Reliability of goniometric measurements and visual estimates of knee range of motion obtained in a clinical setting.

Physical therapists frequently assess passive range of motion (PROM) of the knee as part of their examination of patients with knee complaints. The universal goniometer is frequently used to measure PROM of the knee. Many times, however, a visual estimation of the PROM is made in lieu of using a goniometer. If PROM measurements are to be useful to the physical therapist, the reliability of the measurements must be established. Reliability is defined as the consistency of a measurement. [1] Because PROM measurements are taken on a patient several times, and often by different therapists, intratester and intertester reliability are essential if the measurements are to be meaningful. [2]

Few studies have examined the reliability for PROM measurements of the knee taken in a clinical setting. Rothstein and colleagues [3] examined the intratester and intertester reliability of PROM measuments taken on 12 patients' knees. Twelve randomly paired physical therapists measured passive knee flexion and extension on patients who required knee PROM measurements as part of their physical therapy examination. Therapists were allowed to used their own techniques while taking measurements. The authors reported good intratester reliability for measurements of flexion and extension and good intertester reliability for flexion measurements. Intertester reliability for knee extension measurements was reported as being poor.

A post hoc analysis revealed intertester reliability for knee extension measurements improved when paired therapists used the same patient position during the measurements. This finding suggest patient position needs to be controlled during knee extension measurements to minimize error. However, as the authors indicated, because the study was done in one clinical setting on a small sample, the results may not be generalizable to other clinics.

Based on our experience, therapists frequently visually estimate the PROM of their patients' knees. Therapists also may take goniometric measurements and visual estimates of a patient's knee PROM during the course of treatment. Goniometry and visual estimation of a patient's range of motion (ROM) are parallel forms of the same test. Parallel-forms reliability is determined by comparing the measurements taken with two methods and describes the extent to which measurements obtained with two methods are interchangeable. [4] The parallel-forms reliability for goniometric measurements and visual estimates of the PROM of the knee has not been examined.

Opinions regarding the usefulness of visual estimates of PROM vary. The American Academy of Orthopaedic Surgeons (AAOS) [5] and Rowe [6] have suggested that visual estimation is more accurate than using a goniometer to measure ROM when bony landmarks are not easily seen or palpated. By contrast, Moore, [7] Minor and Minor, [8] and Salter [9] have started that goniometric measurements are more reliable than visual estimates. None of these authors provided data to support their arguments.

There has been only one study that has examined the reliability of visual estimates of PROM of the knee. Marks and associates [10] investigated the reliability for visual estimates of knee ROM taken by three physicians on rheumatoid arthritic patients. Marks et al reported good intratester and intertester reliability. Although this study was done in a clinical setting, the measurements were obtained by physicians, which may not reflect the reliability of measurements taken by physical therapists.

The reliability for visual estimates of knee PROM has not been examined in a physical therapy setting. In addition, the parallel-forms reliability of goniometric measurements and visual estimates of the PROM of the knee has not been examined. Finally, even though Rothstein et al [3] reported the reliability for goniometric PROM measurements of the knee taken on patients, the sample size was small. A study that examines the reliability of goniometric measurements and visual estimates of knee PROM obtained on a large sample of patients would further elucidate the usefulness of these measurements.

Our study was divided into two parts. Part 1 was designed to replicate part of the study of Rothstein et al, [3] but to use a different statistical test. Rothstein et al [3] used a less conservative form of the intraclass correlation coefficient (ICC), which, we believe, underestimates error. [11] In part 1, we examined the intratester and intertester reliability for goniometric PROM measurements of the knee. The purpose of part 2 of our study was to examine the parallel-forms intratester reliability for goniometric measurements and visual estimates of knee flexion and extension. In addition, the parallel-forms intertester reliability of goniometric measurements and visual estimates of the knee was examined. The intertester reliability for visual estimates of knee flexion and extension was also examined. Because we believed that therapists would likely be biased by their first measurements, we chose not to examine the intratester reliability for visual estimates of the knee.

