Kinematics of shoulder abduction in the scapular plane

Clinical

Kinematics of shoulder scapular plane.

abduction

On the influence

velocity

1Michiels

Dr med,

Orthopadische

Klinik

of abduction

J Grevenstein der Johannes

Vol. 10, No. 3, pp. 137- 143, 1995 Copyright @ 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0268~Otl33/95 $10.00 + 0.00

Biomechanics

in the

and external

load

Dr med Gutenberg-Universitat,

Mainz,

Germany

Summary In this paper the kinematics of arm abduction in the scapular plane of 38 healthy experimental subjects is reported. The ratio of the glenohumoral and the scapulothoracic components of the motion were determined and the influence of the abduction speed and of the external load on it were investigated. The investigation of the effect of abduction speed involved all 38 subjects. Each of them performed one slow and one fast abduction. Statistical analysis showed that there were large differences between individuals, but that for any one individual the abduction process is essentially reproducible. For the one individual there is a strong,linear relationship between glenohumeral and scapulothoracic rotation. The slope of the regression of the glenohumeral component on total arm abduction varied from 0.75 to 0.5 with a sample mean of 0.66. It means that only two-thirds of arm abduction occurs in the glenohumeral joint, the remaining third taking place via scapular rotation. In slowly performed abductions the slope of the regression was significantly greater than in the high-speed movements, but differences were very small. In the investigation of the effect of external load, statistical analysis indicated that the slope of the intraindividual regression is largely independent of the load.The possibility of observer bias was analysed, too. It was found that the standard deviation of the abduction parameters determined by different observers was between 6 and 10% of that between different experimental subjects. Relevance A knowledge of the kinematics of the shoulder joint is necessary to understand subacromial pathology and in particular impingement problems. The scapulohumeral rhythm depends on the balanced and coordinated function of the muscles involved. The activation pattern of the abductor muscles, as reflected in shoulder kinematics, seems to be individual and stored as an engram. Key words: C/in.

Kinematics,

Biomech.1994;

shoulder Vol.

abduction,

10: 137-143,

scapulohumeral

rhythm,

Early kinematic investigations of the shoulder complex suggested a decoupling of the glenohumeral (GH) and scapulothoracic (ST) joints, although some authors recognized early that the mobility of the scapula must play an important role in arm abduction. The integrated, coupled and interdependent motion of the Received: 2 December 1992 Accepted: 18 April 1994 Correspondence

and

reprint

to: Dr. med. I Michiels, Essen,, Hufelandstr. 55D-45122

requests

plane,

shoulder

load

April

Introduction

Orthopldische Universitltsklinik Essen, Germany

scapular

humerus, scapula, and clavicle was described 1934 by Codman by the term ‘scapulohumeral rhythm’*. Although there is a total of seven joints involved in the shoulder complex3, most work on this problem has been restricted to the relationship between glenohumeral (GH) and scapulothoracic (ST) rotations4-5. The analysis of Inman et a1.6 indicated a proportional relationship valid over the whole abduction range. However, the experimental work of others provided evidence for departures from linearity’-lo. Doody et a1.8 found evidence that the state of muscle contraction can affect the flow of the movement.

138

C/in. Biomech.

1995; 10: No 3

All the above-mentioned experiments were based on sequences of static X-ray exposures or of static clinical measurements in well-defined positions. In order to obtain each exposure the experimental subject must be positioned, and remain fixed in that position for its duration. In such studies it is impossible to rule out the theoretical possibility of the ‘setting’ of the scapula into an unnatural posture. Because of the shortcomings and uncertainties of previous studies we carried out a kinematic analysis of the shoulder complex in the scapular plane and formulated following questions: (1) How do the glenohumeral and scapulothoracic components of shoulder abduction in the scapular plane behave during a continuous movement? (2) Is the ‘scapulohumoral rhythm’ dependent on (a) the speed of abduction, or (h) an external mechanical load? Methods Sequences of X-rays of the shoulder complex of healthy volunteers without any history of trauma or impairment of the shoulder were obtained during abduction in the scapular plane using a Siemens Polytron loo0 VR real-time image digitizer and processing system. This particular machine has the advantage that exposure of the subjects to radiation is between six and seven times less than with conventional X-ray techniques”. For most of the experiments the imaging rate was 1.92 s-’ (0.52 s between images). For a few abduction series, rates of 3.1 :;-I or 1 s- i were used (0.32-s or l-s intervals respectively). The exact time difference being mentioned on each image, it was possible to calculate the angular velocity in degree s-‘. Because of the limited vertical range of movement of the machine the experimental subjects had to sit down for the investigations. Otherwise they were positioned as described by Freedman and Munro7 and as outlined in Figure I. The image sequences were evaluated by the method described by the same authors. On a magnified positive reprint of each X-ray image,three straight lines and three angles between them were determined (Figure 2). The three angles are not interdependent, as AO, = hOon t- AC&,. However. in order to detect

