Reliability of Measuring Trunk Motions in Centimeters MARGARET ROST, SANDRA STUCKEY, LEE ANNE SMALLEY, and GLENDA DORMAN A method of measuring trunk motion and two related motions using a tape measure and a stepstool was developed by physical therapists at our hospital. The purpose of this study was to assess the reliability of this method. Three s of six motions performed by subjects were each measured by three physical therapist raters on two separate days. The motions were forward, backward, right side, right rotation, right straight leg raising, and right prone knee. Reliability, standard deviation, and standard error were calculated for each motion. Only forward exhibited good single measurement reliability. Reliability coefficients for all motions were higher for the average of three successive measurements or for the measurement of a motion on successive days by the same rater. Measurements of rotation and straight leg raising, despite the improvement, continued to have low reliability. Analysis of variance was used to determine the significance of the differences between means for each motion across three raters, three s, and two days. By looking at the analysis of variance and reliability estimates together, the authors identified two types of constant error affecting the data. Key Words: Evaluation test, Range of motion, Test reliability. Many studies have been reported of attempts to accurately measure range of spinal mobility. Several types of problems have been encountered. Moll and associates developed a method of measuring spinal extension using a tape measure, plumb bob, and skin marks on the lateral trunk. Anderson and Sweetman described a method of measuring trunk flexion and extension using a flexi-rule and fixed landmarks on the back. Sturrock et al and Hart et al both used a spondylometer to measure spinal mobility. 3, Disadvantages of using reported methods in the physical therapy clinic should be noted. In each of the studies cited, selected trunk motions were studied rather than all motions. Specifically, none of these The authors were staff therapists at Orthopaedic Hospital, Los Angeles, CA, when this study was conducted. Ms. rost is currently working in private practice, 67 E Washington Blvd, Pasadena, CA 906 (USA). Ms. Stuckey is now Administrative Physical Therapist Research, Orthopaedic Hospital, 00 S lower St, Los Angeles, CA 9007. Ms. Smalley is the Pain Center Physical Therapist, Orthopaedic Hospital, Los Angeles, CA. Ms. Dorman is Staff Therapist, Orthopaedic Hospital, Los Angeles, CA. This article was submitted ebruary 0,98, and accepted ebruary, 98. investigators measured rotation, side, straight leg raising, or prone knee. Also, special equipment that is not commonly found in physical therapy settings was needed to perform the measurements. Anderson and Sweetman's study required that the subjects lie prone and assume a position of back extension by extending the elbows and rising up into the cobra position of Yoga. This method relied on upper extremity strength of the subject and did not evaluate the extent to which the compressive force of gravity contributes to pain. Macrae and Wright used a tape measure and skin distraction to test forward flexion, 5 and Moll and different associates in two studies reported on measuring trunk flexion, extension, and lateral flexion using a tape measure. 6, 7 But none of these studies investigated mobility for all trunk motions or the reliability estimates of the measurement techniques. Reynolds 8 compared three methods of measuring spinal mobility: using a spondylometer, using a goniometer, and using the skin distraction technique described by Moll and Wright. 6 He found that these methods had distinct limitations, such as the need for special equipment and subject discomfort in maintaining the testing positions required for accurate Volume 6 / Number 0, October 98 3
