Investigation of the Validity of ~ostural Evaluation Skills in Assessing - Lumbar Lordosis Using - Photographs of Clothed Subjects1

Transcription

1 Investigation of the Validity of ~ostural Evaluation Skills in Assessing - Lumbar Lordosis Using - Photographs of Clothed Subjects1 JEAN M. BRYAN, MA, MPT2, EILEEN MOSNER, MPP, RONALD SHIPPEE, PhD2, MARGARET A. STULL, MD4 Journal of Orthopaedic & Sports Physical Therapy The purpose of this study was to evaluate the validity of physical therapists' visual postural evaluation skills in assessing lumbar lordosis using photographs of clothed subjects compared to actual radiographic measurements. The study also addressed the efficacy of the use of a plumb line in postural assessment. Forty-eight physical therapists (raters) participated. After completing a demographic questionnaire, raters were given a set of pictures showing sagittal views of three subjects and asked to rank order the subjects from most to least amount of lumbar lordosis. Raters were then given a second set of pictures with different subjects to evaluate. During this second trial, raters were given a plastic overlay with a hairline to use as a plumb line. Each rater evaluated one set of subjects with 7-8' of difference in lordosis and one set with O of difference as measured radiographically. The raters' rank order of the pictures was interpreted as either correct or incorrect. For 96 trials, 9 responses were correct for an accuracy rate of 9.3%. The use of the hairline as a plumb line did not improve the raters' accuracy. A chi-square test showed no relationship between ratings except under circumstances related to the gluteal prominence body contour. The strongest trend in interrater reliability identified was that the raters' perception of increased lordosis may have been influenced by gluteal prominence. The results indicate low validity in assessing relative amounts of lumbar lordosis using photographs of clothed subjects. This preliminary study indicates the need for more research in this area using both pictures and live subjects. Visual standing postural evaluation is an important part of most patient evaluations. Specifically the amount of lumbar lordosis is assessed during any evaluation of a patient complaining of low back pain. However, subjectively quantifying varying amounts of lumbar lordosis is difficult. Bohannon (1) suggested that postural assess Data for this study was collected at the 1988 Texas Physical Therapy Association Annual Conference with approval through the state Research Committee. The opinions or assertions contained herein are the prlvate views of the authors and are not to be construed as official or reflecting the vlews of the Army. the Department of Defense, or the US. Government. Faculty. US Army-Baylor University. Graduate Program in Physical Therapy. Academy of Health Sciences. Ft. Sam Houston. TX Staff Therapist. U.S. Army Hospital. West Point. NY Department of Radiology. Georgetown University Reservoir Road. Washington. DC /90/ $02.00/0 THE JOURNAL OF ORTHOPAEDIC AND SPORTS PHYSICAL THERAPY Co~nght by The Orthopaedic and Sports Physical Therapy Sections of the American Physical Therapy Association ments of lumbar lordosis can be reduced to a dichotomous option for research purposes-i.e., the patient either has a normal amount of lordosis or an abnormal amount. He further addressed the possibility of rating such assessments by ranking observations on an ordinal scale-i.e., less than normal, normal, greater than normal. Individual definitions of such terms are usually experienced based and have questionable interrater reliability. In addition, the validity of these terms compared to actual radiographic measurements of lumbar lordosis has not been addressed. At present, radiographs are the only valid assessment of lumbar lordosis (8); however, radiographs are costly as well as time consuming and are not without health risks. In the last few years, numerous studies have evaluated other, more objective measures to quantify lumbar lordosis. Such measures include the flexible ruler (2, 6, 12), tape measure (3, 5), parallelogram goni- 24 BRYAN ET AL JOSPT 12:1 July 1990

2 ometer and standard goniometer (3, 5), and inclinometer or gravity goniometer (3, 8, 9). While interrater reliability for these measures tends to be high, the validity of these measures compared to radiographs has not been as well addressed. In the case of the flexible ruler, validity studies have had conflicting findings with one study showing a high correlation (6) and another showing a low correlation (2). Researchers have not shown any of these measures to be clearly superior to the others. More research, especially on validity of these measures, is indicated. Until a reliable, readily available measure of lordosis is shown to have high validity compared to radiographs, clinicians will continue to use visual assessment and subjective definitions of lumbar lordosis in patient evaluations. To