Reliability of passive wrist flexion and extension goniometric measurements: a multicenter study.

Wrist injuries frequently cause secondary limitation of motion at the wrist. The goniometer is used to measure wrist active range of motion (AROM) and passive range of motion (PROM) for documentation purposes and to assist in making clinical decisions. Many of these clinical decisions are based on how much wrist range of motion (ROM) a patient needs for his or her activities of daily living (ADL) and return-to-work requirements. When wrist PROM is limited, therapists utilize a wide range of passive treatments to address this restriction. These treatments include, but are not limited to, PROM stretches,[1] joint mobilization,[2] dynamic splinting,[3] static progressive splinting,[4] and serial casting.[5] To assess the effectiveness of these passive treatments, a passive measure, such as goniometry, must be used. Does a change in a goniometric reading, however, indicate an actual change in a patient's passive joint motion, or is the measurement device unreliable or invalid? Only with a reliable and valid ROM measuring instrument will a therapist know whether a passive treatment is efficacious.[6]

Multiple goniometric techniques are used for measuring ROM of wrist flexion and extension. These techniques differ in respect to placement of the goniometric arms. For example, the American Medical Association[7] and the American Society of Hand Therapists[8] suggest volar and dorsal approaches to the measurement of extension and flexion, respectively, whereas ulnar[9-11] and radial[10,12] goniometric alignments are also suggested in the literature. Consequently, with therapists using different ROM measuring techniques, critical comparisons of treatments that affect ROM are impossible, as it is unknown how measurement results vary from one goniometric technique to another.

Although functional goals are paramount, goniometric results certainly influence treatment decisions. For instance, a goniometric measurement taken on the ulnar side of the wrist may satisfy a ROM goal for ADL, but a radial or volar/dorsal measurement may not meet that ROM criterion. The clinician who uses the ulnar goniometric approach may discontinue attempts to gain more motion, whereas the clinician who uses the other measuring techniques may continue to try and regain motion. This potential difference is goniometric results may have significant clinical implications, particularly in determining functional ROM after palliative/stabilizing procedures such as partial wrist fusions,[13] total wrist arthroplasties,[14] and alignment reconstructive procedures at the wrist.[15]

The reliability of goniometric measurements of joint motion has been assessed under many conditions for various joints.[16-22] Most researchers have either attempted to control variables that are not typically controlled in a clinical setting or used subjects without any pathology. Their results, therefore, have limited application to the clinician. Rothstein et al[23] have provided a protocol for the study of intrarater and interrater reliability of goniometric measurements under clinical conditions.

Two studies of wrist goniometric reliability have been performed,[24,25] with only one study having been done under clinical conditions.[24] In both studies, reliability of measurements obtained with the ulnar and radial measuring techniques was assessed. The results of these studies were conflicting, as greater reliability was seen with an ulnar technique in one study and with the radial technique in the other study. Horger[24] found high intratester and intertester reliability, especially when an ulnar technique was used. The most reliable measurements in that study were obtained by a variation of the previously described ulnar technique,[9-11] which used the third metacarpal for goniometric alignment rather than the more traditionally used mobile fifth metacarpal. Duffin and Zoeller[25] found greater reliability with the radial rather than the traditional ulnar technique.

Heretofore, there has been no comparison of all three goniometric techniques in terms of measurement results and reliability. It was imperative that all measuring techniques be assessed to determine which goniometric technique had the greatest reliability. Therefore, this study was designed to ensure that all techniques were evaluated within a clinical environment. The purpose of this study was twofold: (1) to determine whether passive wrist flexion and extension goniometric measurements obtained using radial alignment, ulnar alignment, and volar/dorsal alignment were similar or dissimilar and (2) to examine which of these three PROM wrist flexion and extension goniometric techniques had the greatest intratester and intertester reliability.

Method

Subjects

One hundred forty patients (141 wrists) from eight hand/upper-extremity clinics around the United States (Appendix) participated in this multicenter study. Patients were included in the study if they were referred to one of the eight clinics and if wrist PROM would normally have been included in their assessment. All subjects read and signed an informed consent statement before admission to the study. Subject data collection included each patient's age, sex, and diagnosis. Whether the subject was treated prior to or immediately after wrist PROM was measured was also noted. Although each clinic was required to collect data on 25 subjects, five of the eight clinics collected data on 21 or fewer subjects (Tab. 1). The data on one subject could not be deciphered from the recording form and were discarded. The average number of subjects per clinic was 17.5 (range = 7-25). The average age of the subjects was 41.5 years (range = 6-81). An average of 65% (range = 48%-85%) of the subjects were male, and an average of 35% (range = 15%-50%) were female (Tab. 1). All data collection was done in one clinic session.

