If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Evidence suggests that patients with brachial plexus birth injury are more likely to retain midline function following a teres major tendon transfer without a concomitant latissimus dorsi transfer. Both procedures increase shoulder external rotation and abduction, but whether increased loss of midline frequency following double transfer is due to glenohumeral (GH) joint motion or scapulothoracic (ST) compensation is unknown. We hypothesized that double tendon transfers would exhibit greater GH external rotation than single tendon transfers, thus requiring greater ST rotation to internally rotate the shoulder, while GH and ST contributions to elevation remained equivalent between both groups.
Twenty-six postsurgical children with C5/C6 brachial plexus birth injuries participated in this study. Thirteen patients with single tendon transfers were matched with 13 with double tendon transfer. Coordinate systems of the thorax, scapula, and humerus were measured utilizing motion capture in 6 arm positions. Joint angles were calculated by the helical (ST) and modified globe method (GH and humerothoracic [HT]). Differences between groups were compared with repeated measures of multivariate analyses of variance for each position. Pending significant multivariate analyses of variance, univariate analyses of variance determined joint differences between transfer groups.
Joint rotations from neutral were similar between groups in 5 of 6 tested positions, with double tendon transfers consistently demonstrating 15°–20˚ more internal rotation at the GH and HT joints. Still, only the internal rotation position showed statistically significant differences in GH and HT joint angles. The ST joint angles were similar in this position (45.2˚ and 48.5˚).
The arc of motion for patients with double tendon transfer was more internally rotated than in patients with single tendon transfer at the GH and HT joints for all positions. However, both groups demonstrated little active rotation from neutral. Based on this data, teres major-only tendon transfers may not reduce the risk of loss of midline function.
Latissimus dorsi and/or teres major tendon transfers, often with concomitant joint reductions, are commonly used to improve shoulder abduction and external rotation in children with brachial plexus birth injuries (BPBIs). The literature consistently shows that tendon transfers improve both of these motions based on patients’ Mallet classification scores
Despite these shortcomings, the classification has been used to show that procedures designed to increase external rotation can also lead to loss of midline function (a drop in score to < 3 for internal rotation).
examined the tendon transfer’s role in loss of midline function by comparing single (teres major only) versus double (teres major and latissimus dorsi) tendon transfers. While double tendon transfers had significantly greater improvement in external rotation Mallet scores (+ 2.1 compared to + 1.5 in single transfers), they observed higher percentages of loss of midline function (35.7% and 14.3%, respectively). These findings suggest the nuances of tendon transfers can adversely impact shoulder function when measured by the Mallet classification.
The extent to which transfer technique influences functional outcomes depends on the underlying mechanisms driving shoulder motion, which remain unknown. In theory, tendon transfers improve function through greater active GH motion. However, this has not been objectively measured, and the effect on ST motion remains unknown.
suggested expressing caution when interpreting Mallet classification scores, because GH and ST joint contributions can vary while producing the same score. In fact, the ST joint can compensate for inadequate or counteractive GH motion to achieve the same score as individuals with better GH motion in some positions.
Using motion capture, this study aimed to determine the joint contributions to functional differences between single and double tendon transfers, specifically as they relate to external rotation, abduction, and loss of midline function. We hypothesized that double tendon transfers would exhibit greater GH external rotation than single tendon transfers, thus requiring greater ST motion to achieve internal rotation. We also hypothesized the GH and ST contributions to elevation would be equivalent between groups, thus both comparisons would provide insight into the potential inverse relationship of tendon transfers on midline function.
