CHILDREN S ORTHOPAEDICS Surgical correction of shoulder rotation deformity in brachial plexus birth palsy

Transcription

1 CHILDREN S ORTHOPAEDICS Surgical correction of shoulder rotation deformity in brachial plexus birth palsy LONG-TERM RESULTS IN 118 PATIENTS T. Hultgren, K. Jönsson, F. Roos, H. Järnbert-Pettersson, H. Hammarberg From Karolinska Institute, Stockholm, Sweden T. Hultgren, PhD, MD, Consultant Hand Surgeon K. Jönsson, MD, Research Assistant F. Roos, MD, Consultant Hand Surgeon H. Hammarberg, MD, PhD, Consultant Hand Surgeon Karolinska Institute, Department of Clinical Science and Education, Section for Hand Surgery, Södersjukhuset S Stockholm, Sweden. H. Järnbert-Pettersson, PhD, Statistician Karolinska Institute, Department of Clinical Science and Education, Södersjukhuset S Stockholm, Sweden. Correspondence should be sent to Mr T. Hultgren; tomas.hultgren@ sodersjukhuset.se 214 The British Editorial Society of Bone & Joint Surgery doi1.132/31-62x.96b $2. Bone Joint J 214;96-B: Received 15 January 214; Accepted after revision 4 July 214 We present the long-term results of open surgery for internal shoulder rotational deformity in brachial plexus birth palsy (BPBP). From 1997 to 25, 27 patients (17 females, 1 males, mean age 6.2 (.6 to 34)) were operated on with subscapularis elongation and/or latissimus dorsi to infraspinatus transfer. Incongruent shoulder joints were relocated. The early results of these patients has been reported previously. We analysed 118 (64 females, 54 males, mean age 15.1 (7.6 to 34)) of the original patient cohort at a mean of 1.4 years (7. to 15.1) post-operatively. A third of patients with relocated joints had undergone secondary internal rotational osteotomy of the humerus. A mixed effects models approach was used to evaluate the effects of surgery on shoulder rotation, abduction, and the Mallet score. Independent factors were time (pre-and postsurgery), gender, age, joint category (congruent, relocated, relocated plus osteotomy) and whether or not a transfer had been performed. Data from a previously published short-term evaluation were reworked in order to obtain pre-operative values. The mean improvement in external rotation from pre-surgery to the long-term follow-up was 66.5 (95% confidence interval (CI) 61.5 to 71.6). The internal rotation had decreased by a mean of 22.6 (95% CI to -26.5). The mean improvement in the three-grade aggregate Mallet score was 3.1 (95% CI 2.7 to 3.4), from 8.7 (95% CI 8.4 to 9.) to 11.8 (11.5 to 12.1). Our results show that open subscapularis elongation achieves good long-term results for patients with BPBP and an internal rotation contracture, providing lasting joint congruency and resolution of the trumpet sign, but with a moderate mean loss of internal rotation. Cite this article: Bone Joint J 214;96-B: Brachial plexus birth palsy (BPBP) frequently results in an internal rotation deformity of the shoulder despite an otherwise good neurological recovery. 1-3 The internally rotated position of the limb is cosmetically and functionally disturbing. 4,5 To bring the hand to the face the shoulder is abducted, elevating the elbow; the so-called trumpet sign, 6 and patients may have difficulty with activities of daily living. A number of surgical procedures have been suggested and evaluated, 1,7-11 but there are few large studies that have analysed the long-term results of surgery In a previous publication, we examined the results for a series of 27 patients with BPBP and shoulder deformity at one year following open surgery. 3 In this study we revisit the same series seven years or more after the primary operation, considering longterm functional outcome and re-operation rate. Our interest was whether the surgery continued to provide improved limb mobility over an extended period of time. In addition, we wanted to know if the long-term mobility was affected by the addition of a latissimus dorsi transfer, the relocated joints remained congruent, how many of the patients with relocated joints required additional surgery and how their mobility compared with the other groups. Finally, we wondered if a reduction of internal rotation in the shoulder affected limb function. Patients and Methods From 1997 to 29, operations were performed on a consecutive series of 274 patients with unilateral BPBP and an internally rotated shoulder deformity. Open subscapularis elongations and latissimus dorsi to infraspinatus transfers were performed either separately or in combination. Concomitant relocation of an incongruent joint was possible in 97 of 15 (92.4%) of the patients. The indications for surgery, choice of operating procedures and surgical techniques have been described in detail in our previous publication. 3 The inclusion criterion for the present study was a minimum follow-up time of seven years VOL. 96-B, No. 1, OCTOBER