[TABULAR DATA OMMITED]

Testers

Goniometric measurements were obtained by 32 therapists from eight different hand/upper-extremity clinics (4 therapists per clinic). For inclusion in the study, each clinic was required to have 4 therapists present at one time to allow randomization of testers. In four clinics, 1 of the measuring therapists left the study. The randomization of therapists precluded the inclusion of new therapists into the study; therefore, data collection was terminated prematurely. Data collection was discontinued either when a clinic recorded data on 25 wrists or 1 of the testers left the study. The 32 testers had an average of 10 years' experience (range = 1.5-20 years). The testers were 25 occupational therapists, 6 physical therapists, and 1 combined occupational/physical therapist. Of the 32 testing therapists working in hand clinics, 17 were certified hand therapists (Tab. 2).

                                 Experience (y)
Clinic(a)   Profession(b)   (%)   X      SD     Range

A           PT               50   14.5   1.5    13-16
            OT               50   15.5   0      15.5
            CHT             100   15     1.2    13-16
B           PT               50    3     0.5    2.5-3.5
            OT               50    9     2.5    6.5-11
            CHT              50    9     2.5    6.5-11
C           PT                0
            OT              100   11.5   5.1    4-18
            CHT              50   16     2     14-18
D           PT                0
            OT              100    4.5   0.8    3.5-5.5
            CHT               0
E           PT               50   18     3     15-21
            OT               50    7     5.8    1.5-13
            CHT              50   18     3     15-21
F           PT                0
            OT              100    8     5.1    2.5-16
            CHT              25   16     0     16
G           PT               25   12     0     12
            OT               75    9     5.4    2-16
            CHT              50   14     2     12-16
H           PT                0
            OT              100    11    4.5    7-18
            CHT             100    11    4.5    7-18

(a) See Appendix for listing of clinics.
(b) Pt = physical therapy, OT = occupational therapy, CHT = certified hand thera
pist.

Instrumentation

Each participation clinic was issued two plastic, 15.2-cm (6-in) goniometers to use for all measurements.(*) Each goniometer's accuracy was asssessed by measuring 10 randomly chosen, computer-generated angles between 0 and 180 degrees. All goniometric measurement angles were in agreement with the computer-generated angles.

One side of the goniometers' numerical scale was covered with moleskin to blind the measurer from reading the scale. This precaution prevented the tester from viewing the measurement results, but allowed a recorded to view the reverse side of the goniometer and record the results (Fig. 1).

Procedure

This study used a modified version of a goniometric measurement method originally described by Rothstein et al.[23] At each clinic, measurements of subjects were performed by randomly paired sets of testers.

A patient was identified as a potential subject by one of the testing therapists. The therapist then obtained consent from the patient, collected subject data, and decided whether goniometric measurements were going to be performed prior to or after that day's therapeutic session. That therapist was also the first testing therapist. Prior to taking any goniometric measurements, the first testing therapist randomly chose the second tester, thus establishing the measuring pair. This entire procedure was repeated for every subject at every clinic.

The first tester then measured the subject's passive wrist extension and flexion in the following order: (1) radial goniometric technique, (2) ulnar goniometric technique, and (3) volar/dorsal goniometric technique (Figs. 2-5). Operational definitions were provided so that each of the testers could use them as guidelines (Tab. 3). A recorder, trained in reading the goniometer, at each clinic read and documented each goniometric measurement while ensuring that the tester did not see the result. After measuring the subjects in the defined order, the first tester remeasured the same subject in the same order after a 30- to 60-second interval. The second tester of the measuring pair then repeated all of the measurements twice while the recorder documented the results. The elapsed time between the first tester's and the second tester's measurements was 2 to 3 minutes.