Materials and Methods
Twenty-six postsurgical children with upper trunk (C5–C6) BPBIs (average age 6.0 ± 1.9 years) were recruited for this study (Table 1). Data from a preliminary study showed an effect size of 1.17 (Cohen’s d) for HT external rotation displacement (how much the humerus was externally rotated relative to the trunk) in the internal rotation modified Mallet position. This Mallet position was chosen because it was commonly used to assess midline function before and after surgery. The HT rotation displacement was chosen to be consistent with the Mallet classification, which did not distinguish between GH and ST contributions to motion. A power level of 0.8, alpha value of 0.05, and a large effect size of 1.17 indicated that 26 patients (13 for each tendon transfer group) were needed for this study. Most participants’ data were collected retrospectively from postoperative follow-up where both motion capture and Mallet scores were measured. To fulfill matching requirements, 3 patients were recruited prospectively in accordance with institutional review board requirements and matched according to preoperative Mallet scores. Thirteen patients with single tendon transfers (teres major only) were matched with 13 patients with double tendon transfers based on the following criteria: (1) each pairing was either within 3 points for each shoulder subscore of the Active Movement Scale or within 1 point on each modified Mallet classification scores at their last preoperative outpatient visit (the largest difference in total score between pairs was 3 and 2 for each scale, respectively), (2) each pairing received the same GH joint reduction during tendon transfer surgery (closed or surgical), and (3) each pairing was separated by less than 2 years in age at the time of motion capture data collection. Additionally, all patients were 3 years of age or older, at least 6 months beyond their tendon transfer surgeries at the time of motion analysis (mean, 4 years; range, 6 mo to 6 y after surgery) and received no other treatments between surgery and the time of motion analysis.
Table 1Patient Demographics and Surgical Histories
Age at time of motion analysis
6.0 ± 1.9
6 ± 1.9
5.9 ± 2.0
8 M, 18 F
4 M, 9 F
4 M, 9 F
Age at time of tendon transfer
1.8 ± 1.2
2.0 ± 1.3
1.6 ± 1.1
Shoulder joint reduction
Interventions before tendon transfers
Prior botox and closed reduction
Prior nerve grafting
Post data collection
Unable to reach midline postoperatively
De-rotational humeral osteotomy following participation
No significant differences found between groups for interventions before tendon transfer following Fisher exact test.
All tendon transfer procedures at our institution were performed with the same technique regardless of whether 1 or 2 tendons were transferred, except for the joint reduction. Patients with an irreducible GH joint based on ultrasound or inadequate external rotation in the operating room (less than 30°) had surgical reductions performed. For patients who required surgical GH reduction, the anteroinferior portion of the GH joint capsule was released as described by Abzug et al.
Specifically, the joint capsule was incrementally released, with care to not over release, until satisfactory humeral external rotation was achieved. Upon completion, the patient’s arm was placed into 135° of abduction, maximum external rotation as the tendon transfer was tensioned, and the arm placed in a spica cast.
reported that the latissimus dorsi and teres major tendons were intimately associated and could be difficult to separate. In the authors' experience, this is an extremely rare occurrence, but it may be necessary to transfer both tendons if they cannot be separated. This did not occur in any of the patients in this cohort.
The patients of 2 surgeons were included in this study. Both surgeons used the same surgical technique. However, one surgeon preferentially performed single tendon transfers, unless the teres major tendon could not be safely separated from the latissimus dorsi tendon. The other surgeon performed double tendon transfers for patients with C5–C6 injuries and single tendon transfers for patients with C5–C7 or global injuries due to concern for loss of midline function.
Patients were seated, and retroreflective markers were placed on the sternal notch, T2 spinous process, T8 spinous process, and the medial and lateral epicondyles of the humerus by a trained physical therapist. Markers for the scapula were placed on the acromion process, trigonum spinae, and inferior angle. Patients were asked to hold their arms in 9 positions, including a neutral position or the natural resting position of the arm at the side (Table 2). The trigonum spinae and inferior angle markers were repalpated for each position to account for the change in the scapula’s position. Marker positions were recorded for 1 second in each position with a 12-camera motion capture system (Vicon) collecting at 60 Hz. Modified Mallet scores were recorded by a trained hand surgeon at the same visit as their motion capture collection (Table 3).
