Goniometric Reliability in a Clinical Setting

Shoulder Measurements
DAN L. RIDDLE, JULES M. ROTHSTEIN, and ROBERT L. LAMB The purpose of this study was to examine the intratester and intertester reliabilities for clinical goniometric measurements of shoulder passive range of motion (PROM) using two different sizes of universal goniometers. Patients were measured without controlling therapist goniometric placement technique or patient position during measurements. Repeated PROM measurements of shoulder flexion, extension, abduction, shoulder horizontal abduction, horizontal adduction, lateral (external) rotation, and medial (internal) rotation were taken of two groups of 50 subjects each. The intratester intraclass correlation coefficients (ICCs) for all motions ranged from .87 to .99. The ICCs for the intertester reliability of PROM measurements of horizontal abduction, horizontal adduction, extension, and medial rotation ranged from .26 to .55. The intertester ICCs for PROM measurements of flexion, abduction, and lateral rotation ranged from .84 to .90. Goniometric PROM measurements for the shoulder appear to be highly reliable when taken by the same physical therapist, regardless of the size of the goniometer used. The degree of intertester reliability for these measurements appears to be range-of-motion specific. Key Words: Physical therapy, Shoulder.

The universal goniometer frequently is used to measure active range of motion (AROM) and passive range of motion (PROM). Often, range of motion is measured on a patient many times and possibly by more than one physical therapist. Both the intratester and intertester reliabilities of ROM measurements, therefore, must be known if clinical decision making is to be made based on these measurements. Despite the frequent use of goniometric measurements, the clinical reliability of these measurements has been examined only for a few motions.1,2 The clinical reliability for PROM measurements of the shoulder complex has not been examined. Without an acceptable degree of reliability, clinical goniometric measurements are of little value and actually may be misleading.3 Most of the researchers who have examined the reliability of goniometric measurements have attempted to control variables that are not controlled ordinarily in a clinical setting.4-8 The results of these studies, therefore, may not necessarily reflect the reliability of goniometry as it is used clinically. Boone and colleagues, for example, standardized subject positions and goniometric placement techniques and used healthy subjects when determining the intratester and intertester reliabilities for AROM measurements of the upper and lower extremity joints.4 Low examined the intratester and
Mr. Riddle is Assistant Professor, Department of Physical Therapy, Medical College of Virginia, Virginia Commonwealth University, PO Box 224, MCV Station, Richmond, VA 23298. This study was completed in partial fulfillment of Mr. Riddle's master's degree, Department of Physical Therapy, Medical College of Virginia. Dr. Rothstein is Assistant Professor, Department of Physical Therapy, Medical College of Virginia, Virginia Commonwealth University, PO Box 224, MCV Station, Richmond, VA 23298 (USA). Send all correspondence to Dr. Rothstein. Dr. Lamb is Associate Professor and Director of Graduate Studies, Department of Physical Therapy, Medical College of Virginia. This article was submitted January 8, 1986; was with the authors for revision 11 weeks; and was accepted July 23, 1986. Potential Conflict of Interest: 4.

intertester reliabilities of active elbowflexionand wrist extension measurements by having 50 testers measure one healthy subject in a standardized test position.5 Two groups of researchers have examined the reliability of goniometric measurements on patients without controlling patient positions or goniometric placement techniques during measurement. Rothstein et al examined the intratester and intertester reliabilities for goniometric measurements on patients for whom knee or elbow PROM measurements were appropriate clinically.1 A high degree of reliability was found for all knee and elbow measurements with the exception of the intertester reliability of passive knee extension. A post hoc analysis revealed patient position had a significant effect on the intertester reliability for passive knee extension. This finding suggested that failure to control patient position may affect the reliability for some goniometric measurements. Hellebrandt et al used patients as subjects and multiple physical therapists as testers to examine the reliability of AROM measurements of the upper extremity.2 The intratester reliability for these measurements demonstrated a variation of 3 degrees or less for 60% of the time. The authors stated that the measurements for medial (internal) rotation and abduction of the shoulder were less reliable than other motions, but the data were not reported. The intratester and intertester reliabilities of clinical PROM measurements of the shoulder complex have not been determined for adult patient populations. The effect of goniometer size on the reliability of shoulder PROM measurements also has not been examined. A study describing the clinical reliability of these measurements would give physical therapists an indication of the usefulness of shoulder PROM measurements. The purpose of this study was to examine the intratester and intertester reliabilities of goniometric measurements of PROM for the shoulder complex on appropriate patients
PHYSICAL THERAPY