2 1412 T. HULTGREN, K. JÖNSSON, F. ROOS, H. JÄRNBERT-PETTERSSON, H. HAMMARBERG Operated on (274) Follow-up at one year (27) Eligible for long-term follow-up (27) Follow-up at > seven years (118) Fig. 1 Lost to follow-up (4) Less than seven years post-op (63) Could not come (15) Did not respond (74) Flowchart of patient inclusion. after the initial operation. Ethical approval was granted for the study by the Regional Ethics Committee in Stockholm. Eligible patients were contacted with written information and invited to take part in the study: from a potential cohort of 27 patients, 118 consented to the assessment (Fig. 1). The mean follow-up time was 1.2 years (7. to 15.1). The mean age at follow-up was 15.1 years (7.6 to 33.7) and there were 64 females and 54 males. The nerve injury was classified as C5 C6 or C5 C7 in 116 of the examined patients, the inferior trunk was also involved in the remaining two patients. A total of six of the upper trunk group had undergone nerve reconstruction and one of the two patients with lower trunk involvement had the nerve injury reconstructed. Shoulder joint relocation had been achieved in 63 of 66 patients with incongruent joints. There were three failed relocations, which were left subluxed; the remaining 52 joints were congruent. Since the one year follow-up, 23 of the 63 had also required an internally rotating osteotomy of the proximal humerus. The indication for the osteotomy in 18 patients was an active internal rotation of < 3, causing difficulties in their daily activities. In the other five patients, recurrent intermittent subluxation of the humeral head had started between three and six years after the index operation. The subluxation would occur in extreme internal rotation, sometimes during sleep. A total of three of these five patients also had a posterior stabilisation procedure according to Hawkins and Janda. 15 All five joints were stable at the long-term follow-up. We have not seen any other patients in the series with late re-subluxation. Six patients with relocated joints had < 3 of active internal rotation but were not troubled by the restriction and did not desire an osteotomy. Latissimus dorsi transfer was an isolated procedure for an initial eight patients with full passive mobility. Five of these patients subsequently developed a rotational contracture necessitating a second operation with subscapularis elongation. All eight patients remain in their original subgroup ( transfer only ) in the statistical analysis. Shoulder mobility (thoraco-humeral movement) and Mallet scores 6 were measured pre- and post-operatively, as described in our previous publication. 3 We used the modified three-grade Mallet score as described by Birch, 16 where each of the five specified movements are graded from 1 to 3, giving an aggregate score of from 5 to 15. The examinations were performed by three surgeons who had not been involved in the primary surgery, two of whom are coauthors (KJ and FR). In order to assess functional impairment due to a limitation of the internal shoulder rotation, the participants were requested to grade their ability to perform three activities: to button and unbutton their shirt or blouse, to put their hand in their trouser pocket, and to wash the opposite axilla without the aid of the good hand. Each activity was given three grades: easy, difficult or impossible. The grades were then compared with the values for internal rotation in order to determine if there was a correlation between the range of internal rotation and impairment of the specified function. Wrist flexion was also analysed as it assists many activities involving internal rotation of the shoulder. Statistical analysis To study any inclusion bias between the re-examined and unexamined groups, we compared their shoulder function at one-year post-surgery, using the independent samples t-test for external rotation, internal rotation and abduction and the Mann-Whitney U test for the Mallet scores. A mixed-effects model approach was used to evaluate the effect of surgery on shoulder function over time, to allow correlation of repeated observations on the same individual. Furthermore, mixed-effects models are able to accommodate missing data and integrate time-varying factors, which are issues in the present study. 