TABLE Motions and Measurement Techniques Motion Starting Position Landmarks Verbal Cue Measurement Data Recorded (cm) orward Standing on stool, heels together, knees straight, Tip of right 3rd finger to top of stepstool. "Bend forward can, keep your knees straight." inal distance Distance. If finger below stool, value was negative. Backward Standing on stool, heels together, Spinous process of C7 to line joining right and left PSIS. "Bend backward can." inal distance Excursion between the initial and final distances. Right side Standing on floor, heels together, knees straight, Tip of right 3rd finger to top of stool. "Bend to the side can, let your arm hang, keep your fingers straight." inal distance Distance. Right rotation Sitting on plinth, knees together, hips flexed to 90, arms folded across chest. Left posterior clavicular prominence to right greater trochanter. "Sit up straight, turn to the right can." Initial and final distances Excursion between the initial and final distances. Right straight leg raising Supine on plinth, legs in neutral. Lateral malleolus to the top of the plinth. "Relax, let me raise your leg." inal distance Tape was perpendicular to top of table. Distance. Right prone knee Prone on plinth, legs in neutral, Lateral malleolus to top of plinth. "Relax, let me raise your leg." inal distance recorded after 90. Distance. goniometric measurements, and that Moll's technique was generally inaccurate. None of the techniques that have been described in the literature seemed to be appropriate for use in a physical therapy clinic where important concerns are for quickly and accurately measuring all trunk motions and other related motions and for keeping patient discomfort at a minimum. The physical therapists at Orthopaedic Hospital developed a simple technique for measuring trunk and related motions using a tape measure, an examination table, and a stepstool. The purpose of this study was to examine the reliability of the method when different therapists measured the same subject, when the same therapist measured one subject on different days, and when the same therapist measured one subject over three successive s. METHOD ig.. orward Bending. Standing, heels together, knees straight; measurement from third fingertip to top of stepstool. 3 our trunk motions and two related lower-extremity motions were measured on subjects by each of three physical therapist raters, ranging in professional experience from.5 to 6.5 years. The motions were forward, backward, right side bendphysical THERAPY
RESEARCH ig.. Backward Bending. Standing, heels together, knees straight; measurement from spinous process of C7 to level of posterior superior iliac spine. ing, right rotation, right straight leg raising, and right prone knee. (The designation "right" will not be used again; all of the motions were measured unilaterally.) The subjects were adult volunteers selected from hospital personnel. Initially, there were 7 subjects, men and 3 women; however, three subjects ( men and woman) were dropped from the study. One missed the second measurement session, and two exceeded the forward flexibility limits established (ie, they were able to reach the floor from the 0 -cm*-high stepstool. No volunteers were accepted who had seen a physician for low back pain within the past five years or had undergone back surgery. The mean age of the subjects was 33.8 years (range, 0-55 years); their mean weight was 68 kg (5 lbs) (range, 5-05 kg); and their mean height was 7 cm (range, 57-95 cm). The subjects were divided into groups and randomly assigned to the initial and successive raters so that each subject was measured by all raters on each day. Before his measurement session, each subject was instructed in preliminary stretching exercises consisting of five s of each of the tested motions. The raters followed written directions that specified the initial postures, the verbal instructions to be given to the subjects, and the landmarks used for the placement of the tape measure. Table includes these instructions. igures through 6 illustrate the measurement positions for each motion. The raters used identical equipment: an examination table, a *.5 cm = in. Volume 6 / Number 0, October 98 ig. 3. Right Side Bending. Standing, heels together, knees straight; measurement from the third fingertip to floor. ig.. Right Rotation. Sitting, arms crossed, hands on shoulders; measurement from left posterior clavicular prominence to right greater trochanter. metal tape measure, and a stepstool. Each motion was repeated and measured by each rater three times successively in the following sequence: forward, backward, side, rotation, straight leg raising, and prone knee. The measurement sessions were repeated one week later at the same time of day. 33
