conducted in R version 4.3.1 (R Foundation for Statistical Computing) [9] with a significance level of P<0.05.
RESULTS
Of an initial cohort of 5,824 rTSAs, 605 were excluded for incomplete data, 580 were excluded for revision indication, 393 were excluded for fracture indication, six were excluded for a diagnosed infection, and five were excluded based on the use of an endoprosthesis. This left a cohort of 4,235 rTSAs performed between October 2007 and October 2022.
Among these patients, there was a mean age of 72.0±7.6 years. The group included 2,510 females, 1,714 males, and 11 unspecified. The average preoperative pain lying on the operative shoulder was 7.0, with 47.2% of the total cohort having average or below average pain. The mean preoperative pain scores were 4.6±2.1 and 9.1±0.9 for the average or below and above average cohorts, respectively (P<0.001). The differences in demographics, diagnosis, and comorbidities are shown in Table 1. On univariate analysis, the above average pain cohort had significantly lower age at surgery (P=0.007); more females (P<0.001); higher body mass index (BMI) (P<0.001); greater frequencies of osteoarthritis diagnosis (P=0.003), and osteonecrosis diagnosis (P=0.043), diabetes (P=0.001), tobacco use (P<0.001), previous injections (P=0.007), and analgesic use (P=0.046); and lower frequency of cuff tear arthropathy diagnosis (P=0.005) compared to the average or below average pain cohort. Upon including these parameters in a multivariate logistic regression model, the following variables remained significantly associated with above average preoperative pain lying on the operative shoulder: sex (P<0.001, OR=1.65), BMI (P<0.001, OR=1.03), osteoarthritis diagnosis (P=0.014, OR=1.21), diabetes diagnosis (P=0.022, OR=1.23), tobacco use (P=0.001, OR=1.60), and previous injections (P=0.021, OR=1.16) (Table 1).
Univariate and multivariate logistic regression analysis of demographics, diagnosis, and comorbidities for rTSAs with average or below and above average preoperative pain lying on the operated shoulder
Regarding postoperative and improvement in pain lying on the operative shoulder, cohorts were formed based on achieving or not achieving the PASS, MCID, or SCB thresholds, which were determined to be 2.0, 1.8, and 3.9, respectively. These respective thresholds were achieved by 76.7%, 86.6%, and 71.0% of the total cohort. The average postoperative pain was 0.5±0.7 in the cohort achieving PASS versus 5.4±2.2 (P<0.001) in the cohort not achieving PASS, while improvement in pain was 6.2±2.9 versus 2.6±2.8 (P<0.001). Similarly, for MCID, the average postoperative pain was 1.1±1.8 versus 4.5±3.7 (P<0.001) in the achieving and non-achieving cohorts, respectively, while improvement was 6.3±2.5 versus 0.3±1.5 (P<0.001). The cohorts achieving and not achieving SCB had postoperative pain of 0.9±1.4 versus 3.3±3.4 (P<0.001) and improvement of 7.1±1.9 versus 1.2±1.8 (P<0.001).
Univariate analysis identified several differences in demographics, diagnosis, comorbidities, preoperative pain when lying on the operative shoulder, and surgical factors between the cohorts achieving or not achieving each threshold (Table 2). Universally, groups not achieving the thresholds had a significantly lower frequency of osteoarthritis diagnosis (P<0.001 for PASS, MCID, SCB), no comorbidities (P=0.028, P=0.007, P=0.008 for PASS, MCID, SCB), and subscapularis repair (P<0.001 for PASS and MCID, P=0.001 for SCB) and significantly higher frequency of previous shoulder surgery (P<0.001 for PASS, MCID, and SCB), cuff tear arthropathy diagnosis (P=0.002 for PASS and MCID and P<0.001 for SCB), heart disease (P=0.005 for PASS and P=0.001 for MCID and SCB), and tobacco use (P<0.001, P=0.018, P=0.033 for PASS, MCID, SCB). Other variables that were universally significant were sex and preoperative pain when lying on the operative shoulder. However, the cohort not achieving the PASS threshold had significantly more females (P=0.026) and higher preoperative pain (P<0.001), while the cohorts not achieving MCID and SCB had significantly fewer females (P<0.001) and lower preoperative pain (P<0.001). Variables only significant in the PASS analysis were younger age (P<0.001), more frequent diabetes diagnosis (P<0.001), less combined liner/tray offset (P=0.026), and less frequent expanded glenosphere usage (P=0.011) in the not achieving cohort. Variables only significant in the MCID and/or SCB analyses included glenosphere diameter (P<0.001 for MCID and SCB) and number of baseplate screws (P<0.001 for SCB).