17 The pre-operative and one-year post-operative values presented all refer to the larger groups of patients examined at those time points. 3 The outcome factors studied were external and internal shoulder rotation, shoulder abduction, aggregate and separate movement Mallet scores. Independent factors were: the time points (pre-surgery, post-surgery at one year and at seven years), gender, age at operation, shoulder joint congruency and whether or not a latissimus dorsi transfer had been performed. The model strategy used to evaluate the effect of surgery has been described before. 3 In brief, we first studied unadjusted associations for each independent factor. Two-way interactions between the time point and each of the other four independent factors were tested. If the interaction was significant (p <.5), we created a new factor consisting of the combinations of the levels from the original factors. Adjusted associations were then calculated by including all THE BONE & JOINT JOURNAL

3 SURGICAL CORRECTION OF SHOULDER ROTATION DEFORMITY IN BRACHIAL PLEXUS BIRTH PALSY 1413 Table I. Mixed-effect model regression analysis of factors associated with active external rotation ( ). Estimated means (95% confidence interval) and p-values given for comparison with the reference level (i.e. first factor level). Values have been calculated for the larger groups that were examined pre-operatively (262) and at one year post-operatively (263), and for the smaller group of 118 which was available at seven years post-operatively (64 females, 54 males)) Factor Factor level Patients (%) Unadjusted Adjusted 1 Adjusted 2 Adjusted 3 Interaction time * age Pre-operative, < 2 years 51 (18.9) Reference Reference Reference Reference Time point Pre-operative 27 Reference Post-operative one yr (79.3 to, 88.3) p <.1 Post-operative 7 yrs (61.5 to 71.6) p <.1 Gender Female 142 (52.6) Reference Reference Reference Reference Male 128 (47.4) -2.6 (-9.7 to 4.6) p = (-5.8 to 1.9), (-5.8 to 2.) p = (-5.7 to 2.) p =.34 Age group < 2 yrs 51 (18.9) Reference Reference Reference 2 to 19 yrs 215 (79.6) 5.7 (-3.2 to 14.7), (-3.6 to 1.7), (.1 to 11.) p =.46 2 yrs 4 (1.5) (-59.6 to -7.2) p = (-44.1 to -8.5) p = (-43.5 to -7.9) p =.5 Joint category Congruent 165 (61.1) Reference Reference Reference no osteotomy 82 (3.4) -8. (-16. to.) p = (-15.4 to -6.2) p < (-14.8 to -5.7) p <.1 plus osteotomy 23 (8.5) -4.3 (-16.1 to 7.4) p = (-18.4 to -4.1) p = (-18.3 to -4.1) p =.2 Transfer procedure Elongation only 177 (65.6) Reference Reference Reference Elongation plus transfer 76 (28.1) -4.9 (-13.1 to 3.2) p = (-11.9 to -2.8) p = (-12. to -2.8) p =.2 Transfer only 17 (6.3) -6.3 (-21.1 to 8.5) p = (-2.4 to -4.2) p = (-2.5 to -4.2) p =.3 2 to 19 yrs 215 (79.6) 23. (14.8 to 31.1) p <.1 2 yrs 4 (1.5) 9.6 (-16.4, 35.6) p =.47 Post-operative one yr, < 2 yrs 51 (18.9) (15.9 to 124.7) p <.1 2 to 19 yrs 215 (79.6) 1.1 (92. to 18.1) p <.1 2 yrs 4 (1.5) 53.3 (29.2 to 77.5) p <.1 Post-operative 7 yrs, < 2 yrs 29 (1.7) 81.2 (7.3 to 92.), p <.1 2 to 19 yrs 89 (33.) 88.1 (79.7 to 96.6) p <.1 2 yrs Interaction time * joint Pre-operative, congruent 165 (61.1) Reference Reference Reference Reference no osteotomy 82 (3.4) (-32.7 to -18.8) p <.1 plus osteotomy 23 (8.5) (-35.2 to -12.7) p <.1 Post-operative one yr, 165 (61.1) 72.8 (67.5 to 78.2) p <.1 congruent no osteotomy 82 (3.4) 73.6 (67. to 8.3) p <.1 plus osteotomy 23 (8.5) 82.9 (72.3 to 93.6) p <.1 Post-op 7 yrs, congruent 55 (2.4) 6.2 (53.4 to 67.) p <.1 no osteotomy 4 (14.8) 59.5 (51.9 to 67.) p <.1 plus osteotomy 23 (8.5) 45. (35. to 55.) p <.1 Interaction time * transfer Pre-operative, elongation only 177 (65.6) Reference Reference Reference Reference Elongation plus transfer 76 (28.1) -8.3 (-15.5 to -1.) p =.25 Transfer only 17 (6.3) -8. (-21.3 to -5.2),.23 Post-operative one year, 177 (65.6) 85.2 (79.6 to 9.7) p <.1 elongation only Elongation plus transfer 76 (28.1) 75.1 (68.2 to 82.) p<.1 Transfer only 17 (6.3) 63.7 (51.3 to 76.2) p<.1 Post-operative 7yrs, 83 (3.7) 64.9 (58.7 to 71.1) p <.1 elongation only Elongation plus transfer 27 (1.) 