TABLE Mean Measurements (N ) for Each Motion,, Time, and (in cm) s s Motion ig. 5. Right Straight Leg Raising. Supine, passive elevation leg; measurement from lateral malleolus to table. Two types of data were obtained from an analysis of variance: information about the reliability or consistency of the measurements across raters, days, and s and information about significant mean differences and interactions between the means of the measurements. Information on reliability and significant mean differences were studied to determine the sources of error and whether corrections for that error needed to be done either by taking the average of the results or by changing the measurement techniques. or each motion, reliability coefficients, standard deviation values, and standard error values were calculated for each of the following variables using the formula outlined in Winer9: rater, day,, rater x day, and rater x day x. The reliability coefficient calculated for rater represents the reliability across raters when day and were held constant, that is, the reliability of measurements of a motion taken by different raters on the same day and on the same. The reliability coefficient for day represents the reliability across days when rater and were held constant. The coefficient calculated for indicated the reliability of orward 0.30 7.79 8.5 6.79 3 8.35 7.7 Backward 8.08 8.9 7.85 8.3 3 8.38 8.6 Right side. 5.9 5. 5. 3.90.5 Right rotation 5.35 5.58 8.9 8.65 3 7.3 7.60 Right straight leg raising 8.67 8.5 8.85 85.33 3 83.3 83.8 Right prone knee 37.77 37.77 3.85 3. 3 38.35 38.9 3 3 6.79 5.90 6.90 7.63 7.9 6.85 6.8 5.69 6.5 5..8 5.7 8. 8.33 8.9 7.95 7.9 8.5 8.0 8.3 8.98 8.8 8. 9.0 5.98.9.50.8.90 3.83 3.60 3.56. 3.98 3.69 3.5 5.58 8.65 7.75.9 8.67 8.3.5 8.8 8.3. 8.8 8.0 8.73 8.85 83.08 8.96 85.0 79.3 79. 79.60 83.7 7.85 75. 75.06 37.73 33.75 33.0 3.98 3.7 36.9 36.33 36.0 38.33 37.56 37.9 37.5 measurements across three successive s when day and rater were held constant. x day represents the reliability calculated for different raters taking measurements on different days with the held constant. inally, rater x day x is the reliability coefficient when rater, day, and all vary. In the clinical setting, determining the reliability across raters corresponds to the situation of different therapists measuring the patient's motion on the same day and on the same. Determining rater x day x reliability corresponds to the clinical situation of comparing the measurement by one therapist of a motion on one day with the measurement by another therapist's taken on another day. This comparison will be referred to as single measurement reliability. or the purpose of this study and in attempting to apply the accuracy of the measurements to the clinical situation, the authors chose to designate reliability coefficients greater than r =.80 as being good and those less than r =.80 as being poor.0 RESULTS The means found for each measurement are summarized in Table ; reliability, standard deviation, ig. 6. Right Prone Knee Bending. Prone, passive knee and standard error are summarized in Table 3; and the analysis of variance in Tables -9. flexion; measurement from lateral malleolus to table. 3 PHYSICAL THERAPY
RESEARCH TABLE 3 Reliability, Standard Deviation, and Standard Error for the Mean Measurements of all Motions Motions orward x day x day x Backward x day x day x Side x day x day x Rotation x day x day x Straight leg raising x day x day x Prone knee x day x day x r.98.9.98.9.8.79.78.96.6.5.9.9.98.8.70.3.7.97.3..68.59.99..3.8.86.99.69.55 s(cm) 9.39 9.5 9.39 9.58 9.69.99.97.87..8..39.9.5.58.6.57..9.5 6.09 6.09 5. 7.8 7..89.78.60 5.36 5.39 SEM (cm).33.33.33.87.5 0.9 0.93 0.38.38.69.9.3 0.7.9.50. 0.85 0.7.3.0 3.3 3.90 0.55 5.37 5.98.05.79 0..98 3.60 TABLE Analysis of Variance orward Bending Error Error x day Error x x x day x a p <.05. b p <.0. 58.58 55.90 05.5 0.3 373.09 06.5 9.0 53.9 33.8 53.86 0.8 6.65 6.6 55.38 were found between days in forward, side, straight leg raising, and prone knee. Significant differences between the means across s were found in all motions except for rotation and straight leg raising. Significant interactions were found between rater and day for the motions of rotation, straight leg raising, and prone knee. Significant interaction between rater and s occurred for the motions of forward and prone knee. Significant interactions between day and time occurred only for the motion of rotation. By examining the mean differences (in the absence of interactions) and the reliability estimates, the authors were able to identify two types of constant change that affected the error measurements. The 3 9 9 9.9.3 05.5.0 86.5.8.55.55 8.3.67 5.09 3.5.57.69.56.66 a.60 b 0.39.97 b.5 0.93 orward had the highest single measurement (rater x day x ) reliability (.8), showing good reliability across raters, days, and successive s. Side had the next highest single measurement reliability of.70. All other measurements had poor single measurement reliability estimates. Measurements of forward, side, and prone knee showed good reliability across raters. orward, side, and prone knee had good measurement reliability across days. All motions had good measurement reliability across successive s. When the rater and day varied and the was held constant, forward and side measurements were the only ones to maintain good reliability estimates. No significant differences between the means for raters were found in forward. In all other motions, these differences between means for raters were significant. Significant differences in the means TABLE 5 Analysis of Variance Backward Error Error x day Error x x x day x a p <.05. b p <.0. 33.30 08.0. 89.60 5.05 7.5.07 33.37.37 3.05 0.70 7.99.7 8.7 3 9 9 Bending 6.65.5. 8. 7.5 0.60 0.5 5.07 0.3 0.35 0.35 0.39 0.5 0.3 3.68 a 0..57 b 0. 0.98 0.90.75 Volume 6 / Number 0, October 98 35
Error Error day Error x day a p <.0. TABLE 6 Analysis of Variance Side Bending 0.0 370.9 73.5 0.9.80 53.56 7. 76.0. 7.53 0.06 3. 0.7 0.97 3 9 9 60.0 8.05 73.5 9.7 5.90.6 3.6 3.83 0.53 0.5 0.03 0.70 0.8 0.5 7.5 a 8.88 a 5.07 0.9.0 0.0 0.0 TABLE 8 Analysis of Variance Straight Leg Raising Error Error x day Error x x x day x a p <.0. 07.8 06.8 75.63 08.66 6.8 8.69 503.7 76.3 3.6 65.39.69.3 3.60.9 3 9 9 53.9 5.3 75.63 5.59 3..8 75.6 3.09 0.9.80 0.85.00 0.90.5 0.0 a 9.9 a.7 3. a 0.5 0.8 0.58 authors have chosen to define stretching effect as a good reliability across s, a significant increasing difference in the means of range of motion between s, but no significant interaction present between the s. Learning effect is defined as a good reliability across days, a significant difference in the means between days, with a concomitant increase in range of motion. To the best of the authors' knowledge, these effects, in relation to trunk range of motion measurements, have not previously been discussed in the literature. orward illustrated stretching effects because the range of motion increased across three s with good reliability, and that change was significant. orward also illustrated learning effects in that the differences in means were significant between days, there was good reliability across days, and the range of motion improved on the second day. Similarly, side also exhibited both stretching and learning effects. Backward exhibited stretching effect but did not exhibit learning effect. Prone knee showed significant differences in means between s and between days, but there were no accompanying significant trends of increasing motion. Therefore, the authors concluded that these differences were random rather than results of learning or stretching. Rotation showed large random error and did not exhibit significance in the mean differences for rater, day, or in any factor. The mean straight leg raising was significantly different between days but the difference was random, so no learning or stretching effects were noted. Error Error day Error day a p <.05. b p <.0. TABLE 7 Analysis of Variance Rotation 978.80 3.68 0.3 58.7 0.79 0.5.9 75.5.86 36.9 5.0 3.66.09 3.96 3 9 9 89.0 6.8 0.3 6.90 0.0 0.5.09 5.99 0.7 0.0.60 0.30 0.7 0.36 7.5 b 0.0 0.89 3.69 a.78 8.75 b 0.76 TABLE 9 Analysis of Variance Prone Knee Bending Error Error day Error day a p <.05. b p <.0. 557.6 677.7 0.65 567.9 3.69 35.7 75.58 655.89 6.79 6.85.77 37.7.9 6.9 3 9 9 78.8.73 0.65.69 6.8 0.76 375.79.6.70 0.68.38 0.8 0.8 0.68 8.9 b 5.70 a 8.95* 6.36 b.8 a.69 0.7 36 PHYSICAL THERAPY
RESEARCH DISCUION Obtaining reliable measurements of trunk motion depended on two factors: eliminating random error inherent in the measurement process and accounting for constant error. Two methods of assessment were found to be effective in controlling random error: taking an average of successive s and using the same rater to measure the motion on different days. Using the same rater and averaging across successive s should improve the reliability of all measurements. Other factors need to be considered when controlling for random error. More specific definitions of bony landmarks would decrease the amount of error. The bony landmarks chosen might have affected the accuracy of the measurements more than the authors originally expected. or example, some of the landmarks were difficult to palpate on obese subjects. Some landmarks were not designated