Figure 2. Analysis of the X-rays requires the definition of three straight lines building three interdependent angles (AOnn = AC&., + ACQr). The body axis is parallel to the edge of the image. The glenoid tilt is determined by a line joining the most cranial and caudal points of the glenoid cavity. The third line is the humeral shaft axis. Onn, arm abduction angle; @or.,, glenohumeral angle; Osr, scapulothoracic angle.

possible sources of error all three angles were measured. Then for each abduction series a regression analysis of @on ver.suS OAA was performed. The parameters calculated included the slope as well as the intercept of the regression line and the squared correlation coefficient. The computed regression line may be represented by the equation: OGH = a + b x OAA

where a represents the intercept and b the regression slope. The influence of abduction speed and of external load was investigated in the following manner: Influence of abduction speed (Measurements

Figure

1. Positioning

of the experimental

subject.

1 to 76)

A group of 38 experimental subjects was asked to carry out a slow and then a fast abduction in the scapular plane, both of them flowing movements. The subject was allowed to choose both the speed of the abduction and the maximum abduction angle for several reasons. Firstly, keeping to a prescribed speed might influence the movement by unnatural muscular braking or accelerating corrections. For similar reasons it was decided not to guide the abducted arm between two vertical rails as described by Bagg and Forrest”.

Michiels

and

Grevenstein:

Kinematics

of arm

abduction

Table 1. Statistics of the angular range in degrees in the 76 image sequences investigating the influence of abduction speed (a) and in the 15 image sequences investigating the influence of external load (b); all values in degrees Arm

abduction @AA

a, Abduction Minimum Maximum Average b, External Minimum Maximum Average

speed

Glen-Hum

Scap-Thor @ST

@G/f

Maximal abduction angle

experiments

2 23 10.11

-14 21 5.96

-14 23 4.12

110 174 152.25

-9

118 148 138.27

load experiments 5 14 10.2

4 20 10.67

7 -0.60

Secondly, if the abduction is forced to reach an almost vertical position of the arm it might lead to movement of the trunk. Although a lateral deviation of the trunk can be avoided by performing a bilateral abduction, it has to be kept in mind that final abduction is accompanied by extension of the spine.The number of images per series for this group varied between 3 and 14 (mean: 6.43; median: 6), 489 images were analysed. Influence of external load (Measurements

77 to 92)

A second group of five experimental subjects was used to study the influence of external load on the course of the movement.All subjects were right-handed. Each subject carried out the abduction movement three times: without load, with a l-kg load, and with a 2-kg load. The mean number of images per series was 10.93 (range 3-15, median 10) in all 164 images. Table 1 gives the ranges of variation of the three measured angles for each of the experiments. For the abduction speed measurements (A) the abduction angle was approximately 150”, and for the external load measurements (B) it was approximately 140”. It may be deduced that in series A the change in abduction between successive images was approximately 25”, and in series B approximately 1.5”. Investigator-related variability

To assess observer bias the images of five abduction movements, chosen at random from the experiments A and B, were independently evaluated by five investigators. Each of the investigators determined the angles OAA, OST, and 0 GH as described above. From these values the regression line relating OGH to OAA was further calculated for each of the series. Results influence of abduction speed (Measurements

1 to 76)

The abduction series of the 36 investigated subjects are represented in Figure 3, where the glenohumeral part is

Figure 3. The abduction series in the velocity experiments, plotting OGH against OAA. a, high-speed abductions; b, low-speed abductions. Upper part, the joined measuring points; lower part, the corresponding regression lines.