65.5 (56.8 to 74.3) p <.1 Transfer only 8 (3.) 61.4 (46.3 to 76.5) p <.1 Adjusted 1, adjusted for all variables: joint, muscle transfer, gender and interaction time*age. Adjusted 2, adjusted for all variables: age, muscle transfer, gender and interaction time*joint. Adjusted 3, adjusted for all variables: joint, age, gender and interaction time* muscle transfer the independent factors that were not included in the interaction and the new factor for the interaction. We limited the number of subgroups to obtain a sufficient number of patients in each. The short-term study had not demonstrated any clinically relevant differences in the results for subjects who underwent operation between the ages of two and 19 years, 3 so three age groups were used for the long-term evaluation: < two years, two to 19 years, and 2 years or older at operation. All of the incongruent joints were combined into one group and a subdivision defined, depending on whether or not a secondary rotating osteotomy had been performed. Abduction and Mallet scores were analysed in the same way, to study the differences over time adjusted for other factors, and the results are summarised in the text. Because the Mallet score is an ordinal scale we implicitly assume that a unit increase from one to two is equal to an increase from two to three. All tests were two-sided and the results were considered significant if a p-value was <.5. Wrist flexion of both sides was compared using the paired t-test. All analyses were performed using IBM SPSS version 21 software (IBM Corp., Armonk, New York). Results Age distribution, gender, types of joint and surgical procedures are summarised in Tables I and II, which present the results of the mixed-effect model regression analysis for VOL. 96-B, No. 1, OCTOBER 214

4 1414 T. HULTGREN, K. JÖNSSON, F. ROOS, H. JÄRNBERT-PETTERSSON, H. HAMMARBERG Table II. Mixed-effect model regression analysis of factors associated with active internal rotation ( ). Estimated means (95% confidence interval) and p-values given for comparison with the reference level (i.e. first factor level). Values have been calculated for the larger groups that were examined pre-operatively and at one year post-operatively (27), and for the smaller group of 118 which was available at seven years post-operatively (64 females, 54 males)) Factor Factor level Patients % Unadjusted Adjusted 1 Adjusted 2 Adjusted 3 Interaction time * age Pre-operative, < 2 yrs 49 (18.6) Reference Reference Reference Reference Time point Pre-operative 262 Reference Post-operative one yr (-15.7 to -22.6) p <.1 Post-operative 7 yrs (-18.7 to -26.5) p <.1 Gender Female 139 (52.8) Reference Reference Reference Reference Male 124 (47.1).7 (-2.7 to 4.1) p =.48.7 (-1. to 2.4) p =.42.7 (-1. to 2.4) p =.44.7 (-1. to 2.4) to.43 Age group < 2 yrs 49 (18.6) Reference Reference Reference 2 to 19 yrs 29 (79.5) 5.7 (1.4 to 1.) p = (-5.7 to -.7), (-5.8 to -.7) p =.11 2 yrs 4 (1.5) 19.8 (4.1 to 35.6) p = (-6.3 to 8.5) p = (-6.3 to 8.5) p =.77 Joint category Congruent 162 (61.2) Reference Reference Reference no osteotomy 78 (29.7) 7.6 ( 4. to 11.3) p < (-2.5 to 1.6) p = (-2.5 to 1.6) p =.65 plus osteotomy 23 (8.7) 19.5 (-14. to 24.9) p < (-6. to.8) p = (-6. to.8) p =.14 Transfer procedure Elongation only 171 (63.3) Reference Reference Reference Elongation plus transfer 75 (28.5).6 (-3.3 to 4.5) p = (-2.2 to 1.8) p = (-2.2 to 1.9) p =.88 Transfer only 16 (5.9) 8.8 (1.6 to 15.9) p = (-5.8 to 1.5) p = (-5.8 to 1.5) p =.25 2 to 19 yrs 29 (79.5) -4.3 (-6.8 to -1.7), p =.1 2 yrs 4 (1.5) -.7 (-8.3 to 6.8), p =.85 Post-op one yr, < 2 yrs 49 (18.6) (-41.2 to -27.5), p <.1 2 to 19 yrs 29 (79.5) (-23.8 to -15.7), p <.1 2 yrs 4 (1.5) -.7 (-24.7 to 23.2) p =.95 Post-op 7 yrs, < 2 yrs 29 (11.) (-35.5 to -19.7)to. 2 to 19 years 89 (33.8) (-3.9 to -2.6), p <.1 2 yrs Interaction time * joint Pre-operative, Congruent 162 (61.6) Reference Reference Reference Reference no osteotomy 78 (29.7).7 (-1.4 to 2.8) p =.5 plus osteotomy 23 (8.7) -.1 (-3.6 to -3.4) p =.96 Post-opeartive one year, 162 (61.6) -9. (-12.3, 3.4) p <.1 congruent no osteotomy 78 (29.7) (-33.6 