Each subject's passive wrist flexion and extension was therefore measured three different ways a total of four times (two times per tester) by two different testers. No therapeutic activity was performed between any of the measurements. None of the testers had access to the measurement results, nor were they permitted to watch any of the other testers measure subjects. All eight clinics were instructed to collect data on 25 different subjects; however, the final number of subjects per clinic varied (Tab. 1). On completion of data collection, all results were returned to us for analysis.

Data Analysis

In an attempt to determine whether the radial, ulnar, and volar/dorsal goniometric techniques produced similar or dissimilar results on the same subject, an analysis of variance (ANOVA) was used. The ANOVA allowed a comparison of the mean goniometric results of the three techniques by clinic. If significance was noted (P = .05), a Tukey Honestly Significant Difference Multiple-Comparison Test was performed to determine which techniques were different from one another. This test is one of the more conservative multiple-comparison designs and has stringent statistical criteria that must be met before deeming variance as significantly different.[26] The results of this test may help the clinician decide whether these three different techniques could be used interchangeably.

The intraclass correlation coefficient (ICC),[27] which is based on an ANOVA, was used to provide an estimate of agreement both within raters (model 3,1) and between raters (model 2,1). In the 3,1 model, the tested raters are the only raters of interest, thus making it an appropriate model for asssessment of within-tester reliability.[27] Shrout and Fleiss[27] support this view, as they do not feel it is reasonable to generalize one rater's score to a larger population using the 3,1 model. The 2,1 model also uses an ANOVA; however, it partitions the variance into differences between subjects, errors, and raters. This partitioning permits generalization of results beyond the raters in this study. The use of model 2,1 is suggested when assessing between-rater reliability.[28] For the determination of intertester reliability, the mean of each therapist's two trials was used. The ICC reliability index has been commonly utilized as the primary determinant of reliability in previous wrist goniometry studies.[24,25]

The amount of measurement error, or the standard error of measurement (SEM), was also used to quantify reliability.[29] The SEM may be the most desired index of reliability, as it provides a number that represents the way a single score will vary if a test is administered more than once. The SEM's clinical revelance is enhanced by the fact that it is expressed in the metric unit of the measurements.[29]

A generalizability statistical model,[30] an extension of the intraclass reliability model, was used to identify the percent contribution of numerous components to error variation. Sources of error analyzed included goniometric technique, therapist, diagnostic category, treatment, patient, and miscellaneous error. All of these statistical calculations were organized and processed on the Statistical Analysis System computer program.[31] Descriptive statistics, measures of central tendency (mean, standard deviation, and range), were used to represent all of the clinics with a statistical value.[32]

Results

There was a wide range of passive wrist flexion and extension among the subjects. Overall, the subjects' goniometric measurements varied from 5 to 95 degrees in both flexion and extension.

The ANOVA and the Tukey Honestly Significant Difference Multiple-Comparison Test revealed significant differences (P = .05) among the mean goniometric results of the specified techniques for flexion in six of the eight clinics and for extension in three of the eight clinics (Tab. 4).

[TABULAR DATA OMITTED]

The ICC results for intratherapist and inthertherapist flexion and extension for each clinic are displayed in Figures 6 through 9. The descriptive statistics pertaining to the ICC for all clinics combined are presented in Table 5.

[TABULAR DATA OMITTED]

For the ICC results exceeding .90, the flexion measurements of all techniques were always better than the extension measurements. The dorsal goniometric technique exceeded the .90 ICC mark 75% and 100% of the time for intratherapist flexion and intertherapist flexion, respectively. For both intratherapist and intertherapist extension, the volar technique exceeded the .90 ICC mark 38% of the time. The ulnar goniometric technique exceeded the .90 ICC mark 38% and 50% for intratherapist and intertherapist flexion, respectively, whereas it never exceeded .90 for either intratherapist or intertherapist extension. For measurements taken on the radial side, the .90 ICC mark was exceeded 25% of the time for intratherapist flexion and 50% of the time for intertherapist flexion. Both intratherapist and intertherapist extension radial measurements never exceeded the .90 ICC mark.

The SEM results for intratherapist and intertherapist flexion and extension for each goniometric technique are presented in Table 6. For intratherapist flexion and extension, the SEM of the volar/dorsal technique was lowest 100% and 75% of the time, respectively, as compared with the SEMs of the other techniques. For both intertherapist flexion and extension, the SEM for the volar/dorsal technique was the lowest 80% of the time.