668

RESEARCH

using two sizes of goniometers in a clinical setting. We sought to answer the following research questions: 1. What is the degree of intratester reliability for the measurements obtained with large and small goniometers for seven shoulder motions (shoulder flexion, extension, abduction, horizontal abduction, horizontal adduction, lateral [external] rotation, and medial rotation)? 2. What is the degree of intertester reliability for the measurements obtained with large and small goniometers for the seven shoulder motions? 3. Are the measurements obtained with the two sizes of goniometers essentially interchangeable for each of the seven shoulder motions? METHOD Subjects Two groups of 50 patients each referred to the Department of Physical Therapy, Medical College of Virginia Hospital, Richmond, Va, were studied. Criteria for admission to the study were that shoulder PROM had been recorded in the patient's chart as a part of the physical therapy evaluation and that the patient was 18 years of age or older. All subjects read and signed a consent form before admission to the study. Subject data collected were patient diagnosis, sex, age, height, and weight. This information was obtained so that, if necessary, these variables could be used in a series of a posteriori analyses to determine whether the factors had an effect on reliability. Because measuring all the shoulder motions for reliability in one session would have been impractical, we conducted two testing sessions (Part 1 and Part 2) with each part consisting of an examination of selected shoulder motions in 50 subjects. Some patients were subjects in both testing sessions, whereas other patients participated only in one session. The subjects in Part 1 consisted of 24 men and 26 women. The mean age of these subjects was 49.0 (s = 14.5) years with a range of 21 to 77 years of age. The subjects in Part 2 consisted of 30 men and 20 women. The mean age of these subjects was 48.2 (s = 14.2) years with a range of 19 to 77 years of age (Tab. 1). Testers Goniometric measurements were made by all full-time staff physical therapists in the department who were treating adult patients. The 16 physical therapists had a mean of 6.3 (s = 3.0) years of clinical experience and had graduated from 13 different schools. Additional therapist data collected were the number of years at Medical College of Virginia Hospital and the specialty area in which the therapist practiced. This information was collected so that, if necessary, these factors could be examined a posteriori to determine whether they had an effect on the reliability of goniometric measurements. Instrumentation We usedfivelarge plastic goniometers* with 10-in movable arms and five small plastic goniometers* with 5-in movable

TABLE 1 Characteristics of the Patient Sample Variables Men Women Mean age (yr) Age range (yr) Mean height (cm) Height range (cm) Mean weight (kg) Weight range (kg) Number of right shoulders measured Number of left shoulders measured Diagnostic category 1 (postsurgical)a Diagnostic category 2 (postfracture)b Diagnostic category 3 (general orthopedics)c Diagnostic category 4 (neurological)d
a b

Part 1 (n = 50) 24 26 49.0 14.5 21-77 170.4 11.2 144.8 1 9 3 73.1 14.5 43.6 127.3 20 30 6 6 26 12

Part 2 (n = 50) 30 20 48.2 14.2 19-77 170.7 13.7 142.2 1 9 3 72.7 16.1 43.6 127.3 20 30 5 2 19 24

Postsurgical rotator cuff repair. Postfracture of humerus or scapula (nonsurgical). c Adhesive capsulitis, tendinitis, bursitis, arthritis. d Postonset spinal cord injury (quadriplegia), closed head injury, or cerebrovascular accident.

Figure. Two sizes of universal goniometers used in this study with the scales covered with white adhesive paper.

arms (Fig. 1). The scales of all of the goniometers were marked in 1-degree increments. Before the study, the accuracy of the goniometers was examined by using them to measure 10 randomly chosen, computer-generated angles between 0 and 180 degrees drawn by a graphics plotter. The angles measured by the goniometers were in agreement with the computergenerated angles. To prevent tester bias, the goniometer scales were covered with white adhesive paper before the study. Doing this prevented the therapists from seeing the values when they were taking a measurement but allowed the recorders to view the reverse side of the goniometers and to read the numbers. The goniometers were examined before each measurement session

* International Standard Goniometer, Orthopedic Equipment Co Inc, Bourbon, IN 46504. 1 in = 2.54 cm. Convacare, Inc, PO Box 19747, Raleigh, NC 27619.