to -23.6), p <.1 plus osteotomy 23 (8.7) (-6.9, -42.9) p <.1 Post-op 7 yrs, congruent 55 (2.9) (-28.1 to -16.6) p <.1 no osteotomy 4 (15.2) (-29.3 to -15.4) p <.1 plus osteotomy 23 (8.7) (-4.9 to 21.7) p <.1 Interaction time * transfer Pre-operative, elongation 171 (65.) Reference Reference Reference Reference only Elongation plus transfer 75 (28.5) -.4 (-2.5to 1.6) p =.68 Transfer only 16 (6.1) -3.2 (-6.9 to.6) p =.96 Post-operative one yr, 171 (65.) (-24.9 to -17.3) p <.1 elongation only Elongation plus transfer 75 (28.5) (-23.7 to -12.1) p <.1 Transfer only 16 (6.1) -3.3 (-15.5 to 8.9) p =.59 Post-operative 7 yrs, 83 (31.6) (-27.2 to -18.) p <.1 Elongation only Elongation plus transfer 27 (1.3) (-35.8 to -19.3) p <.1 Transfer only 8 (3.) -12. (-27.4 to 3.3) p =.12 Adjusted 1, adjusted for all variables: joint, transfer, gender and interaction time*age. Adjusted 2, adjusted for all variables: age, transfer, gender and interaction time*joint. Adjusted 3, adjusted for all variables: joint, age, gender and interaction time*transfer. We have reworked our data for the whole group but only had a reduced number at long-term follow-up. Internal rotation was only evaluated in 263 of 27 patients external and internal rotation, with the estimated fixed effects, their 95% confidence intervals (CIs) and their significance. Figures 2 and 3 represent raw means and 95% confidence intervals. There was an inclusion bias at the long-term follow-up with respect to the condition of the joint, with a higher proportion of relocated joints in the examined group: 63 of 118 (53%), compared with 19 of 89 (21%). We also found an inclusion bias between the examined and the unexamined groups in the long-term study with respect to their results at the one-year follow-up. The examined patients had slightly poorer function at one-year following surgery in both abduction and internal rotation, with a difference in means of 11.9 (95% CI of difference:.2 to 23.7; p =.47) for abduction and 12. (95% CI of difference: 5.2 to 18.8; p <.1; t-test) for internal rotation. There were no statistically significant differences between the two groups at the one-year follow-up with respect to external rotation (difference of means: -3.6, 95% CI of difference: -1.1 to 2.8; p =.27) or Mallet scores (aggregate Mallet score: p =.42, hand-to-mouth: p =.8, hand-to-head: p =.24 and hand-to-back: p =.53). External rotation. The factors associated with external rotation are presented in Table I. The unadjusted mean THE BONE & JOINT JOURNAL

5 SURGICAL CORRECTION OF SHOULDER ROTATION DEFORMITY IN BRACHIAL PLEXUS BIRTH PALSY 1415 Mean external rotation ( ) Congruent joint only Joint category Fig. 2 + osteotomy Pre-op Post-op one yr Post-op seven yrs Bar chart showing mean external rotation by joint category at the three time points: pre-operatively (pre-op), post-operatively (post-op) at one year and post-op at more than seven years. Mean internal rotation ( ) Congruent joint only Joint category Fig. 3 + osteotomy Pre-op Post-op one yr Post-op seven yrs Bar chart showing the mean internal rotation by joint category at the three intervals: pre-operatively (pre-op), post-operatively (post-op) at one year and post-op at more than seven years. improvement in external rotation from pre-surgery to more than seven years post-surgery was 66.5 (95% CI 61.5 to 71.6). The adjusted mean improvement in external rotation was similar in the youngest and two to 19 year old age groups (81.2, 95% CI 7.3 to 92. and 88.1, 95% CI 79.7 to 96,6, respectively, compared with the reference). Means and 95% CIs for external rotation at the three time points by joint category are presented in Figure 2. The adjusted mean improvement was similar for the congruent joints (6.2, 95% CI 53.4 to 67. ) and the relocated joints with no osteotomy (59.5, 95% CI 51.9 to 67. ), but slightly less for the relocated joint group with osteotomy (45., 95% CI 35. to 55. ), compared with the reference. The improvement in external rotation was similar for males and females. The outcome for the three transfer groups was also very similar at the long-term follow-up, with means of 64.9 (95% CI 58.7 to 71.1 ) for the elongated only group, 65.5 (95% CI 56.8 to 74.3 ) for the elongated and transferred group and 61.4 (95% CI 46.3 to 76.5 ) for the group with a transfer only, compared with the