Univariate analyses of rTSA cohorts achieving or not achieving PASS, MCID, and SCB thresholds for pain lying on the operative shoulder at 1-year follow-up
On multivariate analysis (Table 3), only four factors remained universally significantly associated with failure to achieve PASS, MCID, or SCB: frequency of previous shoulder surgery (PASS: P<0.001, OR=1.97; MCID: P<0.001, OR=1.83; SCB: P<0.001, OR=1.97), frequency of tobacco use (PASS: P=0.006, OR=1.55; MCID: P=0.019, OR=1.63; SCB: P=0.003, OR=1.73), preoperative pain when lying on the affected shoulder (PASS: P<0.001, OR=1.21; MCID: P<0.001, OR=0.68; SCB: P<0.001, OR=0.58), and frequency of subscapularis repair (PASS: P=0.004, OR=0.79; MCID: P=0.005, OR=0.74; SCB: P=0.002, OR=0.75). Additional significant factors for failure to achieve PASS were age (P<0.001, OR=0.98) and frequency of heart disease (P=0.001, OR=1.46), diabetes (P=0.010, OR=1.32), and expanded glenosphere usage (P=0.002, OR=0.57). The only additional significant factors associated with failure to achieve MCID or SCB were frequency of osteoarthritis diagnosis (P=0.037, OR=0.80), presence of comorbidities (P=0.017, OR=0.78), and number of baseplate screws (P=0.042, OR=1.12) in the SCB analysis.
Multivariate logistic regression analyses for rTSA cohorts achieving or not achieving PASS, MCID, and SCB thresholds for pain lying on the operative shoulder at 1-year follow-up
DISCUSSION
This series sought to analyze factors that might affect patients while lying on their operative side after primary rTSA with 1 year of follow-up. A multivariate analysis of 4,235 patients revealed three factors associated with failure to achieve clinically relevant thresholds for postoperative pain (PASS) and improvement in pain while sleeping on the affected shoulder (MCID and SCB): higher preoperative pain while lying on the symptomatic shoulder, prior shoulder surgery, and tobacco use. Failure to achieve PASS was associated with higher preoperative pain, while failure to achieve MCID and SCB was associated with lower preoperative pain. Failure to achieve all three thresholds was linked to higher rates of tobacco use and prior shoulder surgery. The only modifiable intraoperative factor associated with failure to achieve postoperative and improvement thresholds for pain while sleeping on the operative shoulder at 1 year postoperatively was not repairing the subscapularis.
Chronic pain leading to the use of opioids prior to shoulder arthroplasty has been shown to lead to inferior results after rTSA compared to patients who are not on opioids [10]. The influence of preoperative opioids on postoperative pain is an issue with serious implications for patients; up to 40% of patients undergoing either aTSA or rTSA had been prescribed narcotics within 3 months prior to their surgery [11]. A recent systematic review found that the use of preoperative opioids was an independent predictor of pain after rTSA [12]. Evaluating factors that affect postoperative pain is further complicated by the consistent postoperative pain (VAS ≥2) present in 18.9% of patients undergoing aTSA or rTSA [13]. In a large series of 4,158 patients with rTSA, Parada et al. [14] also showed that pain was the third most common complication after surgery. Studies that examine pain as a postoperative outcome will be influenced by the large percentage of patients who experience postoperative pain after shoulder arthroplasty, and that a significant number of patients utilize opioid medications in the preoperative period. Our study corroborated this issue with the association of preoperative pain as an independent risk factor for postoperative pain while lying on the side after rTSA.
