Three-dimensional kinematic evaluation of scapulohumeral rhythm after reverse shoulder arthroplasty: a systematic review and meta-analysis

Felipe F Gonzalez; Raphael Fonseca; Gustavo Leporace; Rafael Pitta; Marcos N Giordano; Jorge Chahla; Leonardo Metsavaht

doi:10.1016/j.xrrt.2021.10.009

. 2021 Dec 6;2(1):8–16. doi: 10.1016/j.xrrt.2021.10.009

Three-dimensional kinematic evaluation of scapulohumeral rhythm after reverse shoulder arthroplasty: a systematic review and meta-analysis

Felipe F Gonzalez ^a,^b,^c,^∗, Raphael Fonseca ^a, Gustavo Leporace ^b,^c, Rafael Pitta ^a, Marcos N Giordano ^a, Jorge Chahla ^d, Leonardo Metsavaht ^b,^c

PMCID: PMC10426534 PMID: 37588296

Abstract

Background

The movement of the arm relative to the trunk results from 3-dimensional (3D) coordinated movements of the glenohumeral (GH) and scapulothoracic (ST) joints and dictates the scapulohumeral rhythm (SHR). Alterations in SHR increase joint overload and may lead to low functional scores, pain, and failures in patients undergoing reverse total shoulder arthroplasty (RSA). The goal of this systematic review and meta-analysis was to examine 3D SHR kinematics after RSA and compare it to that of asymptomatic shoulders.

Methods

A systematic review and meta-analysis of articles in English were performed using PubMed, Embase, Cochrane Library, and SciELO. Additional studies were identified by searching bibliographies. Search terms included “Reverse shoulder arthroplasty”, “3D”, and “scapula”. It was selected cross-sectional studies that reported SHR with 3D motion analysis systems in patients who underwent RSA and asymptomatic controls. Two authors independently performed the extraction of articles using predefined data fields, including study quality indicators.

Results

Data from four studies were included in quantitative analysis, totaling 48 shoulders with RSA and 63 asymptomatic shoulders. Pooled analyses were based on random-effects model (DerSimonian-Laird). A statistically smaller SHR ratio was observed in the RSA group than that in the control group (P < .00001), meaning a greater contribution of ST joint in relation to GH joint for arm elevation. The standardized mean difference was −1.16 (95% confidence interval: −1.64, −0.67). A sensitivity analysis with three more studies that had imputed data on control group did not change the direction of the effect. The standardized mean difference on sensitivity analysis was −0.60 (P = .03; 95% confidence interval: −1.13, −0.06). It was detected as “not important heterogeneity” within the comparison (I²: 22%). Chi-square was not statistically significant (Chi²: 3.85), and I² was 22%. Tau² was not zero (Tau²: 0.05). Sensitivity analysis showed an I² of 74%, which might represent substantial heterogeneity, Chi-square was not statistically significant (Chi²: 23.01), and Tau² was not zero (Tau²: 0.37).

Conclusion

This study found that RSA shoulders have an increased contribution of ST joint during arm elevation, compared with asymptomatic shoulders. More movement in ST joint in proportion to GH joint increases GH joint contact forces, which could lead to component loosening or other complications. Further studies should address the clinical implications of this kinematic finding.

Keywords: Scapulohumeral rhythm, 3D motion analysis, Reverse shoulder arthroplasty, Contact forces, Complication, Kinematic

Since its conception, reverse total shoulder arthroplasty (RSA) has gained significant popularity, especially due to the substantial clinical and functional improvement in patients with rotator cuff arthropathy.⁴⁴ Designed in 1993, it was developed as an alternative to anatomic shoulder replacement in patients with massive rotator cuff tears.³⁵ Recently, RSA indications have been expanded to acute three- or four-part proximal humerus fractures in the elderly, primary osteoarthritis, chronic anterior dislocation, and failed anatomic total shoulder arthroplasty.⁴¹ As a consequence of its worldwide use, there has been an increase in the number of complications and reoperations.³⁶

Multiple RSA complications are described in the literature. Complications on the glenoid side, humeral side, implant issues, and muscular complications, account for more than twenty different types of complications reported.³⁵ Often, they are misunderstood and misdiagnosed, and this can lead to many reoperations on the same patient.⁵ Moreover, instability or dislocation of implants is frequently attributed to an imbalance of the muscles around the reconstructed joint⁵; however, those complications might be attributed to kinematic differences between the anatomic and prosthetic joints. A comprehensive understanding of RSA biomechanics is crucial to identify how complications can be avoided and/or treated in the event they are already present.