A peak shoulder elevation trial was used for comparing overhead motion between groups. When the modified Mallet and motion capture trials were collected, the patient’s arm was not restricted to the plane of motion being assessed. Consequently, peak shoulder elevation trials were determined by the position (abduction, flexion, or elevation) with the greatest HT elevation angle for each patient.
Coordinate systems for the thorax, humerus, and scapula were constructed such that the axes aligned with those recommended by the International Society of Biomechanics.
Custom software (LabView, National Instrument Corporation) was used to calculate ST, GH, and HT joint angles in each position. Joint displacements were calculated as the joint rotations to those positions from the starting neutral position. Scapulothoracic joint angles were calculated using helical angles, while the GH and HT joint angles were calculated using the modified globe method.
A modified globe method was selected over the International Society of Biomechanics-recommended Euler sequence to use an order-independent approach that produced joint angles consistent with clinical observations for all the positions tested.
Joint displacements for each position were compared with separate repeated measures of multivariate analyses of variance (MANOVAs). Factors consisted of group (single and double) and joint (HT, GH, and ST). The internal and external rotation MANOVAs used joint internal/external rotation displacements from neutral, while the peak shoulder elevation MANOVA consisted of the joint elevation displacements from neutral (upward rotation for ST joint and elevation for GH and HT joints). Pending significant differences between groups or the group-joint interaction, post hoc 1-way analysis of variance with a Bonferroni correction (α = 0.017) were performed between groups for the internal/external rotation and the shoulder elevation displacements across joints. Main effect differences between joints were not of interest in this study; therefore, no post hoc tests were performed following significant main effect differences for joints. This approach was repeated to compare the joint angles in each position.
A Fisher exact test was performed on the nonrandom association between interventions prior to tendon transfer and the tendon transfer groups to indicate if prior interventions could be having an impact on our results. Additionally, a mixed ANOVA was used to compare the pre- and postoperative Mallet scores. Factors consisted of group (single and double), and the repeated measures factor was position and time (pre- and postoperative).
There were no significant differences between joint displacements during internal rotation (P = .415), external rotation (P =.348), or peak shoulder elevation (P = .137). Since the MANOVA was nonsignificant, no further statistical analysis was performed on joint displacement data (Table 4).
Table 4Joint Displacements From Neutral
Peak shoulder elevation
All positions are described in degrees of internal rotation except the external rotation position, which is described in degrees of external rotation.
Peak shoulder elevation is described in degrees of scapulothoracic (ST) upward rotation and glenohumeral (GH)/humerothoracic (HT) elevation.
Boldface text indicates measurement differences greater than 10˚.
The MANOVA for joint angles during internal rotation yielded significant differences between groups (λ2,23 = 3.720, P = .040). The post hoc ANOVAs revealed double tendon transfers had significantly more internal rotation at the GH and HT joints (P = .010 and P = .007, respectively). In contrast, the ST internal rotation angles were not significantly different between surgical groups (P = .328).
Conversely, the MANOVAs for external rotation (P = .508), the neutral position (P = .180), and peak shoulder elevation (P = .137) joint angles did not reveal significant differences, so no further testing was implemented (Table 5).
The Fisher exact test found no significant associations for prior treatment and tendon transfer type (P = .197), and the mixed ANOVA found no significant main effect differences (P = .274), so no further analysis was performed.
Although treatment of patients with BPBI has historically focused on improving external rotation and abduction, recent research suggests that double tendon transfer procedures can adversely affect internal rotation.
Yet, the present study’s single tendon transfer group was 15°–20˚ more externally rotated at the GH and HT joints (Table 5) in motion capture measurements for all positions. While 15°–20° is substantial and visibly apparent, the Mallet classification was unable to discern this difference, likely due to lack of precision. This reinforces the current limitations in our understanding of tendon transfers.