evaluate the spinal postural curves, including lumbar lordosis, the clinician should assess the patient from the side view in the sagittal plane. According to Kendall and McCreary (7), to evaluate postural deviations a clinician must be able to visualize the "ideal posture" as a basis for comparison. They advocate the use of a plumb line for postural assessment to standardize points of reference. They also state that postural alignment should be based on skeletal alignment and not body contours. However, they state that body contours are correlated to some degree with variations in skeletal alignment and draw the conclusion that the experienced clinician can estimate the skeletal alignment by observing body contours (7). To date, very little research has addressed the accuracy of these visual postural evaluations as compared to radiographic measures. Mosner et al. (1 0) investigated the issue of body contours and differences in lumbar lordosis between black and white females. Their findings indicated that clinicians' postural assessments of lumbar lordosis may actually be incorrectly based on body contours rather than skeletal landmarks. Thus, the research question is, how accurate are physical therapists at visually evaluating and ranking varying amounts of lumbar lordosis? The purpose of this study was to evaluate the validity and reliability of physical therapists' postural evaluation skills in visually assessing varying amounts of lumbar lordosis using photographs of clothed subjects as compared to actual radiographic measurements. lnterrater reliability was also evaluated. The study addressed the efficacy of the use of a "plumb line" in postural assessment. METHODS Fortyeight physical therapists (raters) were recruited and participated in this study at the 1988 Annual Conference of the Texas Physical Therapy Association. Prior to participation, raters com- pleted a demographic questionnaire. Of the 48 raters, the average amount of experience as a physical therapist was 14.6 years. Thirty-seven (77%) stated they treated primarily musculoskeletal patients. Twenty (42%) said they "always" performed a standing postural evaluation, and 22 (46%) said they "usually" did. From fixed choices of Excellent, Good, Fair, or Poor, 6 raters (12%) self-ranked their ability to assess posture as Excellent and 33 (69%) as Good. Twenty-six raters (54%) had received advanced training in postural evaluation. After completing the questionnaire, raters were given a set of pictures (photographs) showing sagittal views of three subjects in shorts and T-shirts pulled tight to the back and were asked to rank order the subjects from most to least amount of lumbar lordosis. Raters were then given a second set of pictures of three different subjects and asked to rank order the amounts of lumbar lordosis in the same manner; however, during this second trial, raters were given a plastic overlay with a hairline to use as a plumb line during their evaluation. Each rater evaluated one set of subjects with 7-8' of difference in lordosis and one set with 12-14' of difference. The actual lumbar lordosis was measured from a standing lateral radiograph of the lumbosacral spine as the angle formed at the intersection of lines drawn from the top of the second lumbar vertebra and the top of the sacrum. Three black and three white subjects with varying degrees of lordosis were used in the pictures (Table 1). The research protocol for taking the radiographs and pictures was approved through the Brooke Army Medical Center Department of Clinical Investigations, Ft. Sam Houston, TX. Subjects in the pictures were between the ages of 18 and 35, had no history of back problems, and met the Army height and weight standards (1 3). The subjects grasped a pole in front of them at waist height to remove their arms from the radiographic beam. Stagnara et al. (11) showed that this modification to the normal postural evaluation stance did not significantly change the amount of lumbar lordosis. The pictures were taken in the same position as the radiographs and were full length sagittal views. The subject's face was covered but the ear showed. The subjects' gluteal prominence, a measure of the distance from the greater trochanter to the most posterior gluteal aspect as described by Mosner et al. (1 0) is also shown on Table 1. All measures were taken in shorts and T-shirts. Pictures of these six subjects were divided into four sets of three subjects each as shown in Table 2. Raters viewed either sets A and B or sets C and D. In this arrangement, each rater evaluated a set of pictures where the amount of lordosis varied from 7 to 8' (40, 48, and 55O of JOSPT 12: 1 July 1990 VALIDITY OF POSTURAL ASSESSMENT 25