[TABULAR DATA OMITTED]

The greatest effect on variance among all measurements, as analyzed via the generalizability model, was accounted for by the patient 65% (range = 38%-80%) of the time and then by inherent error within the study (ie, study design, recorder error, and so forth) 16% (range = 12%-25%) of the time (Fig. 10).

The diagnostic category had an effect on reliability an average of 9% (range = 0%-49%) of the time (Fig. 11). The most frequently encountered diagnosis was postsurgical 72% of the time. This postsurgical category included subjects who underwent procedures such as carpal tunnel release, open reduction and internal fixation of the carpus or radius/ulna, limited carpal fusions, and tendon transfers. The postfracture diagnostic group was encountered 16% of the time and included radius/ulna fractures, carpal fractures, and metacarpal fractures that were not openly reduced or internally fixated. General orthopedics was seen 8% of the time and consisted of distal radius/ulna, carpal, and soft tissue strains, sprains, and tears. Burns were also included in this category. The neurological category was encountered least often at 4% of the time and included peripheral nerve entrapments, central nervous system disorders, and iatrogenic neuropraxias (Tab. 1).

The effect of whether treatment was administered prior to or after the goniometric measurement on realiability was averaged at 6% (range = 0%-30%) (Fig. 11). An average of 35% of the subjects were treated before being measured, and 65% were measured after treatment (Tab. 1). The actual goniometric technique effect on reliability was 3% (range = 1%-5.5%), and the therapist effect was 2% (range=0%-6.5%) (Fig. 12).

Discussion

Overall, many factors can effect wrist goniometric measurements. With this in mind, the results of the three goniometric techniques were significantly different on many occasions. In terms of reliability, the results of the ICC and the SEM show the volar/dorsal technique to be the most reliable of the three different techniques.

Despite the fact that all three goniometric techniques are utilized for measuring wrist PROM, they produced significantly different results when compared in this study. From the results of the ANOVA and the subsequent Tukey Multiple-Comparison Test, it is apparent that in many instances the results of the three techniques were significantly different. This finding has clinical implications, as one therapist may be making clinical decisions based on one technique, whereas another therapist may be making treatment decisions based on a different goniometric technique. Although these results address one of the purposes of the study, they do not address the question of which is the most reliable technique. For this purpose, the ICC was used as a statistical index of reliability. When inherent variability is high, as it was in this study with wrist PROM measurements ranging from 5 to 95 degrees, one can have great confidence in the ICC as strong index of [reliability.sup.28] Additionally, using the SEM as an index of reliability allows the reader to determine whether the amount of measurement error is clinically significant.

Reliability denotes the stability of a measure and whether one tester, or two testers, can obtain similar measurements of the same variable on separate occasions. Most researchers and clinicians establish their own definitions of what is "acceptable reliability." One must consider the nature of the measurement and whether measurement stability is necessary for a clinical decision to be made. For instance, clinicians might tolerate lower reliabilities when the measuring technique is only one of many different ways to measure an attribute. Higher reliabilities may be required, however, when one measurement alone dictates how a clinician will approach the treatment plan. High ROM measurement reliability at the wrist is particularly useful, as changes in ROM may determine the success or failure of a treatment. Further standards for tests and measurements related to reliability have been previously established[33]

Of the three goniometric techniques, the volar/dorsal technique was the most reliable, with consistently higher ICC and lower SEM results than the radial and ulnar techniques both within and between testers. This finding differs from those of previous reliability studies. In those studies, the volar/dorsal technique was not assessed and consequently could not be compared with the radial and ulnar techniques. We felt a critical analysis of all three techniques was warranted, especially as the volar/dorsal technique is recommended by the American Medical Association[7] and the American Society for Hand Therapists.[8]

Additionally, it appears that anyone who is measuring wrist ROM should know which goniometric technique has the greatest reliabilty, because the result can have profound implications for the patient. The goniometric result may determine what type of further treatment is warranted. It may deem a surgical result as successful or unsuccessful.