Recent studies have focused on the contribution of the scapulothoracic (ST) joint motion to total humerothoracic (HT) motion or arm motion, which is composed of ST and glenohumeral (GH) joint motion.²⁰^,²¹^,³⁴^,⁴³ Scapulohumeral rhythm (SHR) has been used to quantify the relative motion of the scapula and humerus to total shoulder motion and is defined by GH elevation divided by ST superior rotation motion (Fig. 1).²^,⁷^,¹⁹^,³⁴ SHR has been widely used as a sensitive measure of shoulder dysfunction and can provide important information regarding current or future shoulder injuries.³^,²⁴^,²⁷^,⁴²

Scapulohumeral rhythm (SHR) during shoulder abduction. Note: SHR is the relation between the movement of GH and ST joints during arm elevation and the movement of the humerus relative to the trunk. *Dorsal view* of left shoulder girdle. (A) Shoulder abduction at 0°; (B) shoulder abduction at 90°; (C) maximum shoulder abduction. : humerothoracic (HT) elevation; : scapulothoracic (ST) superior rotation.

Inline graphic — Scapulohumeral rhythm (SHR) during shoulder abduction. Note: SHR is the relation between the movement of GH and ST joints during arm elevation and the movement of the humerus relative to the trunk. *Dorsal view* of left shoulder girdle. (A) Shoulder abduction at 0°; (B) shoulder abduction at 90°; (C) maximum shoulder abduction. : humerothoracic (HT) elevation; : scapulothoracic (ST) superior rotation.

Many of the studies analyzing the HT motion after RSA use 2D methods, 3D computed tomographic models, or cadaveric biomechanical tests to quantify the range of motion.⁶^,⁸^,¹⁸^,³¹^,³³^,⁴³ Two-dimensional methods to quantify GH joint motion are limited and may result in inaccurate or incomplete descriptions of joint position. Moreover, 3-dimensional (3D) computed tomographic models or cadaveric biomechanical results may not be extrapolated to an in vivo setting. On the other hand, 3D skin-based motion systems have shown to be accurate and reliable for scapular and humeral kinematics in vivo.¹⁴^,¹⁶^,²²^,²⁹ Although studies that used 3D motion analysis to measure shoulder function after RSA show altered SHR, not all of them reported SHR as the primary outcome or compared it to a control group.¹^,²^,⁴^,⁷ To the best of our knowledge, there is no systematic review that addresses these methodological issues. Therefore, an accurate understanding of 3D SHR on patients submitted to RSA is still lacking. The goal of this systematic review and meta-analysis was to overcome these methodological limitations and examine 3D SHR kinematics after RSA and compare it to that of asymptomatic shoulders.

Materials and methods

Eligibility criteria

Types of studies

Studies evaluating SHR in the scapular abduction plane with 3D motion analysis systems in patients who underwent RSA and in asymptomatic shoulders were included. No design, publication date, and publication status restrictions were imposed. The search was restricted to studies in English language.

Types of participants

Participants of any age, any gender, any race, any socioeconomic status, any functional level, at any time after surgery, and with any indication for RSA were included in this analysis. There were no exclusion criteria based on participants' characteristics.

Types of exposure

This review was limited to studies looking for SHR in patients with RSA.

Types of outcome measure

The primary outcome measure is a performance-based outcome, the SHR.

Information sources

Studies were primarily identified by research in electronic databases. Secondary identification was by expert consultation and scanning reference lists in similar articles. This search was applied to PubMed, Embase, Cochrane Library, and SciELO in January 17, 2020, at 2 pm.

Search strategy

We used the following search terms combined in each search database with technical adaptations of Boolean operators for each database: ((Shoulder Replacement∗) OR (Shoulder arthroplast∗) OR (Shoulder Prosthesis)) AND ((Range of Motion) OR (Range of Movement) OR Flexion OR Extension OR Abduction OR Adduction OR Rotation OR Sagittal OR Coronal OR Axial OR Biomechanics OR Kinematic∗ OR Scapul∗)) AND ((3-dimensional OR 3D OR three-dimensional)).