Previous studies have shown that non-arthroplasty procedures on the affected arm prior to arthroplasty result in significantly higher VAS and less improvement in pain, as well as a higher rate of complications after shoulder arthroplasty [15-17]. Frank et al. [15] demonstrated that patients with prior surgery had mean postoperative VAS scores of 2.0, while patients without previous surgery had mean VAS scores of 0.9. It is difficult to establish which prior surgeries affect the ability for a patient to sleep on their shoulder after rTSA. Conflicting data exist about the effect of rotator cuff repair prior to rTSA; several studies have reported worsened pain [16,18], while others have not found a negative effect [19,20]. The current study reiterates an association with previous surgery and continued shoulder pain, showing that patients who underwent previous surgery were less likely to achieve PASS and improvement in MCID and SCB than were those who did not.
Previous literature has identified an association between tobacco use and diminished postoperative pain improvements following shoulder arthroplasty [12,21]. Walters et al. [21] demonstrated that current smokers had a mean postoperative VAS of 2.5 versus 1.8 in non-smokers and 1.0 in previous smokers. Wells et al. [22] demonstrated a decrease in VAS score from 7.1 to 4.3 in a smoker group and 5.8 to 1.8 and 1.5 in non-smoker and previous smoker groups, respectively. This study also demonstrates the negative effects of tobacco use and the ability to achieve an optimal outcome. Other than previous surgery, tobacco use was the only other risk fracture for failure to achieve all three thresholds.
Depending on the quality of the rotator cuff and implant design, repairing the subscapularis is not always feasible. However, the decision to repair it remains a topic of significant controversy. In a series of patients who underwent rTSA using the same platform system studied herein, Friedman et al. [23] evaluated outcomes with an average follow-up of 37 months. A total of 341 patients underwent subscapularis repair, while 251 did not receive repair. The authors did not demonstrate a significant difference in outcome scores or complication rates between the groups. However, there was a small, yet statistically significant, improvement in certain motion planes.
Multiple other studies of lateralized humeral designs have demonstrated limited to no improvement in functional outcome scores with subscapularis repair [23,24]. Franceschetti et al. [25] found improved postoperative abduction and internal rotation with the addition of a subscapularis repair in lateralized humerus rTSA; however, they found no difference in other planes of motion or improvement in pain. When considering the effect on a medialized humeral implant, Mocini et al. [26] found improved functional outcomes and pain scores with subscapularis repair, but no difference with subscapularis repair in lateralized humeral designs. All patients in the current study were treated with medialized glenoid/lateralized humerus components, and variations in the individual humeral-side and glenoid-side components within this system did not affect pain while lying on the shoulder. Despite the controversy of subscapularis repair as it relates to outcomes and function, this study demonstrated that subscapularis repair was the only intraoperative finding associated with an improvement in pain levels while sleeping on the operative shoulder at 1 year postoperatively.
Vegas et al. [8] sought to determine the effects of shoulder arthroplasty on sleep improvement. The authors found that patients who underwent aTSA compared to those who underwent rTSA reported significantly less difficulty sleeping on the affected side 1 year postoperatively. Patients diagnosed with osteoarthritis were more likely to achieve PASS, MCID, and SCB. In contrast, those with cuff tear arthropathy were more likely to have higher preoperative shoulder dysfunction and an inability to sleep on the affected side. This finding suggests that a retracted rotator cuff etiology may be the cause of sleep disturbances, and that repairing the previously intact subscapularis may help prevent related discomfort. However, it remains unclear why improved pain relief was observed with subscapularis repair and a lateralized humerus design. This improvement may reflect enhanced prosthesis stability with subscapularis repair or the presence of an additional soft tissue bolster that provides greater comfort. Further research is needed to determine the source of this improvement before any general recommendations can be made.