Additionally, it may play a major role in a patient's partial or permanent impairment rating. When a wrist PROM measurement will be used as a primary determinant in making a clinical decision, or will be an independent variable in a research study, the volar/dorsal approach may be the goniometric technique of choice. The volar/dorsal technique consistently was above the .90 ICC level as compared with the other techniques. In addition, the volar/dorsal technique consistently had the highest ICC value and the lowest SEM both within and between testers. If, however, a gross anatomical volar or dorsal abnormality prevents accurate goniometric alignment, the radial or ulnar technique would suffice. There should be no alternating of measurement techniques on the same patient.

Unlike other reliability studies, this study revealed a trend toward intertester ICC and SEM results being slightly better than intratester results. Clinically, the differences noted may be insignificant; however, it does question traditional wisdom that repeated measurements made by different therapists be interpreted more conservatively than measurements made by the same therapist.

The percent effect of various factors on the reliability of wrist goniometric measurements was analyzed by the generalizability theory. Classical reliability theory partitions a score or measurement into a true component and an error component, with any deviation from the true score being deemed external random error. The generalizability theory forces one to interpret reliability in a multidimensional fashion. The underlying premise is that not all variations from trial to trial should be attributed to random error. Instead, the generalizability theory is used to identify other factors that might influence test scores. With the identification of other relevent effects on test scores, the therapist should be able to explain, predict, and control for these factors and thus level less variance unexplained as simply "error."[28] The generalizability statistical analysis allowed quantification of which factors contributed to variance and ultimately to reliability.

In our study, the patients' wrist motion provided the greatest amount of variance. This finding was not unexpected , as some subjects had near-normal PROM, whereas others had severely limited motion at the wrist. Unexplained error had the second greatest effect on reliability and may have included testing conditions, such as time of day, or positioning of the subject, among other factors. Deficiencies in the study design can also be categorized under "unexplained error." The effect of diagnostic category and treatment can certainly influence variance among measurements. Scarring, bony o r soft tissue deformities, or pain may cause imprecise goniometric placement. Inconsistent external force application, which is required for PROM measurements, may also affect the results. The effect of the actual goniometric technique and the effect of the therapist on variance is small but not inconsequential, and should be considered when measuring wrist ROM.

Other potential sources of error during goniometric measurements may have been poor visualization of bony landmarks. Perhaps when edema or an aberrant bony prominence was encoutered, the testers attempted to be more precise in placement of the goniometer's arms, hence the greater reliability of the volar/dorsal technique. These anatomical obstructions, however, may cause the therapist to question what is actually measuring, or the validity of the measurement. Horger[24] noted external force application most often as a potential source of error in wrist flexion and extension PROM measurements. In our study, however, the potential for more consistent external force application for a PROM measurement during the volar/dorsal method may have been one of the reasons it had greater reliability. In the volar/dorsal technique, the distal arm of the goniometer is placed where a passive external force is applied to the wrist, thus allowing greater repeatability of the passive force application. Other potential sources of error may include inconsistent positioning of the patient for the measurement or pain while moving the wrist passively. All potential sources of error may have been more profound in measuring extension rather than flexion, as flexion measurement reliability was greater. All of these potential sources of error have previously been reported in the literature.[21,24,34,35]

In this study, an effort was made to sample a broad range of patients and therapists in an attempt to allow generalization of results beyond this study. The design of the study ensured the assessment of all three goniometric techniques. In Horger's study,[24] the therapists were allowed to choose whichever technique they preferred. In Horger's study, all specialized therapists (those who practice at hand clinics) chose the ulnar technique, whereas all nonspecialized therapists (those who do not practice at hand clinics) chose the radial technique. The results of Horger's study, therefore, may be an assessment of specialty versus nonspecialty therapists rather than an assssement of goniometric techniques. The study by Duffin and Zoeller[25] used nondisabled subjects and was limited to only one rater, who measured wrist flexion and extension radially and ulnarly. In our study, a broad range of clinical experience among the therapists, including certified hand therapists who practice in a hand/upper-extremity environment, ensured heterogeneity. Multiple clinics across the United States were used to allow for regional differences. Finally, the randomization "blinding" of the tester minimized bias within the study.

Although this was a multicenter study, we believe separate analysis of each clinic was required because randomization of testing pairs occurred within clinics but not between clinics. Testing pairs had an opportunity to measure subjects only within their own clinic. Therefore, the ANOVA, Tukey multiple-Comparison Tests, ICC, SEM, and generalizability statistics were utilized for each individual clinic. The descriptive statistics (mean, standard deviation, and range) were calculations of general trends across all clinics.