Method

Subjects

Subjects for this study were 43 patients referred to the Physical Therapy Department, Medical College of Virginia Hospital (MCVH), Virginia Commonwealth University, Richmond, Va. Criteria for admission to this study were that each patient was at least 18 years of age and that the patient's examination would normally include PROM measurements of the knee. Both knees of 7 of the 43 patients were measured; therefore, a total of 50 sets of measurements were obtained.

The subjects consisted of 29 males and 14 females. The ages of the subjects ranged from 18 to 80 years ([unkeyable] = 39.5, SD = 15.0). Age, sex, height, weight, diagnosis, and extremity tested were recorded for each subject (Tab. 1). These data were collected for a posteriori analyses to determine whether any of these factors may have influenced reliability. All subjects signed a consent form prior to participation in the study.

Testers

All testers were full-time staff physical therapists who treated adult patients at MCVH. The 14 therapists had a mean of 7.2 years of experience (SD = 4.0) and had graduated from 12 different physical therapy schools. Additional information collected was the age, sex, area of specialty, and the number of times per week each therapist visually estimated the PROM of the knee and recorded the value of the chart (Tab. 2). These factors were collected for a posteriori analyses to determine whether any of these variables may have influenced reliability.

[TABULAR DATA OMITTED]

Instrumentation

Three plastic goniometers, (*) each with a 12.7-cm (5-in) moveable arm and a scale marked in 1-degree increments, were used to take measurements. We chose this size of goniometer because we believe these types of goniometers are frequently used on patients with knee problems. The accuracy of each goniometer was assessed prior to the beginning of the study by measuring 10 randomly selected angles drawn by use of a proctractor. The three goniometers measured all angles accurately. The goniometer scales were covered with contact paper on the side facing the testers so they could not read the measurements. The other side was left uncovered, which allowed the recorder to read the measurements.

Procedure

The procedure chosen for this study was a modification of the procedures described by Rothstein et al. [3] Each therapists was given a different random list of names of the other therapists. When a therapists identified a patient as being appropriate for the study, that therapist would take the first set of measurements. The second tester was then determined from the first tester's random list. This method provided a random pairing of therapists for each patient admitted to the study.

A recorder (MAW) was responsible for reading and recording all measurements and for recording the patient position used for each measurement. Patient position was recorded for a posteriori analyses to determine whether positioning influenced reliability.

When an appropriate patient was identified, the therapist who identified the patient (the referring therapist) notified the recorder. The recorder then identified the second tester by proceeding down the random list of the referring therapist to the name of the next available therapist. The referring therapist first visually estimated the PROM of the knee flexion and extension, in that order. Visual estimates were done first because we felt that if the goniometer was used first, the angle made by the goniometer arms might influence by the visual estimate. The therapist recorded the measurements on a piece of paper provided by the recorder and then handed the paper to the recorder. The referring therapist then used the blinded goniometer to measure passive knee flexion twice and knee extension twice, in that order. When the arms of the goniometer had been aligned to the therapist's satisfaction, the goniometer was handed to the recorder. The recorder read the value from the goniometer and recorded the value. After recording each measurement, the recorder positioned the goniometer arms back to the zero-degree position. After each measurement, the subject's limb was repositioned in its starting position.

After the referring therapist obtained the six measurements, the second tester (retest therapist) took the six measurements in the same order as

[TABULAR DATA OMITTED]

the referring therapist. To minimize bias, the retest therapist did not observe the referring therapist taking measurements. Throughout the study, all therapist were allowed to use their own methods for positioning the patient and the goniometer.