specifically enough for the raters to believe they could be consistent. or example, the - to -cm difference between the proximal and most distal aspect of the spinous process of C7 might have accounted for as much as 50 percent of the excursion range of backward measured in some subjects. This amount of variation would preclude accurate assessment of backward regardless of the method used. As did the other studies cited previously, this study showed that one technique of measurement does not appear to be adequate in accurately measuring all trunk motions and the two associated motions. orward, side, and prone knee lend themselves well to this measurement technique. In view of its good measurement reliability, prone knee as a tension sign should be considered for wider clinical use. Backward showed poor measurement reliability in all instances except across s. Redefining the landmarks might improve the measurement. There is such a limited amount of excursion in backward that any minor error will strongly affect the accuracy of the measurement. Rotation and straight leg raising exhibited the same types of inconsistencies of measurement as backward. In addition, other problems encountered in straight leg raising were that the raters had to lift and hold the weight of the lower extremity, maintain relaxation of the subject, and judge the end of range as indicated by rotation of the pelvis. Measurements of both of these motions were so inconsistent for the normal subjects in this study that perhaps therapists in the clinic should use these motions only to assess the presence or absence of pain with movement rather than the extent of limitation caused by pain. The standard error measurement was valuable in discriminating between error and actual changes in the range of motion. Clinically, this value would assist the therapist in determining progress. or example, the increase in range of motion must exceed the standard error of measurement to show true progress. The special effects of stretching and learning were identified. They can be used to qualify progress, or their effects can be controlled. The clinical implications of both of these effects need to be studied in more detail. It would appear that averaging over three successive measurements or taking the measurement on the same each time would negate the stretching effect the authors defined. Learning effect was present over the first two measurement sessions. More investigation is necessary to determine how extensive the learning effect is over time. SUMMARY In summary, a standard measurement system for all trunk motions and two related motions was studied. Problems were identified that have not been mentioned in previous studies measuring back motion. Stretching and learning effects, as defined by the authors, were identified. The measurement technique evaluated in this study was determined to be effective for forward, side, and prone knee. The technique was not effective for backward, trunk rotation, or straight leg raising. Acknowledgment. Special appreciation is expressed to Karen Hoffman, PhD, Associate Professor of Medical Education, University of Southern California, Los Angeles, CA, for her development of the statistical rationale and computer programming necessary for the data analysis. REERENCES. Moll JMH, Liyanage SP, Wright V: An objective clinical method to measure spinal extension. Rheumatology and Physical Medicine :93-3, 97. Anderson JAD, Sweetman BJ: A combined flexi-rule/hydrogoniometer for measurement of lumbar spine and its sagittal movement. Rheumatol Rehabil :73-79, 975 3. Sturrock RD, Wojtulewski JA, Hart D: Spondylometry in a normal population and in ankylosing spondylitis. Rheumatol Rehabil :35-, 973. Hart D, Strickland D, Cliffe P: Measurement of spinal mobility. Ann Rheum Dis 33:36-39, 97 5. Macrae I, Wright V: Measurement of back movement. Ann Rheum Dis 8:58-587, 969 6. Moll JMH, Wright V: Normal range of spinal mobility. Ann Rheum Dis 30:38-386, 97 7. Moll JMH, Liyanage SP, Wright V: An objective clinical method of measuring lateral spinal flexion. Rheumatology and Physical Medicine :5-39, 97 8. Reynolds PMG: Measurement of spinal mobility: A comparison of three methods. Rheumatol Rehabil :80-85,975 9. Winer BJ: Statistical Principles in Experimental Design, ed. New York, NY, McGraw-Hill Book Co, 97, pp 83-96 0. Currier DP: Elements of Research in Physical Therapy. Baltimore, MD, The Williams & Wilkins Co, 979, pp 59 Volume 6 / Number 0, October 98 37