7475A Plotter, Hewlett-Packard Co, 16399 W Bernardo Dr, San Diego, CA 92127.

Volume 67 / Number 5, May 1987

669

to ensure their proper functioning. During the study, two small goniometers were discarded because of slippage at the axis, and they were not replaced. Procedure The procedure used in this study was a modification of the method originally described by Rothstein et al.1 Our study was divided into two parts to minimize patient discomfort and therapist time constraints. In Part 1, PROM measurements were made of shoulder flexion, extension, and abduction. In Part 2, passive horizontal abduction, horizontal adduction, lateral rotation, and medial rotation were measured. We allowed the subjects to participate in one or both parts of the study because no comparisons were made across Parts 1 and 2. A random pair of therapists measured each subject admitted to the study. Each physical therapist participating in the study was given a unique random list of all other therapists participating in the study. When the therapists identified one of their patients as being appropriate for the study, they measured that patient. The first therapist then progressed down their list to determine the second therapist. All data were recorded by the primary author (D.L.R.) or one of the clinical supervisors. The supervisors were trained in all aspects of data recording. The recorders did not advise therapists in any way while measurements were being taken. The recording form was a modified version of the form suggested by Miller.9 The recorder estimated and noted the following for each PROM measurement: 1) subject body position (supine, prone, sitting, side lying, standing); 2) elbow position (0, 45, or 90 offlexion);and 3) whether the movement occurred in the appropriate cardinal plane. The size of the goniometer used for each measurement also was recorded. In addition, humeral position (0,45, or 90 of abduction) and forearm position (pronated, supinated, neutral) were estimated and recorded by the recorder during rotation measurements. This information was used a posteriori to identify factors that affected the reliability of the measurements. For Part 1, a physical therapist identified an appropriate patient for the study and notified a recorder to record the data. The therapist measured passive shoulderflexion,extension, and abduction in that order. All therapists used their own techniques of goniometer placement and patient positioning. All measurements were made before the patient's treatment on the day data were collected. The therapist chose whether to use the large or small goniometer for the first measurement for each motion. The two sizes of goniometers each were used twice to measure each motion. The same goniometer was not used consecutively. Each motion, therefore, was measured four times. After therapists aligned the goniometers to their satisfaction, they gave the goniometers to the recorders. The upper extremity then was brought back to the starting position by the therapist. The recorder wrote the data on the data sheet, repositioned the goniometer to the zero-degree position, and then returned the goniometer to the therapist. After all 12 measurements were taken by thefirsttherapist, the recorder notified the next available therapist from the list of randomly paired therapists. To prevent the possibility of bias, the second therapist scheduled to obtain measurements was not permitted to observe the first therapist taking measurements. The second therapist took 12 measurements, following the same order of PROM measurement and using the 670

TABLE 2 Intratester and Intertester Reliability in Part 1Measurements of Shoulder Flexion, Extension, and Abduction
Joint Motion Flexion Device large small interdevicec large small interdevicec large small interdevicec Intratester ICC a (n = 100) Intertester ICC b (n = 50)

Extension

Abduction

.98 .98 .98 .94 .94 .94 .98 .98 .98

.89 .87 .27 .26 .87 .84

a The intratester intraclass correlation coefficients were calculated by comparing the first measurements of each tester; therefore, the sample size always will be 100. b The intertester intraclass correlation coefficients were calculated by comparing the first measurements of each pair of testers; therefore, the sample size always will be 50. c The first measurements obtained with the large and small goniometers by each tester were compared.