139

140

C/in.

Biomech.

1995;

10: No 3

fable 2. Statistics of angular speed in degrees SC’ for the 76 image sequences investigating the influence of abduction speed _.-.._ _.---

Ali (n .:= 76) Fast in = 381 Slow in = 38)

Mean .._.. 52.17

SD

70.00

20.38

34.34

11.77

Median

Min -

Max

11.50

105.13

67.79

35.25

105.13

32.39

11.50

64.10

plotted against arm abduction. The first half (Figure 3a) contains the abductions performed at ‘high’ speed for all 36 subjects and Figure 3b their corresponding ‘low' speed abductions. In both figures the upper part delineates the joined measuring points and the lower one the computed regression lines of @ok on @A,\. The most obvious and important conclusion of the statistical data analysis is the very high correlation for each of the calculated linear regression lines in the low-speed group as well as in the high-speed group. The correlation coefficient was in all cases in the range 0.9851-0.9999, with an average value of 0.9968, and in 73 of the 76 measurements it was above 0.99. Table 2 shows the statistics for the velocities measured in the two types of experiment. For the low-speed series. the range of the mean angular velocity over the whole abduction movement was 11.5” s- i to 64.10” s--I with an average of 34.34” SC’. The average in the high-speed series was 70.00” s -.‘. i.e. twice that of the low-speed series. In each of the speed groups the average and median speed were essentially the same, supporting the assumption of symmetry of the underlying distribution. In all cases the mean abduction speed may be considered to lie in the physiological range, since for throwing sportsi it may be as great as 2000 degree s- i. Table 3 lists the corresponding statistics for the intercepts (a) and slopes (b) of the regression lines for both speed groups. The slope of the regression line gives the increase in GH rotation by 1 degree of total AA. Hence expressed in per cent (times 100) it may be Table 3. Statistics of the regression intercept for the 76 image sequences influence of abduction speed Mean .._-- ._..... - -... _I-

SD

Median

line slope investigating

and the

considered to represent the fraction of the GH part to total abduction. Statistical investigation of the effect of abduction speed on slope and intercept was performed using the Wilcoxon signed-rank test. For the intercepts no significant difference between high- and low-speed abductions in the same subject could be revealed (P = 0.4982). On the other hand comparison of the slopes of the two groups reveals a statistical difference (P = 0.0356), but the relevance of that difference (mean -0.016) seems questionable. Influence of external load (Measurements

77 to 92)

The 15 abduction series used to study the effect of external load are represented in Figure 4. Here again the upper part shows the joined measuring points and the lower part the corresponding regression lines. These measurements were also evaluated starting from a regression analysis of the interindividual slope of the OGH versus OAA p lot. All correlation coefficients were greater than 0.98, demonstrating that the GH component stays in a constant relationship to AA throughout the whole abduction movement. Using Hotelling’s one-sample test for comparing the slopes obtained with zero, l-, and 2-kg load respectively,the P value was calculated to be 0.2509, indicating that the characteristics of the abduction are essentially not influenced by the external load. In addition it was not

i*c ,

, 2

-_____---

Min

,,’

Max

intercept a Ail (rr ;- 76) Fast (n -- 38) Slow In 38'1

1.222

9.1 f

0.720

--lg.874

20.833

:.139

8.62

0.526

-- 17.771

18.666

1,305

9.69

0.720

-19.874

20.833

0.662

0.055

0.662

0.489

0.809

0.654

0.056

0.658

0.489

0.766

0.670

0.054

0.675

0.586

0.809

Slope b A!: in 2: 76) Fast in =-G381 Slow in= 38)

Figure4. The abduction series in the loading experiments, plotting OcH against OAA. Upper part, the joined measuring points; lower part, the corresponding regression lines.