Theoretically, tendon transfers convert active internal rotators into active external rotators. An abundance of literature has shown improved HT function after surgery, often attributed to improved “active” GH rotation, but without objective measurement of actual transferred muscle use or GH rotation.
The small GH displacement arcs (Fig. 2) and similar joint displacements between our groups, regardless of transfer type, indicate that active GH rotation remains substantially limited after surgery. This suggests that the tendon transfer does not increase active external rotation, but it likely reorients a similar arc of motion into a more externally rotated position. In our study, the neutral position showed a more externally rotated GH joint in the single tendon transfer group, which suggested that the more externally rotated resting shoulder position was maintained throughout the testing positions.
Based on previous evidence, we expected greater external rotation in the double tendon transfer group, however, the results did not support this expectation.
This may be due to unforeseen and possibly unmeasurable factors influencing loss of midline function. For example, a child’s passive external rotation during the procedure dictates the extent of incremental joint release. Therefore, the degree of preoperative external rotation and/or the magnitude of the release may predominantly influence rotation regardless of tendon transfer type. Preoperative passive external rotation has already been shown to be a good predictor of outcomes for closed joint reductions.
Other factors that may have contributed to postoperative GH external rotation include concomitant subscapularis release (2 in single transfer group, none in the double group), prior botulinum toxin injections (8 in single transfer group, 3 in double group), or extent of GH dysplasia (not recorded).
Emphasis on surgically achieved improvements in external rotation and abduction, with limited focus on the reciprocal midline function may have previously underestimated the impact of joint release in tendon transfer patients.
External rotation gains were therefore made at the expense of internal rotation. In part, this may be due to the cohort including concomitant complete subscapularis releases. Our findings show single tendon transfers can have greater GH external rotation than double tendon transfers even though more internal rotators remain intact. Together, these findings suggest the extent of joint release is a possible confounding factor when assessing postoperative movement attributed to tendon transfer type.
found that including a partial subscapularis release in conjunction with anterior capsular release increased external rotation and abduction Mallet scores while maintaining midline function. Partial subscapularis release was instituted after observing a total loss of midline function following complete subscapularis release.
Two patients in our study had more substantial release of the subscapularis (1 also had concomitant pectoralis major release) due to inadequate external rotation during tendon transfer. Both patients had single tendon transfers and demonstrated the largest external rotation joint angles of the entire cohort. Therefore, the extent of joint release may have a greater impact on postoperative joint orientation than the tendon transfer itself. Unfortunately, we could not account for the extent of joint release in this study.
This study has several limitations. Preoperative motion capture data is not available for this cohort, which limits interpretation of the findings. Obtaining reliable preoperative motion capture measurements is impractical because of the patients’ young ages when surgery was performed. Additionally, some comparisons may be underpowered due to small sample size, because the power analysis is based on the internal rotation position (position of most interest). Two surgeons were involved in all procedures, to varying extents, with both present for most procedures. Regardless of tendon transfer type, procedures were performed as described by Abzug,
but 1 surgeon performed fewer closed reductions and more single transfers than the other. Therefore, 1 surgeon may have a lower threshold for performing joint releases. Additionally, the extent of joint release and preoperative treatment strategies are variable within each patient. Brachial plexus birth injuries are individualistic in nature, with no 2 patients exhibiting identical deficits. Therefore, matching controls can reduce group differences but not eliminate them.
Based on the current findings and literature detailed, the precise indications for performing a single versus a double tendon transfer remain unclear. The current study showed approximately 15˚ greater GH and HT external rotation in the single tendon transfer group in all positions and no differences in ST motion. More importantly, although most clinical scores improved, it was apparent that active GH joint motion was minimal in all patients. Consequently, tendon transfers appear to establish a more functional resting GH joint orientation rather than produce more active GH rotation. Other factors, such as extent of joint release, may further influence functional outcomes after tendon transfer surgery. Therefore, we recommend caution when interpreting a single tendon transfer’s ability to preserve postoperative midline function based on functional measurement scores.