Data Analysis

The ICC (1,1), as described by Shrout and Fleiss, [12] was used to describe the degree of reliability of the measurements. We chose this form of the ICC because we believe it best reflects the error that can be expected when a therapist takes a PROM measurement on a patient. [11]

Part 1. The ICCs for intratester reliability for goniometric measurements of knee flexion and extension were calculated by comparing the first and second goniometric measurements taken by each tester. The referring therapists and the retest therapists each obtained 50 paired measurements for each motion; therefore, a total of 100 paired measurements were obtained for each motion. The ICCs for intertester reliability for goniometric measurements of knee flexion and extension were calculated by comparing the first goniometric measurements for each pair of testers.

Part 2. The ICCs for parallel-forms intratester reliability for measurements obtained by use of a goniometer and by visual estimation were calculated by comparing the visual estimate and the first goniometric measurement obtained by each tester. The ICCs for parallel-forms intertester reliability for measurements obtained by use of a goniometer and by visual estimation were calculated in the following way. The visual estimates obtained by the referring therapists were compared with the first goniometric measurements obtained by the retest therapists. In addition, the first gonioetric measurements obtained by the referring therapists were compared with the visual estimates obtained by the retest therapists. The referring therapists and the retest therapists each obtained 50 paired measurements for each motion; therefore, a total of 100 paired measurements were obtained for each motion.

The ICCs for intertester reliability for visual estimates of knee flexion and extension were calculated by comparing the visual estimates for each pair of testers. Therefore, there were 50 pairs of measurements for each motion.

Results

The means, standard deviations, and ranges for all goniometric measurements and visual estimates obtained in this study are summarized in Table 3.

Part 1

The ICCs for intratester reliability of measurements obtained with a goniometer were .99 for knee flexion and .98 for knee extension. The ICC values for intertester reliability of measurements obtained with a goniometer were .90 for knee flexion and .86 for knee extension (Tab. 4).

[TABLE DATA OMITTED]

Part 2

The ICC values for the parallel-forms intratester reliability of goniometric measurements and visual estimates were .93 for flexion and .94 for extension. The ICC values for the parallel-forms intertester reliability of goniometric measurements and visual estimates were .86 for flexion and .82 for extension. The ICC values for intertester reliability of visual estimates were .83 for knee flexion and .82 for knee extension (Tab. 4).

Discussion

Part 1

Goniometric PROM measurements of knee flexion and extension were highly reliable when the same therapist took repeated measurements. This finding of high intratester reliability for goniometric measurements of knee flexion and extension agrees with the results of Rothstein et al. [3]

The intertester reliability for goniometric measurements of knee flexion and extension was also high, but not as high as intratester reliability. To minimize error, PROM measurements of the knee should be taken on a patient by the same therapist. Although the additional error associated with measurements taken by different therapists is small, this small increase in error may affect the usefulness of these measurements.

Rothstein et al reported "relatively poor intertester reliability" [3(p1613)] for goniometric PROM measurements of knee extension, with ICCs ranging from .59 to .80. Our study demonstrated higher intertester reliability for knee extension measurements (ICC = .86). The larger sample size could explain the higher intertester reliability estimate for knee extension measurements. It has also been suggested that the reliability for measurements obtained at one facility may not be generalized to all facilities. [3,13] Both studies, however, were in agreement that the reliability of knee extension measurements was lower than for knee flexion measurements.

Rothstein et al[3] used formula (E4) of the ICC described by Bartko and Carpenter, [14] which is less conservative than the ICC (1,1) used in our study. [15] Therefore, relative to our study, the results obtained by Rothstein et al provided an overestimation of the degree of reliability. The results of our study, however, demonstrated similar or higher reliability estimates compared with those reported by Rothstein et al.