This study assessed only PROM for wrist flexion and extension. Active-range-of-motion goniometric reliability should be assessed in future studies. Although the literature suggests that AROM measurements are more reliable than PROM measurements,[24,36] passive motion measurements are thought to be more valid indicators of changes in periarticular connective tissue.[6] Additionally, a PROM measurement appears to be the most appropriate measurement for assessing widely used passive treatments.[6] The design of this study prohibited the therapists from choosing a goniometric technique to use. By requiring each measurer to use all of the techniques, the study design ensured assessment of all three goniometric methods. In addition, the study design allows generalization of the results beyond the testers in this study. The operational definitions were compiled from textbooks that are used to teach goniometry in physical therapy and occupational therapy curricula. The use of a recorder to read the goniometer did not reproduce the true clinical environment, but minimized bias. Finally, the use of multiple centers resulted in an unequal number of subjects from each clinic and the inability of each measuring pair to have an equal opportunity to measure all subjects. This limitation restricted the use of the ICC statistical analysis to each clinic independently rather than to all clinics as a whole.

Future studies should assess wrist AROM measurement reliability and validity. Additionally, studies that try to further analyze some of the identified sources of wrist measurement error may help provide more specific guidelines for measurement. For example, different measuring techniques for specific diagnostic categories may be useful. Controlling external force application for PROM at the wrist may improve reliability further.

Conclusions

The radial, ulnar, and volar/dorsal goniometric techniques should not be used interchageably, as their results frequently will be inconsistent. Therefore, when given the choice, the volar/dorsal goniometric technique should be used, as it appeared to be the most reliable method both within and between testers for measurements of passive wrist flexion and extension. The clinician should be cautious, however, when interpreting these data, as goniometric reliability for wrist passive flexion and extension measurements were population-specific.

Appendix Clinics Participating in

Multicenter Study

A - Curtis Hand Center, Union Memorial Hospital, Baltimore, MD B - University of Florida, Department of Orthopaedics, Gainesville, FL C - Hand Rehabilitation Center of Gainesville Inc, Gainesville, FL D - Loyola University Medical Center, Department of Occupational Therapy, Maywood, IL E - Hand Rehabilitation Association of San Antonio Inc, San Antonio, TX F - Hand Surgery Associates PC, Denver, CO G - University of Pennsylvania, Medical Center Penn Hand Specialists, Philadelphia, PA H - Michigan Hand Rehabilitation Center Inc, Warren, MI