same size of goniometers as used by the first therapist. The same protocol was followed for Part 2 except the PROM measurements of horizontal abduction, horizontal adduction, lateral rotation, and medial rotation each were obtained four times in that order. Data Analysis Intraclass correlation coefficients (ICC:1,110) were calculated to describe the degree of reliability for the relationships in the three questions we posed. Calculations of the ICCs for intratester reliability, using either the large or small goniometer, were made by comparing thefirstand second measurements taken by each tester (100 paired measurements, therefore, were obtained for each motion because 50 patients each were measured by two therapists). Calculations of the ICCs for intertester reliability, using either the large or small goniometer, were made by comparing the first measurements taken by each pair of testers (50 paired measurements, therefore, were obtained for each motion). Calculation of the ICCs for intratester reliability using the large and small goniometers were made by comparing thefirstmeasurements taken by all testers (100 paired measurements, therefore, were obtained for each motion). RESULTS For Part 1, the ICC values for the intratester reliability for the measurements of flexion, extension, and abduction obtained with either the large or small goniometer ranged from .94 to .98 (Tab. 2). The ICC values for the intertester reliability for the measurements offlexionand abduction obtained with either the large or small goniometer ranged from .84 to .89. The ICC values for intertester reliability for the measurements of extension were .27 obtained with the large goniometer and .26 obtained with the small goniometer. For Part 2, the ICC values for the intratester reliability for the measurements of horizontal abduction, horizontal adduction, lateral rotation, and medial rotation obtained with either the large or small goniometer ranged from .87 to .99 (Tab. 3). The ICC values for the intertester reliability for the measurements of horizontal abduction, horizontal adduction, and PHYSICAL THERAPY

RESEARCH TABLE 3 Intratester and Intertester Reliability in Part 2Measurements of Shoulder Horizontal Abduction, Horizontal Adduction, Lateral Rotation, and Medial Rotation Joint Motion Horizontal abduction Device large small interdevicec large small interdevicec large small interdevicec large small interdevicec Intratester ICCa (n = 100) .90 .93 .87 .95 .96 .96 .99 .98 .98 .94 .93 .93 Intertester ICCb (n = 50) .30 .28 .41 .35 .88 .90 .55 .43

Horizontal adduction

Lateral rotation

Medial rotation

a The intratester intraclass correlation coefficients were calculated by comparing the first measurements of each tester; therefore, the sample size always will be 100. b The intertester intraclass correlation coefficients were calculated by comparing the first measurements of each pair of testers; therefore, the sample size always will be 50. c The first measurements obtained with the large and small goniometers by each tester were compared.

TABLE 4 Intertester Reliability When Subject Body Position Was the Same for Both Testers Who Were Paired Joint Motion Extension Horizontal abduction Horizontal adduction na 29 31 36 Device large small large small large small ICCb .40 .44 .41 .48 .40 .32

The intertester reliability for PROM measurements of flexion and abduction was high for both sizes of goniometers, but not as high as the intratester reliability for these motions. The results indicate that whenever possible the same therapist should take repeated measurements of flexion and abduction on a patient. Although different physical therapists could measure these motions, therapists should recognize that the additional error added to the measurements could make interpretation of values difficult and could result in flawed clinical decision making. The intertester reliability for PROM measurements of extension was poor for both sizes of goniometers. The PROM measurements of shoulder extension taken by two different therapists, therefore, probably would not be clinically useful. Such measurements may be misleading because they contain considerable measurement error. An a posteriori analysis demonstrated that for only 29 of the 50 pairs of measurements was the same patient body position used by both testers for extension measurements. The same patient body position was used by both testers for almost all pairs of measurements of flexion and abduction. An a posteriori ICC revealed that when subject body position was the same for both testers who were paired, the ICC values for extension still were poor (Tab. 4). A failure to control body position probably did not contribute to the poor intertester reliability for the PROM measurements of extension. Extension may not be measured frequently in the clinic, which could have contributed to the poor intertester reliability. Although the cause of the poor intertester reliability for measurements of extension cannot be identified from our data, the results indicate that the same physical therapist should take all repeated PROM measurements of extension on a patient. An a posteriori analysis was performed to examine the effect of patient diagnostic category on reliability (Tab. 5). Patient diagnostic category did not appear to affect intratester reliability. For intertester reliability, the number of subjects in each category was too small for a meaningful analysis, but this relationship warrants further study with a larger sample of patients. Part 2 Clinical PROM measurements of shoulder horizontal abduction, horizontal adduction, lateral rotation, and medial rotation were highly reliable when repeated by the same therapist. Intratester reliability was unaffected by the size of the goniometer used. The intertester reliability for PROM measurements of horizontal abduction and horizontal adduction was poor for both sizes of goniometers. Repeated measurements of horizontal abduction and horizontal adduction, therefore, always should be made by the same therapist. An a posteriori analysis revealed that when subject body position was the same for both testers who were paired, the ICC values for horizontal abduction and horizontal adduction still were poor (Tab. 4). This observation suggests that subject body position was not a factor that reduced intertester reliability. We can hypothesize only that poor intertester reliability resulted because horizontal abduction and horizontal adduction may not be measured frequently in the clinic. The intertester reliability for PROM measurements of lateral rotation was high, but not as high as the intratester reliability for these measurements. Different therapists still can measure the same patient if physical therapists recognize 671