Michiels and Grevenstein:

even possible to detect a trend since in the three distinct situations the minimal values for the slope amounted to 0.6017, 0.6061, and 0.5154 and the maximal values to 0.7371, 0.8089, and 0.7323 respectively (mean values 0.65362, 0.67004, and 0.62736). Investigator-related

variability

To characterize the variability, the variances of intercept and slope, both investigator-independent and investigator-related, were calculated using the variance components estimation procedure. For intercept and slope we obtained 133.7116, and 0.0051 respectively as the investigator-independent component of variance, and 8.5759 and 0.0005 respectively as the investigator-dependent component. It may be concluded that the investigator is responsible for at most one-tenth of the observed variability. Hence essentially all of this variation is real, and the reproducibility of the measurement method is adequate. Discussion The results of the present investigation into the relationship between the GH and ST components of arm abduction differ from the conclusions of previous investigator& “*i3 in some way. There are several possible explanations for these discrepancies. Firstly the analysed abduction movements were not always performed in the same plane. For example Inman et a1.6 and Saha’” chose the frontal plane, although most authors have followed Johnston14 in choosing the scapular plane. Secondly Doody et al.* and Bagg and Forrest” proposed that differences could be the result of the different positions measured and the various intervals between these positions. Thirdly there certainly exists a variability between individuals’,*. A fourth discrepancy may be introduced by the methods used: X-ray analysis”,7,‘3 versus clinical measurements after localization of anatomical bony landmarks on the skin**“. In the light of these differences it is necessary to consider a number of possible systematic sources of error: Measurement error. The location of anatomical features in the X-ray image or on the body must lead to investigator-dependent error. These errors may certainly be greater if based on the localization and measuring of bony landmarks on the skin. For our X-ray measurements we have demonstrated that this source of error is smaller by a factor of 10 compared with the natural variability of the measured parameters. Patient positioning. (a) Most authors consider that scapular rotation takes place in a well-defined, immovable plane. However, this scapular plane is subjected to tilting since the scapula glides cranially and laterally along the tapering thorax9,10,‘5. Habermeyer claims that the spatial orientation of the scapula varies

Kinematics

of arm abduction

141

not only between individuals but also in a single individual, can be very different for different states of muscular activity, and thus may be difficult to define’“. However, Schatz was able to prove, using X-ray analysis, that tilting his shoulder blade preparations by 20”, and thus changing the scapular planecaused at most a 4” tilt deviation of the glenoid socket plane as measured on the X-ray projection’7. The deviations caused by tilting the scapula are therefore small and are indistinguishable from measurement error. (b) Tilts of the spinal column are not registered by the method. Leaning away from the abduction side can be avoided by a symmetrical movement, and straightening out of the breast kyphosis by not carrying through to terminal abduction. Since we carried out the abduction with the palm oriented caudally, the external rotation of the humerus, necessary for the terminal abduction phase”, was not present. Correspondingly the measurement range was restricted in some cases, the mean value of maximal abduction over all 91 measurements being 152.25 degrees. The regression analyses carried out in this investigation gave two values for each sequence of images: the intercept a and the slope b (the coefficient of OAA) that represents the per cent part of the GH component to total arm abduction AA. A comparison of the results in the velocity experiments indicates that for both parameters the interindividual scatter is much larger than the intraindividual value. The relative intraindividual constancy is apparent when the slow abduction results are compared with those for fast abduction (Table 3). Both variables in the two groups and especially the slope b have comparable scatter and comparable values. There is hardly any discussion in the literature of the significance of the intercept a for the abduction movement. We believe that it reflects the ‘rest position’ of the shoulder complex at the time of the experiment, which will be influenced by many factors and hence may be variable at different measuring sessions. The decisive factor is the position of the scapula, which in turn is not only predominantly influenced by the actual activation status of the shoulder girdle musculature15 but may be influenced also by the anatomical configuration of thorax and muscular tissues (i.e. in heavily built and muscular subjects). The variability of OoH in our experiments (from -14 to +23”) reflects these individual differences. Additionally the rest position of the humerus cannot be equated with an abduction angle 0 AA = 0”. The smallest angle between the humerus and the vertical in our set of measurements was admittedly only 2”, but was on average 10.1” (Table 2). The slope (gradient of OoH versus OAA) gives the fraction of the total abduction represented by the GH rotation. Although in this case the interindividual range (from 49 to 81%) is relatively large, the intraindividual differences are quite small. Statistical difference between fast and slow abduction movements is obvious (P = 0.035) but differences are very small, with a

142

Clin.