reference (Table I). Internal rotation. The unadjusted loss of mean internal rotation from pre- to more than seven years post-surgery was 22.6 (95% CI 18.7 to 26.5 ). The patients with relocated joints who were not selected for rotational osteotomy had a mean loss of 28.6 (95% CI 23.6 to 33.6 ) at one year following surgery, while the group that was osteotomised had a mean loss of internal rotation of 51.9 (95% CI 42.9 to 6.9 ) compared with the reference (Table II). Figure 3 illustrates means and 95% CIs for internal rotation at the three time points, by joint category. At the longterm follow-up, the adjusted mean internal rotation was similar in the congruent joint group (-22.3 compared with the reference) and the relocated joint group with no osteotomy ( compared with the reference). At the long-term followup, the group with relocated joints and subsequent osteotomy had regained a substantial amount of the active internal rotation that was lost at one year after the relocation (Fig. 3). Although slightly less than in the congruent joint group at more than seven years, the difference in internal rotation between the two groups was not statistically significant (difference in means 8.9, (95% CI -2.3 to 2.1, p =.12)) There was no significant difference in internal rotation at the long-term follow-up between the group that had an elongation only and the group that had an elongation plus transfer (difference in means 6.4, (95% CI -2.6 to 15.4, p =.16)). There were no significant changes in the long term associated with gender or age. Figures 4 and 5 show the distribution of internal rotation for individual patients and how they judged their ability to perform a specified activity at the long-term follow-up (17 of 118 responded to this questionnaire). Buttoning a shirt could be performed by all 17 patients who responded and was classified as easy by 77% (82 of 17) of the patients. There were no distinct ranges of rotation which defined easy or difficult. A total of three of 17 patients were unable to put their hand in the trouser pocket, but their internal rotation ranged from 55º to 9º and thus did not appear to be a determining factor for that activity. The most difficult of the three activities was washing the other axilla, with seven of 17 patients unable to perform the function without the aid of the other hand. The mean wrist flexion on the affected side for the whole group was 64.5 ( to 95). This was 9.1 less than on the unaffected side (95% CI to -6.7 ; p <.1, paired t-test) Abduction. The unadjusted overall abduction at more than seven-years post-surgery was 13.6 (95% CI 98.1 to VOL. 96-B, No. 1, OCTOBER 214

6 1416 T. HULTGREN, K. JÖNSSON, F. ROOS, H. JÄRNBERT-PETTERSSON, H. HAMMARBERG Count (n) Easy Difficult Impossible Mean abduction ( ) Pre-op Post-op one year Post-op seven yrs Internal rotation ( ) Fig. 4 Bar chart showing active internal rotation and the ability to complete the task of buttoning/unbuttoning a shirt or blouse at more than seven years post-surgery (n = 17). Congruent joint only Joint category Fig. 6 Relocation + osteotomy Bar chart showing the mean abduction by joint category at the three time points: pre-operative (pre-op), post-operative (post-op) at one year and post-op at more than seven years (post-op = 7 yrs). Count (n) Internal rotation ( ) Fig. 5 Easy Difficult Impossible Bar chart showing active internal rotation and the ability to wash the opposite axilla without using the uninvolved hand at more than seven years post-surgery (n = 17) ) and the change from pre-surgery to seven-years post-surgery was -7.4 (95% CI -2. to ; p =.7, mixed models). There were substantial differences in abduction between the three joint categories, but no statistically significant changes were noted over time in either of the subgroups (Fig. 6). For instance, the adjusted abduction for the relocated joint group with osteotomy at more than seven years post-surgery was 37.5 lower than for the congruent joint group (95% CI 19.6 to 55.5 ; p <.1, mixed models). When adjusted for all factors and time*age, the youngest age group had improved their abduction at the long-term follow-up, but the change was not statistically significant (1.5 ; 95% CI: 1. to 22.; p =.74, mixed models). When adjusted for all factors and time*transfer, there was a significant loss of abduction in the