Patient positioning has been considered a factor that may influence reliability

Table 4. Intratester and Intertester Reliability for Knee Flexion and Extension Measurements

                       Intratester     intertester
Method                 N (a)  ICC (b)  N     ICC
Flexion
  Goniometer           24  .99        50     .90
  Visual estimation    ... ...        50     .83
  Between method (c)   100 .93       100     .86
Extension
  Goniometer           100 .98        50     .86
  Visual estimation    ... ...        50     .82
  Between methods      100 .94       100     .82
  (a) N = number of paired measurements.
  (b) ICC = intraclass correlation coefficient (1,1).
  (c) Visual estimate was compared with first goniometric
measurement for each tester.

for some measurements. [3] Rothstein et al[3] demonstrated that the intertester reliability for knee extension PROM measurements noticeably decreased when different patient positions were used by both therapists. An a posterior analysis of our study demonstrated that the interter reliability for goniometric measurements of knee extension decreased only slightly when patient position was different for paired therapists (Tab. 5). The error associated with knee extension measurements was also slightly greater when paired therapists used different patient positions during measurements (Tab. 5). This finding is different from that of Rothstein et al and suggests that patient position during knee extension PROM measurements contributes only slightly to the error associated with these measurements. Although the error associated with using different patient positions during knee PROM measurements is only slightly greater than the error present when the same position is used, we recommend standardizing patient positioni to minimize error.

Part 2

The parallel-forms intratester reliability for PROM measurements obtained by use of a goniometer and by visual estimation was high for knee flexion and extension measurements. The reliability, however, was not as high as the intratester reliability for goniometric measurements of knee flexion

Table 5. Intertester Reliability When Paired Testers Used the Same and Different Patient Positions

                               ICC (a)
Joint                                       Visual
Motion      N (b)  Position    Goniometer   Estimation
Flexion     35     Same        .92          .85
            15     Different   .85          .80
Extension   38     Same        .87          .83
            12     Different   .84          .82
  (a) ICC = intraclass correlation coefficient (1,1).
  (b) N = number of paired therapists.

and extension. Interchanging goniometric measurements and visual estimates of knee flexion or extension, therefore, may introduce a small amount of additional error. A therapist's ability to detect changes on the order of a few degrees, on average, may be a few degrees, on average, estimating a patient's PROM. The additional measurement error could result in a therapist making an incorrect decision when assessing for a small change in a patient's PROM.

The parallel-forms intertster reliability for PROM measurements obtained by use of a goniometer and by visual estimation was fair for knee flexion and extension measurements. The intertester reliability for goniometric measurements was slightly higher than the parallel-forms intertester reliability for these measurements. When different therapists must measure a patient's knee PROM, error can be minimized if both therapists use a goniometer.

The intertester reliability for measurements of knee flexion and extension obtained by visual estimation was fair, but better than we expected. Based on our clinical experience, we would have expected lower reliability. This finding, that different therapists can fairly reliably estimate knee ROM, is in general agreement with the findings of Marks and colleagues. [10] However, the intertester reliability for goniometric measurements was higher than for visual estimates of knee flexion and extension. Therapists will minimize error in their measurements if they take measurements on a patient with a goniometer. This finding supports claims by Moore, [7] Minor and Minor [8], and Salter [9] that measurements obtained with a goniometerare more reliable than visual estimates of PROM.

An a posteriori analysis similar to the study of Riddle et al [13] was performed to determine the effect of different diagnoses on reliability (Tab. 6). Patient diagnosis did not appear to affect intratester reliability, except for parallel-forms reliability of measurements taken on below-knee amputees. Intertester

[TABULAR DATA OMITTED]

reliability was generally poor (ICCs = .03-76) for measurements taken on below-knee amputees. Because the involved knees of patients with below-knee amputations have shorter distal limb segments, therapists may have difficulty aligning a goniometer or estimating the knee's position during PROM measurements. The small sample of amputees does not allow us to make conclusions about the reliability of PROM measurements taken on these patients. Our study does suggest that the reliability for PROM measurements of the knee, taken on below-knee amputees, should be investigated further.