Biomech.

1995;

10: NO 3

mean value of 0.016 (1.6%). It has to be taken into account that high and low speed were not clearly defined but were chosen by the subject without any more detailed instructions or handicaps. Nevertheless for each subject the velocity of the high-speed abduction was higher than that of the low-speed one, but the highest velocity in the low-speed group (64.10 degrees s-- “1 was nearly twice as fast as the slowest one in the high-speed series (35,25 degrees s-l). Therefore, and since the difference is in the order of a few per cent, the statistical evidence that slower abduction movements have higher slopes has to be interpreted with caution.The calculation of the regression of slope against computed speed for all measurements could be another possible way of expressing their dependency. Nevertheless this procedure considers the 76 abduction samples as not being interrelated and therefore is from the point of view of the statistical analysis debatable. But if interpreted with caution, it could be justified since the differences between fast and slow abductions were minimal. The result in this case was: y = 0.682 3.86 x 10. ’ x x (x = the abduction speed in degrees per second, y = the slope of the plot of GH versus AA). Hence this criterion also could indicate that the slope b is essentially constant, whatever the abduction speed. The value of the residual mean square error parameter R* for this regression (R’ = 0.029) confirms this independence. it seems that each experimental subject performs an arm abduction in his or her own personal way. Under the range of conditions that we have measured, the intraindividual variability is so small that the analysis of two abduction positions for a given subject suffice completely to describe his personal pattern. The two points define the regression line and hence the whole sequence of the movement. Additional measuring points would bring no further information. This follows from the extremely high value for the correlation coefficients, which is in no case was smaller than 0.985. Arguing from these very high correlations it does not appear sensible to divide the abduction movement into early and late phases, as Inman et al.h have done. although it must be admitted that the terminal phase of abduction was not included in our experimental set-up, the mean maximal abduction angle being 152.25 degrees. In some individual movements, irregularities of the scapula rotation were detected in the form of short-term, small, counterrotations. This phenomenon was already described within the first 30” of abduction and called the ‘setting phase of the scapula’6,‘s. During this phase the scapula and humerus are seeking a position of maximal congruence (and hence maximal stability) and in this way converge on an axis in the scapular plane. Nevertheless, in our set of measurements these irregularities are at most 1” and hence are indistinguishable from experimental error. It remains unclear whether these irre@arities are typical for a given subject and would be observed constantly, or are only incidental. Statistically, and considering the

high correlation of the regression lines, they are meaningless. A further analysis and subdivision of the available data, like that carried out by Bagg and is therefore not sensible either. The Forrest’“, systematic errors of the method, which may amount to a few degrees, do not allow the small irregularities to be described as significant. The typical abduction mode for an individual subject does not seem to be influenced by external load either. The differences between different load conditions are not only statistically insignificant (P = 0.25), but not even a consistent trend with increasing load was evident. Therefore we cannot confirm the observation by Johnstoni that the rotation of the scapula occurs earlier when the abduction is performed under load. On averaging over the whole set of measurements it turns out that only two-thirds of the arm abduction occurs in the glenohumeral joint, the remaining one-third taking place via scapular rotation. For a distinct experimental subject it has been possible to demonstrate a fixed, possibly engrammatically stored, relationship between the rotations of the scapula and humerus, which is not influenced by the abduction speed or the externally applied load. The abduction sequence in the scapular plane proceeds as if there were a mechanical coupling between the humerus, scapula, and thoracic wall. However, this coupling takes place exclusively via muscular control. Codman* 50 years ago concluded that the movement at the shoulder girdle depends on the absolute coordination of the muscles involved and his observation gains significance in the context of our experiments. This strong coupling is even more amazing, as the rotation of the scapula constantly changes the operating environment of the supraspinatus and deltoid muscles. the two main abductors involved in the studied movements. Conclusions