a Number of therapist pairs in which both testers used the same subject body position during measurement. b ICC = intraclass correlation coefficient.

medial rotation obtained with either the large or small goniometer ranged from .28 to .55. The ICC values for the intertester reliability for the measurements of lateral rotation were .88 obtained with the large goniometer and .90 obtained with the small goniometer. DISCUSSION Part 1 Clinical PROM measurements of shoulder flexion, extension, and abduction obtained with the universal goniometer were highly reliable when repeated by the same physical therapist. The measurements obtained with the two sizes of goniometers essentially were interchangeable and of equal intratester reliability. Rothstein et al reported similar findings for clinical PROM measurements of the knee and elbow.1 Our results and those of Rothstein et al do not support the recommendations of Moore,11 Esch and Lepley,12 and Norkin and White13 that large goniometers should be used on joints with long bony segments.
Volume 67 / Number 5, May 1987

TABLE 5 Intratester and Intertester Reliability for the Four Diagnostic Categories in Part 1 Diagnostic Category Postsurgical Postfracture General orthopedics Neurological
a

Type of Reliability intratester intertester intratester intertester intratester intertester intratester intertester

na

Flexion Large
b

Extension Small .99 .97 .98 .55 .99 .94 .97 .83
c

Abduction Large 1.0 .89 .98 .71 .98 .87 .99 .94 Small 1.0 .87 .91 .59 .99 .85 .99 .91

Large .97 0 .88 .28 .95 .36 .89 .33

Small .98 0 .86 .25 .96 .31 .89 .46

12 6 12 6 52 26 24 12

.98 .97 .99 .68 .98 .94 .97 .83

For intratester reliability, the first and second measurements for each tester were correlated; therefore, the sample size always will be twice that of intertester reliability. b Large goniometer. c Small goniometer.

TABLE 6 Intratester and Intertester Reliability for the Four Diagnostic Categories in Part 2 Diagnostic Category Postsurgical Postfracture General orthopedics Neurological Type of Reliability intratester intertester intratester intertester intratester intertester intratester intertester na 10 5 4 2 38 19 48 24 Horizontal Abduction Largeb .96 0 1.0 0 .93 .67 .86 .17 Smallc .92 0 .96 0 .93 .72 .93 .10 Horizontal Adduction Large .96 .43 .92 0 .93 .47 .95 .41 Small .99 .40 .96 0 .95 .39 .96 .35 Lateral Rotation Large .93 .78 1.0 0 .99 .93 .98 .85 Small .96 .72 .99 0 .99 .93 .98 .88 Medial Rotation Large .86 0 .52 0 .97 .86 .92 .20 Small .95 0 .82 0 .97 .79 .88 .03

a For intratester reliability, the first and second measurements for each tester were correlated; therefore, the sample size always will be twice that of intertester reliability. b Large goniometer. c Small goniometer.

that additional measurement error will be added to these measurements. The intertester reliability for PROM measurements of medial rotation was poor. Repeated measurements of medial rotation, therefore, always should be made by the same therapist. The same subject body position was used by both testers in all 50 pairs of medial rotation measurements. The same subject body, elbow, humeral, and forearm positions were used by both therapists for 43 pairs of measurements. These data strongly suggest that factors other than patient position reduced intertester reliability for medial rotation measurements. Although the cause of the lack of intertester reliability for these measurements was not identified in this study, we question whether therapists may vary in the way they control scapular motion during PROM measurements of medial rotation. An a posteriori analysis was performed to determine the effect of patient diagnostic category on reliability (Tab. 6). As in Part 1, diagnostic category did not appear to affect intratester reliability. The number of subjects was too small to allow any conclusions regarding the effect of diagnostic category on intertester reliability. In the design of this study, we attempted to replicate what occurs during actual clinical use of goniometric measure672