combined release-plustransfer group (-16.2 ; 95% CI: to -5.6; p =.3, mixed models). Mallet score. The unadjusted mean improvement in the aggregate Mallet score over the long term was 3.1 units (95% CI: 2.7 to 3.4; p <.1, mixed models), from 8.7 (95% CI: 8.4 to 9.) pre-operatively to 11.8 at seven-years following surgery. (95% CI: 11.5 to 12.1). The adjusted improvement was similar within the subgroups, but the end result for patients with congruent joints was 1.7 units higher (95% CI:.7 to 2.7; p <.1, mixed models) than for the group with relocated joints and osteotomy. There was no statistically significant difference in the end results between the two age groups (difference of means:.17, 95% CI of difference: -.66 to.99; p =.69, mixed models). The trumpet sign had been corrected to Mallet grade 3 in 79% (71 of 9) of the patients at long-term follow-up. The mean score for hand-to-back was unchanged from pre- to more than seven years post-surgery (-.1; 95% CI: -.2 to.1; p =.56, mixed models). Discussion This study shows a good long-term improvement in the mean external rotation following subscapularis elongation, with effective correction of the trumpet sign, an acceptable mean loss of internal rotation, and no clinically relevant change in mean abduction. The long-term improvement in external rotation was similar in the group with subscapularis elongation alone THE BONE & JOINT JOURNAL

7 SURGICAL CORRECTION OF SHOULDER ROTATION DEFORMITY IN BRACHIAL PLEXUS BIRTH PALSY 1417 and the group with a combined subscapularis elongation and latissimus dorsi transfer. The patients were not randomised and it is possible that the selection bias was so appropriate it resulted in almost identical outcomes in the two groups. A more likely explanation is that the latissimus dorsi transfer contributed very little to the improvement in external rotation. We no longer perform a latissimus dorsi in combination with the subscapularis elongation. A transfer can be done as a second procedure in the few cases where it is found to be required. 3 For the patients who had a weakness in external rotation but did not have a rotational contracure, the latissimus dorsi to infraspinatus transfer provided useful improvement. The transfer however, could prevent contracture formation in the longer term. Half of the patients thus treated had developed an internal rotation contracture after the first year. Additional surgery was performed after the first year in approximately 25% of the patients with relocated joints in order to adjust the rotational range or to stabilise the joint. Late recurrence of intermittent subluxation was noted in 8% of the examined patients; all of whom were stabilised by a second operation. The overall mean loss of internal rotation in the long term is not dramatic and may reflect an adaptation to a functional rotation range. When considering individual values, however, there was a very large variation in the series. While a loss of internal rotation may be inevitable when the shoulder joint is relocated, there was a small group of patients with congruent joints who achieved < 4º of internal rotation and we have not been able to determine why they lost more movement than the others. Patients with partially healed C5 C6 and C5 C7 lesions generally have good wrist flexion, as shown in this study. Reaching the midline is aided by both wrist flexion and shoulder extension and does not require full internal rotation of the shoulder. All of the patients in our series were able to button a shirt. Reaching across the midline to the axilla requires more internal rotation (plus shoulder flexion) and may be a better test for this particular movement. We also noted that the Mallet hand-to-back test correlated poorly with internal rotation and remained unchanged from pre-surgery to post-surgery in our series. While our study has demonstrated substantial loss of internal rotation over time in some patients after surgery, it also suggests that the impact on daily activities may be less negative than expected. A weakness of this study is that a large proportion of patients were lost to long-term follow-up. However, our analysis for inclusion bias suggests that the long-term results are not exaggerated. Another weakness is that the pre-operative and one year post-operative measurements were made by the surgeon (TH), whereas the measurements at long-term follow-up were made by three independent examiners. This may have contributed to the finding of slightly poorer results at the long-term follow-up. We have endeavoured to minimise the risk for bias by using a pre-set protocol for joint measurements with defined reference points, sighting lines and the consistent use of a goniometer. Other long-term studies on secondary shoulder surgery in BPBP patients have reported deteriorating results over time We have previously recommended that elongation of the subscapularis should not be done in patients with < 7º of internal rotation. 