1. For the one individual there is a strong, probably between engrammatically stored, coupling glenohumeral and scapulothoracic rotation. 2. However, there is considerable variation between individuals. 3. Averaged over a large group of experimental subjects, about *h of arm abduction takes place in the glenohumeral joint and about I/3 via 4. This individual rotation of the scapula. scapulohumeral rhythm does not seem to be influenced by abduction speed or by external load within our measuring range. 5. The statistical analysis of the available data does not support a subdivision of the abduction sequence into distinct phases. Acknowledgements

The authors are very grateful to Prof. Dr M. Thelen, head ofthe Radiological Institute of our university, and to Prof. Dr H.-H. Schild and Miss Nadge, his co-workers, for their support in performing our radiological investigation. Statistical analysis of the

Michiels

data would have been impossible without the help of PD Dr S Wellek, Institute for Medical Statistics and Information. References

1 Braune W, Fischer 0. iiber den Anteil, den die einzelnen Gelenke des Schultergiirtels an der Beweglichkeitdes menschlichen Humerus haben. Abh der math-phys cl d KSachs Gesellsch der Wiss 1888; 24: 395-410 2 Codman EA. The Shoulder: Rupture of the Supraspinatus Tendon and Other Lesions In or About the Subacromial Bursa. Thomas Todd Co., Boston 1934 3 Cailliet R. Shoulder Pain. F.A. Davis Co, Philadelphia

1981 4 Flecker H. Roentgenographic study of the movements of abduction at the normal shoulder joint. Med JAust 1929; 122-3 5 Fisk GH, Colwell G. Shoulder movements in health and disease. Arch Phys Med Rehabi119.54; 35: 149-5.5 6 Inman VT, Saunders JB, Abbott LC Observations on the function of the shoulder joint. J Bone Joint Surg 1944; 26: l-40 7 Freedman L, Munro RR. Abduction of the arm in the scapular plane: scapular glenohumeral movements. J Bone Joint Surg 1966; 48-A: 1503- 10 8 Doody SG, Freedman L, Waterland JC. Shoulder movements during abduction in the scapular plane. Arch Phys Med Rehabill970; 51: 595-604 9 Poppen NK, Walker PS. Normal and abnormal motion of the shoulder. J Bone Joint Surg 1976; 58A: 195-201

and Grevenstein:

Kinematics

of arm abduction

143

10 Bagg SD, Forrest WJ. A biomechanical analysis of scapular rotation during arm abduction in the scapular plane. Am J Phys Med Rehabill988; 67: 238-45 11 Michiels I, Mohr W, Schild H-H et al. Die Riintgendigitalanlage als Instrument zur kinematischen Untersuchung der Schulter - Eine Untersuchung zur Strahlenbelastung. Biomed Tech (Berlin) 1991; 36: 149-52 12 Bonci CM, Hensal FJ, Torg JS. A preliminary study on the measurement of static and dynamic motion at the glenohumeral joint. Am JSports Med 1986; 14: 12-17 13 Saha AK. Theory of shoulder mechanism: descriptive and applied. Springfield, Illinois, CC Thomas, 1961 14 Johnston TB. The movements of the shoulder joint. A plea for the use of the ‘plane of the scapula’ as the plane of reference for movements occurring at the humero-scapular joint. Br J Surg 1937; 25: 252-60 15 Laumann U. Kinesiology of the shoulder electromyographic and stereophotogrammetric studies. In: Bateman JE, Welsh RP, ed. Surgery of the Shoulder. BC Decker, Philadelphia, 1984 16 Habermeyer P Isokinetische Krgfte am Glenohumeralgelenk. Hefte zur Unfallheilk 1990; 202: 1-114 17 Schiitz M. Die Bestimmung des vertikalen Neigungswinkel der Cavitas glenoidalis und dessen Bedeutungfir die Periarthropathia humeroscapularis. [Inaugural Dissertation] Frankfurt am Main, 1984 18 Gagey 0, Bonfait H, Gillot Cl, Mazas F. Anatomie fonctionnelle et m6canique de 1’ClCvationdu bras. Rev Chir Orthop 1988; 74: 209-17