This study examined the reliability of PROM measurements of the knee obtained by use of a goniometer and by visual estimation. The reliability for goniometric measurements of the shoulder, elbow, ankle, and foot have also been examined on appropriate patients. [3,13,16] However, the reliability for visual estimates of the PROM present at these joints has yet to be examined. The reliability fo goniometric measurements and visual estimates also needs to be examined for the wrist, hip, and hand. In addition, an examination of the reliability for active ROM measurements of joints measured clinically is needed

Conclusions

Goniometric PROM measurements of knee flexioni and extension are highly reliable when taken by the same physical therapist. Goniometric measurements of the PROM of a patient's knee taken by different therapists will not be as reliable as when the same therapist takes the measurements. Visual estimates of knee PROM will add slightly more error to the therapist's measurements than those taken with a goniometer. The additional error associated with visual estimates could affect the usefulness of the measurements if a therapist is attempting to detect small changes in a patient's PROM. When different therapists must measure a patient's PROM, therapists can minimize error by using a goniometer and by standardizing patient position.

Acknowledgments

We would like to thank the entire staff of the Medical College of Virginia Hospital Physical Therapy Department, whose efforts made this study possible.

References

[1] Kerlinger FN. Foundations of Behavioral Research. 2nd ed. New York, NY: Holt, Rinehart & Winston Inc; 1973.

[2] Miller JP. Assessment of joint motion. In: Rothstein JM, ed. Measurement in Physical Therapy. New York, NY; Churchill Livinstone Inc; 1985:103-136.

[3] Rothstein JM, Miller PJ, Roettger RF. Goniometric reliability in a clinical setting: elbow and knee measurements. Phys Ther. 1983; 63:1611-1615.

[4] 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.

[5] Joint Motion: A Method of Measuring and Recording Chicago, Ill: American Academy of Orthopaedic Surgeon; 1965:8.

[6] Rowe CR Joint measurement in disability evaluation. Clin Prthop. 1964;32:43-53.

[7] Moore ML. The measurement of joint motion, part I: introductory review of the literature. Phys Ther Rev. 1949;29:195-205.

[8] Minor MA, Minor SD. Patient Evaluation Methods for the Health Professional. Reston, Va: Reston Publishing; 1985.

[9] Salter N. Methods of measurement of muscle and joint function. J Bone Joint Surg [Br]. 1955;37:474-491.

[10] Marks JS, Palmer MK, Burke MJ, et al. Observer variation in the examination of knee joints. Ann Rheum Dis. 1978;37:376-377.

[11] Lovell FW, Rothstein JM, Personius WJ. Reliability of clinical measurements of lumbar lordosis taken with a flexible rule. Phys Ther. 1989;69:96-105.

[12] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[13] Riddle DI, Rothstein JM, Lamb RL Goniometric reliability in a clinical setting shoulder measurements. Phys Ther. 1987;67:668-673.

[14] Bartko JJ, Carpenter WT. On the methods and theory of reliability. J Nerv Metn Dis. 1976;163:307-317.

[15] Lovell FW, Rothstein JM, Personius WJ. Reliability of clinical measurements of lumbar lordosis taken with a flexible rule. Phys Ther. 1989;69:905-105.

[16] Elveru RA, Rothstein JM, Lamb RL. Goniometric reliability in a clinical setting: subtalar and ankle joint measurements. Phys Ther. 1988;68:672-677.

(*) ConvaCare Inc, PO Box 19747, Raleigh, NC 27619.

M Watkins, MS, PT, is Director of Physical Therapy, Glens Falls Hospital, 100 Park St, Glens Falls, NY 12801 (USA). This study was completed in partial fulfillment of the requirements for Mr Watkin's Master of Science Degree in Physical Therapy, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA. Address all correspondence to Mr Watkins.

D Riddle, MS, PT, is Assistant Professor, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia, Virginia Commonwealth University, PO Box 224, MCV Station, Richmond, VA 23298.

R Lamb PhD, PT, is Associate Professor and Chairman, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia, Virginia Commonwealth University.

W Personius, PhD, PT, is Professor and Chairman, Department of Physical Therapy, Shenandoah College and Conservatory-Winchester Medical Center, Winchester, VA 22601.