Study selection

Eligibility assessment was made by two investigators following the reporting guidelines for meta-analysis Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)²³ and Meta-analyses of Observational Studies in Epidemiology (MOOSE).³⁸

Data collection process

We have developed a data extraction sheet containing relevant information. The extraction sheet was refined after the initial round. A total of 3 rounds of research looking for tables, graphs, and text were carried out by 2 investigators.

Risk of bias in individual studies

We assessed each study quality using National Institutes of Health Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies.³⁹ Two investigators, in consensus, rated each study after full-text analysis. No clinical trial was included in this analysis because of eligibility criteria.

Summary measures

The standard mean difference in SHR ratio comparing RSA and control group was the primary measure of treatment effect, using a random-effects model. Quantitative analyses were performed with all included studies. A standard mean difference was calculated instead of mean difference because different equipment and methodologies were used between studies. P value and 95% confidence intervals were calculated. P value was considered statistically significant when ≤.05.

Planned methods of analysis

In two studies, SHR was reported in different parts of the total arm elevation.¹⁹^,³⁴ Mean and standard deviation (SD) were calculated for total arm elevation using the following formula: $\frac{D z}{z} = \sqrt{{(\frac{D a}{a})}^{2} + {(\frac{D b}{b})}^{2} + {(\frac{D c}{c})}^{2}}$ , divided by appropriate constants. Consider mean as “a”, “b”, or “c” and its associated SD or error as Da, Db, or Dc, respectively.¹⁵

One study reported range instead of SD.²¹ SD was estimated using the following formula: SD = (maximum value − minimum value)/4.¹⁵

One study did not report SHR of comparison group on text.⁷ We retrieved it from one figure. The same study did not report SDs of RSA and comparison group. We retrieved it from calculations following Cochrane Handbook for Systematic Reviews of Interventions¹⁵ recommendations using available data.

Heterogeneity was tested with Chi² test and I² statistics. Variance estimation among the effects observed in different studies (between-study variance) was calculated with Tau².

The interpretation of heterogeneity followed Cochrane Handbook for Systematic Reviews of Interventions.¹⁵ Chi² test was interpreted as statistically significant if <0.05. I² statistics followed this guide for interpretation: 0% to 40%, might not be important; 30% to 60%, may represent moderate heterogeneity; 50% to 90%, may represent substantial heterogeneity; 75% to 100%, considerable heterogeneity.

Statistical analysis was performed with Review Manager software, version 5.4.

Sensitivity analysis

A sensitivity analysis was conducted to assess differences in the overall effect adding studies that had imputed data. Other analyses such as subgroup or meta-regression were not performed.

Among the studies included in the sensitivity analysis, one study did not report SHR, but GH elevation and ST upward rotation.¹ The SHR value was obtained by dividing GH by ST. Means and SDs were calculated using the following formulas: $z \pm D z = \frac{a \pm D a \cdot b \pm D b}{c \pm D c}$ and $\frac{D z}{z} = \sqrt{{(\frac{D a}{a})}^{2} + {(\frac{D b}{b})}^{2} + {(\frac{D c}{c})}^{2}}$ . Consider mean as “a”, “b”, or “c” and its associated SD or error as Da, Db, or Dc, respectively.¹⁵

Two studies included in sensitivity analysis did not have asymptomatic shoulders as the control group,¹^,² and one study did not have the control group.⁴ Means and SDs were imputed as “worst-case scenario” based on summary measures reported on other studies included in the quantitative analysis.¹⁹^,²¹^,³²

Risk of bias across studies

Risk of bias across studies was assessed by a funnel plot in a way of displaying potential publication bias.

Results

Study selection

We have identified 540 records through database searching (PubMed, Embase, Cochrane Library, and SciELO). Additionally 6 articles were identified through reference lists of identified articles. After the removal of duplicates, 374 articles had their titles and/or abstracts screened. Owing to a clear misfit of eligibility criteria, 356 articles were excluded. Eighteen articles were assessed for full-text screening in more detail, and 11 were excluded because of the use of a 2D analysis system, 3D models, or for not reporting SHR or any motion pattern that could enable SHR calculation (eg, GH and ST range of motion) (Fig. 2). No unpublished relevant studies were obtained.

PRISMA 2009 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of the study.