References

[1] McEntee PM. Therapist's management of the stiff hand. In: Hunter JM, Schneider LH, Mackin EJ, Callahan AD, eds. Rehabilitation of the Hand: Surgery and Therapy. 3rd ed. St. Louis, Mo: CV Mosby Co; 1990:328-341. [2] Mennell JM. Joint Pain: Diagnosis and Treatment Using Manipulative Techniques. Boston, Mass: Little, Brown and Company Inc; 1964:44-67. [3] Coditz JC. Dynamic splinting of the stiff hand. In: Hunter JM, Schneider LH, Mackin EJ, Callahan AD, eds. Rehabilitation of the Hand: Surgery and Therapy, 3rd ed. St. Louis, Mo: CV Mosby Co; 1990-342-352. [4] Schultz-Johnson K. Splinting: a problem-solving approach. In: Stanley B, Tribuzi S, eds. Concepts in Hand Rehabilitation. Philadelphia, Pa: FA Davis Co; 1992:238-271. [5] Bell-Krotoski JA. Plaster cylinder casting for contractures of the interphalangeal joints. In: Hunter JM, Schneider LH, Mackin EJ, Callahan AD, eds. Rehabilitation of the Hand: Surgery and Therapy. 3rd ed. St. Louis, Mo: CV Mosby Co; 1990:1128-1133. [6] Flowers KR, Michlovitz SL, Assessment and management of loss of motion in orthopaedic dysfunction. In: Postgraduate Advances in Physical Theraphy. Alexandria, Va: American Physical Therapy Association; 1988:1-11. [7] Guides to the Evaluation of Permanent Impairment. 2nd ed. Chicago, Ill: American Medical Association; 1984. [8] Fess EE, Moran CA. Clinical Assessment Recommendations. Garner, NC: American Society of Hand Therapists; 1981. [9] Esch D, Lepley M. Evaluation of Joint Motion: Methods of Measurement and Recording. Minneapolis, Minn: University of Minnesota Press; 1973. [10] Moore ML. Clinical assessment of joint motion. In: Basmajian JV, ed. Therapeutic Exercise. 4th ed. Baltimore, Md: Williams & Wilkins; 1984::194-224. [11] Norkin CC, White DJ. Measurement of Joint Motion: A Guide to Goniometry. Philadelphia, Pa: FA Davis Co; 1985. [12] Scott N, Trombly CA. Evaluation. In: Trombly CA, ed. Occupational Therapy for Physical Dysfunction. 2nd ed. Baltimore, Md: Williams & Wilkins; 1983:126-229. [13] Watson HK, Hempton RF. Limited wrist arthrodesis. J Hand Surg [AM]. 1980;15:320-327. [14] Cooney WP, Beckenbaugh RD, Linscheid RL. Total wrist arthroplasty. Clin Orthop. 1983; 187:121-128. [15] Palmer AK, Dobyns JH, Linscheid RL. Management of post-traumatic instability of the wrist secondary to ligament rupture. J Hand Surg [Am]. 1978;3:507-532. [16] Cobe HM. The range of active motion at the wrist of white adults. J Bone Joint Surg. 1928;26:763-774. [17] Hewitt D. The range of active motion at the wrist of woman. J Bone Joint Surg. 1928; 26:775-787. [18] Hellebrandt FA, Duvall EN, Moore ML. The measurement of joint motion, part 3: reliability of goniometry. Phys Ther Rev. 1949;29:302-307. [19] Hamilton GF, Lachenbruch PA. Reliability of goniometers in assessing finger joint angle. Phys Ther. 1969;49:465-469. [20] Low JL. The reliability of joint measurement. Physiotherapy. 1976;62:227-229. [21] Fish DR, Wingate L. Sources of goniometric error at the elbow. Phys Ther. 1985;65:1666-1670. [22] Solgaard S, Carlsen A, Kramhoft M, Petersen VS. Reproducibility of goniometry of the wrist. Scand J Rehabil Med. 1986;18:5-7. [23] Rothstein JM, Miller PJ, Roetger RF. Goniometric reliability in a clinical setting: elbow and knee measurements. Phys Ther. 1983;63: 1611-1615. [24] Horger MM. The reliability of goniometric measurements of active and passive wrist motions. Am J Occup Ther. 1990;44:342-348. [25] Duffin L, Zoeller R. Reliability and comparison of two goniometric methods for the wrist. Presented at the 66th Annual Conference of the American Physical Therapy Association; June 23-27, 1991; Boston, MA. [26] Marks RG. Analyzing Research Data: The Basis of Biomedical Research Methodology. London, England: Lifetime Learning; 1982. [27] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428. [28] Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. East Norwalk, Conn: Appleton & Lange; 1993:505-528. [29] Rothstein JM. Measurement and clinical practice: theory and application. In: Rothstein JM, ed. Measurement in Physical Therapy. New York, NY: Churchill Livingstone Inc: 1985:1-46. [30] Cronbach LG. The Dependability of Behavioral Measurements: Theory of Generalization for Scores and Profiles. New York, NY: John Wiley & Sons Inc; 1972. [31] SAS/STAT 6.03 Cary, NC: SAS Institute Inc; 1988. [32] Morehouse CA, Stull GA. Statistical Principals and Procedures With Applications for Physical Education. Philadelphia, Pa: Lea & Febiger; 1975. [33] Task Force on Standards for Measurement in Physical Therapy. Standards for tests and measurements in physical therapy practice. Phys Ther. 1991;71:589-622. [34] Moore ML, The measurement of joint motion, part 1: introductory review of the literature. Phys Ther Rev. 1949;29:195-205. [35] Hamilton GF, Lachenbruch PA. Reliability of goniometers in assessing finger joint angle. Phys Ther. 1969;49:465-469. [36] Gajdosik RL, Bohannon RW. Clinical measurement of range of motion: review of goniometry emphasizing reliability and validity. Phys Ther. 1987;67:1867-1872.