3 TABLE 1 Lordasis and gluteal prominence measures for subjects Subjects Lordosis Gluteal prominence 0 cm 1. White female 40, Black female White female Black female White female Black female TABLE 2 Composition of rated picture sets Pire Set Subject (Number' RaceT Lordosis Range A W,W,W 40<48<55 B B, B. B 43 < 55 < 69 C W,W,B 40<48<55 D B. W. B 43 < 55 < 69 Subject numbers as in Table 1. Race: W, white; B, black. lordosis) and one that varied ' (43, 55, and 69' of lordosis). Twenty-three therapists rated picture sets A and B. In the same manner, 25 rated picture sets C and D for a total of 96 trials. The order of presentation of picture sets was randomized to control for the learning effect. The individual pictures in each set were marked with either a circle, a square, or a triangle to avoid any bias due to numbering the pictures. When the raters rated the pictures by amount of lordosis, the investigator recorded the rank order utilizing these symbols. DATA ANALYSIS The raters' rank order of the pictures from the least to the most amount of lumbar lordosis was interpreted as either correct or incorrect. Analysis of this study was based on descriptive statistics for the 96 trials. The chi-square (4) distribution at a significance level of 0.05 was used to evaluate the independence of the raters' ranking of lordosis for each set of pictures. RESULTS For 96 trials, only 9 responses were correct for an accuracy rate of 9.3%. Four therapists rated set A, one rated set B, none rated set C, and 4 rated set D correctly. The use of the hairline as a plumb line did not apparently improve the raters' accuracy as only two of the correct responses were made using the hairline. The chi-square tests for picture sets A and C showed no correlation between ratings (Table 3). However, for picture sets B and D there was a statistically significant correlation between ratings TABLE 3 Frequency distribution of lordosis ratings and chi-square results comparing ratings with radiographic lordosis Lordosis Rating Rating Least Middle Greatest Set A 1 (40) (48) (55) 6 7 Chi-square = (df = 4) p = Set B Subject* 2 (43) (55) (69) 8 12 Chi-square = (df = 4) p = 0.000' Set C 1 (40) (48) (55) 8 10 Chi-square = (df = 4) p = Set D Subject* 2 (43) (55) (69) Chi-square = (df = 4) p = O.OOOT ' Subject numbers as in Table 1 ; lordosis in degrees in parentheses. Statistically significant correlation. at a probability level of ~ The observed rankings of lordosis were in the opposite direction of the expected rankings for sets B and D. The rankings were negatively correlated and incorrect when compared to the radiographic measures of lordosis. For example, in set D, subject 2, with the least amount of radiographically measured lordosis, was rated as having the greatest amount of lordosis in 19 out of 25 trials. DISCUSSION Based on the 9.3% correct response rate, the raters' ability to accurately assess relative amounts of lumbar lordosis was poor. Several factors may have contributed to this low rate of accuracy. The design of this study was not strictly clinical since therapists are usually assessing live patients and not pictures; however, pictures are used for clinical postural assessment and are often used in place of repeated radiographs as an index of effected postural change in a patient. In addition, for purposes of this study, pictures ensured that each rater saw exactly the same view and allowed for a greater number of raters 3bviously, pictures are only a two-dimen+mal representation of the three-dimensional suoject, but if the assessment of postural curves IS truly based on a sagittal view, the photqrapn should suffice. 26 BRYAN ET AL JOSPT 12: 1 July 1990

4 The individuals in the pictures were wearing shorts and T-shirts. The shorts and shirts were arranged so that they were smooth over the skin surfaces of the low back and buttocks. Thus, the picture showed a true reflection of the subject's body contours as seen from the sagittal view. This contour showed the outline of the lumbar paraspinal muscles and not the lumbar spinous processes. In a sagittal view, even without clothes, the evaluator views the outline of those same paraspinal muscles. Even so, the subjects' clothes may have attributed to the decreased accuracy of rating lordosis since clinically visual postural assessment should be done with the area exposed (7). The expected frequency distribution if all raters had correctly rated set A is shown in Table 4. If the raters were consistently accurate, this general pattern would be expected for all four picture sets. This pattern was not demonstrated in any of the four sets as shown in Table 3. As shown by the chi-square test, there was no correlation between ratings in picture sets A and C. As an example, in set A, raters rated subject 1, who had the least amount of lordosis, almost equally as having the least, middle, and greatest amount of lordosis with frequencies of 8, 8, and 7, respectively. This rather even distribution of ratings is an indication of very low interrater reliability in sets A and C. In sets B and D, the chi-square test did show a significant correlation between the ratings. In evaluating the ratings of lordosis for these sets, the observed frequencies were