ments. The nature of a reliability study, however, requires that more repeated measurements be obtained than are likely to be taken during clinical practice. Based on our examination of the pattern of our results, we do not believe that the results were caused by the effect of having subjects measured multiple times. This study was conducted in one clinical setting, so the results may not be generalized necessarily to all clinical settings. We found, however, as suggested by Rothstein et al,1 that the determination of goniometric reliability is feasible even in a busy department. Clinicians, therefore, can determine goniometric reliability in their own clinical settings. This study examined the clinical reliability for shoulder PROM measurements, and the study of Rothstein et al1 examined the reliability for PROM measurements of the elbow and knee. We found no studies that examined the clinical reliability for PROM measurements of the wrist, hand, hip, foot, or ankle joints. The clinical reliability for these measurements must be determined if clinical decision making is to be based on these measurements. This study demonstrated that PROM measurements of the shoulder can be highly reliable when taken by the same physical therapist. The degree of intertester reliability was ROM specific. Reliability was unaffected by the size of the goniometer used.
PHYSICAL THERAPY

RESEARCH

CONCLUSIONS

Based on our study of two groups of 50 patients each in a clinical setting, we conclude that PROM measurements of the shoulder taken by the same therapist can be highly reliable, regardless of whether a large or small goniometer is used. The PROM measurements of flexion, abduction, and lateral rotation, obtained with either a large or small goniometer, can be reliable when different physical therapists measure a patient, but probably are not as reliable as when the same

physical therapist takes these measurements. The PROM measurements of extension, horizontal abduction, horizontal adduction, and medial rotation, obtained with either a large or small goniometer, probably are unreliable when different physical therapists measure a patient, even when the patient's body position is controlled. The PROM measurements obtained with the large or small goniometer essentially can be interchangeable without affecting reliability for each of the seven shoulder motions.

REFERENCES
1. Rothstein JM, Miller PJ, Roettger RF: Goniometric reliability in a clinical setting: Elbow and knee measurements. Phys Ther 63:1611-1615, 1983 2. Hellebrandt FA, Duvall EN, Moore ML: The measurement of joint motion: Part 3. Reliability of goniometry. Phys Ther Rev 29:302-307, 1949 3. Rothstein JM: Measurement and clinical practice: Theory and application. In Rothstein JM (ed): Measurement in Physical Therapy. New York, NY, Churchill Livingstone Inc, 1985, vol 7, chap 1 4. Boone DC, Azen SP, Lin C-M, et al: Reliability of goniometric measurements. Phys Ther 58:1355-1360, 1978 5. Low JL: The reliability of joint measurement. Physiotherapy Canada 62:227-229,1976 6. Mayerson NH, Milano RA: Goniometric measurement reliability in physical medicine. Arch Phys Med Rehabil 65:92-94, 1984 7. Loessin Grohmann JE: Comparison of two methods of goniometry. Phys Ther 63:922-925, 1983 8. Pandya S, Florence JM, King WM, et al: Reliability of goniometric measurements in patients with Duchenne muscular dystrophy. Phys Ther 65:1339-1342,1985 9. Miller PJ: The assessment of joint motion. In Rothstein JM (ed): Measurement in Physical Therapy. New York, NY, Churchill Livingstone Inc, 1985, vol 7, chap 4 10. Shrout PE, Fleiss JL: Intraclass correlations: Uses in assessing rater reliability. Psychol Bull 86:420-428, 1979 11. Moore ML: The measurement of joint motion: Part 2. The technic of goniometry. Phys Ther Rev 29:256-264, 1949 12. Esch D, Lepley M: Evaluation of Joint Motion: Methods of Measurement and Recording. Minneapolis, MN, University of Minnesota Press, 1974, pp 8-9 13. Norkin CC, White DJ: Measurement of Joint Motion: A Guide to Goniometry. Philadelphia, PA, F A Davis Co, 1985, pp 12-16

Volume 67 / Number 5, May 1987

673