This study has several limitations, starting with the variability in arm positioning during sleep. Certain patients never sleep on their side, and even those who do can place the arm in a variety of positions that may not be equal in generating discomfort. Therefore, it is possible that pain while lying on the side may not be a direct corollary to sleep comfort. Outcomes were examined at 1 year postoperatively, because past studies have shown that sleep improvement plateaus by 3 months and sleeping on the affected side plateaus by 6 months after rTSA [8]. However, it is possible that some patients may have improved after the 1 year mark. Although analgesic use was accounted for and showed no significant differences in usage between cohorts, this study was unable to specifically account for opioid use in patients, potentially confounding our results. Additionally, this study did not distinguish between current and former tobacco use, which prevented the authors from evaluating how smoking cessation might be associated with an improved ability to sleep on the affected shoulder postoperatively.
Sleep and sleep quality exist on a spectrum, such as the ability to fall asleep, maintain sleep, depth of sleep, restfulness, etc. [27]. However, in this study, patients were not asked to stratify their sleep by timing or any other factors aside from their pain lying on the affected side. Furthermore, all previous shoulder surgeries were grouped together in our study. Certainly, we would expect that certain surgeries prior to rTSA would have more of an impact than others; future research should seek to evaluate the effects of previous surgery types.
Additionally, the subscapularis repair technique was determined at the discretion of the operating surgeon and potentially introduces variability. The integrity of the repair was not assessed via advanced imaging at the 1 year mark in this study. With a lateralized humeral system, attempting to repair the subscapularis anatomically would result in lateralization and distalization of the insertion compared to its preoperative state. Anecdotally, most surgeons do not attempt a complete anatomic restoration of the footprint and may instead opt to place the tendon more medially. This lack of standardization and validation of the repair should be considered when interpreting subscapularis repair as the lone positive influence in this study. Additionally, it is unknown if these results are generalized to other systems with differing design philosophies. There may be a link to the amount of soft tissue coverage at the level of the joint. Future studies are planned to analyze the volume of the deltoid as well as the volume of the subcutaneous tissue to determine if this “cushion” affects pain while sleeping.
CONCLUSIONS
This multi-center study of rTSA with a single prosthesis demonstrated that multivariate analysis identified three factors linked to the inability to PASS and improvement in sleep-related pain on the affected shoulder (MCID and SCB): higher preoperative pain when lying on the symptomatic shoulder, previous shoulder surgery, and tobacco use. Patients with greater preoperative pain were more likely to fail to achieve PASS, whereas those with lower preoperative pain were more likely to fall short of MCID and SCB. The inability to reach all three thresholds was more common among patients with a history of tobacco use and prior shoulder surgery. The only modifiable intraoperative factor associated with failing to achieve postoperative and improvement thresholds for pain while sleeping on the operative shoulder at 1 year postoperatively was the decision not to repair the subscapularis. This information is useful in counseling patients on the likelihood of meeting their expectations and provides potential new insight into the role of subscapularis repair in rTSA.
Notes
Author contributions
Conceptualization: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Data curation: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Formal analysis: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Investigation: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Methodology: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Project administration: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Supervision: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Validation: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Visualization: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Writing – original draft: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. Writing – review & editing: BLH, JLL, HH, BGG, KAH, EPB, AEB, JAE, SAP. All authors read and agreed to the published version of the manuscript.
Conflict of interest
SAP is a consultant and receives institutional research support from Exactech Inc. (Gainesville, FL, USA). He is a consultant for Arthrex Inc. and is a board or committee member for American Academy of Orthopaedic Surgeons and American Shoulder and Elbow Surgeons. These affiliations have no role in the decision to publish this article. No other potential conflicts of interest relevant to this article were reported.
Funding
None.
Data availability
Contact the corresponding author for data availability.
Acknowledgments
None.