3 In view of the long-term results, we further recommend that the elongation should provide no more than 45 of passive external rotation on the operating table (except when relocating the shoulder joint) and that caution should be observed with patients who have poor wrist flexion. We stress the importance of long-term post-operative surveillance due to the large individual variation and the need for additional surgery in selected patients. In conclusion, this study shows that open subscapularis elongation achieves good long-term results for patients with BPBP and an internal rotation contracture, providing lasting joint congruency and resolution of the trumpet sign, with a moderate mean loss of internal rotation. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by E. Moulder and first proof edited by G. Scott. References 1. Kambhampati SB, Birch R, Cobiella C, Chen L. Posterior subluxation and dislocation of the shoulder in obstetric brachial plexus palsy. J Bone Joint Surg [Br] 26;88- B: Waters PM, Smith GR, Jaramillo D. Glenohumeral deformity secondary to brachial plexus birth palsy. J Bone Joint Surg [Am] 1998;8-A: Hultgren T, Jönsson K, Pettersson H, Hammarberg H. Surgical correction of a rotational deformity of the shoulder in patients with obstetric brachial plexus palsy: Short-term results in 27 patients. Bone Joint J 213;95-B: Westin B, Hultgren T, von Koch L. Obstetric brachial plexus injury: expectations before and satisfaction three months after secondary surgery on the shoulder. Hand Therapy 212;17: Langer JS, Sueoka SS, Wang AA. The importance of shoulder external rotation in activities of daily living: improving outcomes in traumatic brachial plexus palsy. J Hand Surg [Am] 212;37-A: Mallet J. Obstetrical paralysis of the brachial plexus. II. Therapeutics. Treatment of sequelae. Priority for the treatment of the shoulder. Method for the expression of results. Rev Chir Orthop Reparatrice Appar Mot 1972;58:(Suppl1): (in French). 7. Fairbank HAT. Birth Palsy: subluxation of shoulder joint in infants and young children. Lancet 1913;3: Hoffer MM, Wickenden R, Roper B. Brachial plexus birth palsies. Results of tendon transfers to the rotator cuff. J Bone Joint Surg [Am] 1978;6-A: L Episcopo JB. Restoration of muscle balance in the treatment of obstetrical paralysis. N Y J Med 1939;39: Pearl ML. Arthroscopic release of shoulder contracture secondary to birth palsy: an early report on findings and surgical technique. Arthroscopy 23;19: Zancolli A, Zancolli R. Reconstructive surgery in brachial plexus sequelae. In: Gupta A, Kay S, Scheker LR, eds. The Growing Hand. London: Mosby, 2: Cohen G, Rampal V, Aubart-Cohen F, Seringe R, Wicart P. Brachial plexus birth palsy shoulder deformity treatment using subscapularis release combined to tendons transfer. Orthop Traumatol Surg Res 21;96: Kirkos JM, Kyrkos MJ, Kapetanos GA, Haritidis JH. Brachial plexus palsy secondary to birth injuries. J Bone Joint Surg [Br] 25;87-B: VOL. 96-B, No. 1, OCTOBER 214

8 1418 T. HULTGREN, K. JÖNSSON, F. ROOS, H. JÄRNBERT-PETTERSSON, H. HAMMARBERG 14. Pagnotta A, Haerle M, Gilbert A. A long-term results on abduction and external rotation of the shoulder after latissimus dorsi transfer for sequelae of obstetric palsy. Clin Orthop Relat Res 24;426: Hawkins RJ, Janda DH. Posterior instability of the glenohumeral joint. A technique of repair. Am J Sports Med 1996;24: Birch R, Bonney G, Wynn Parry CB. Surgical disorders of the periferal nerves. Chuchill Livingstone 1998; Gueorguieva R, Krystal JH. Move over ANOVA: progress in analyzing repeatedmeasures data and its reflection in papers published in the Archives of General Psychiatry. Arch Gen Psychiatry 24;61: THE BONE & JOINT JOURNAL