in the opposite direction of the expected frequencies. Subject 2 in both these sets had the least amount of lordosis (43') but was consistently rated as having the greatest amount in 37 out of 48 trials (77%). The rankings were negatively correlated and thus were not valid compared to the actual radiographic measurements of lumbar lordosis. One explanation of this high correlation for sets B and D may be the subjects' gluteal prominence measurements. Table 5 shows the subjects in each picture set listed in order of gluteal prominence and the rating of amount of lordosis for each subject. Note that the statistical correlation TABLE 4 Expected frequency distribution if all raters correctly rated Set A from least to greatest amount of lumbar lordosis Lardasis Rating Rating Least Middle Greatest Set A 1 (40) (48) (55) ' Subject numbers as in Table 1; lordosis in degrees in parentheses. TABLE 5 Frequency distribution of lordosis ratings and chi-square results comparing ratings with gluteal prominence Gluteal Prominence Rating Least Middle Greatest Set A 1 (12.7) (12.9) (13.5) Chi-square = (df = 4) p = Set B Subject* 6 (15.4) (17.1) (19.9) Chi-square = (df = 4) p c 0.001' Set C 1 (12.7) (12.9) (17.1) Chi-square = (df = 4) p = Set D 6 (15.4) (13.5) (19.9) Chi-square = (df = 4) p < 0.001' ' Subject numbers as in Table 1 ; gluteal prominence measures in cm in parentheses. ' Statistically significant correlation. is the same as in Table 3 since the chi-square only evaluates the independence of the rater's ranking of subjects and not how the ratings relate to radiographic measures of lordosis or gluteal prominence. Table 3 shows how those ratings relate to actual lordosis and Table 5 relates the ratings to gluteal prominence. In picture sets A and C, the gluteal prominence difference between subjects was not as great, and the raters' choice of the subject with the greatest amount of lordosis was more evenly distributed. The rating is not evenly distributed in sets B and D, as evidenced by subject 2 who had the lowest amount of lordosis (43O) and the largest measure of gluteal prominence (1 9.9 cm). Referring to Table 4, which is still the expected pattern for a perfect correlation, the significant correlation in sets B and D appears to be in a positive direction so that large measures of gluteal prominence were rated as large measures of lordosis. Even though the correlation appears to be positive, the correlation is still not valid when compared to actual radiographic measures of lordosis. Although visual postural assessment of lumbar lordosis should be based on skeletal alignment rather than on body contours (7), in this case the gluteal prominence body contour may have influenced the raters' perception of the JOSPT 12: 1 July 1990 VALIDITY OF POSTURAL ASSESSMENT 27

5 amount of actual lumbar lordosis. Mosner et al. (1) explored this phenomenon in their investigation of the clinical perception that blacks have more lordosis than whites. They found that black females did not have any more lordosis than whites but did have significantly greater gluteal prominence measures. They concluded that if the evaluator is looking at body contours rather than identifying skeletal alignment, a large gluteal prominence would tend to accentuate the actual lordosis and could affect the accuracy of assessing lumbar lordosis (1 0). The larger gluteal prominence seen in subject 2 may well have affected the rater's ranking of amounts of lordosis and may explain the high correlation between ratings seen in sets B and D. Although this study was not specifically designed to investigate clinician bias that blacks have greater amounts of lordosis, some inferences are possible. Picture sets C and D required the raters to differentiate between black and white subjects. In set C raters did not consistently rate subject 4 (black) as having the greatest lordosis. In this case, however, rating subject 4 as greatest was accurate; therefore, raters were apparently basing their ratings on some criterion other than race. In set D, raters did not consistently rate subject 5 (white) as having the least amount of lordosis. The correlation seen in the ratings of set D appears to be related to gluteal prominence as previously discussed and not related to race. Kendall and McCreary (7) discuss visual postural assessment on live subjects with direct visualization of the back, not on pictures. One view of a subject is obviously limiting, but the view offered in the pictures is the "classical" or sagittal view for making these postural assessments (7). But, due to paraspinal muscle bulk, this view does not allow the clinician to appreciate spinal curves as they are reflected on skin surface. With a posterior view, the clinician can view the depth of the lumbar curve. A picture of a posterior view may not allow this same visual perception. Further research to validate the use of pictures in postural assessment is indicated. Subjects should have on bikini type bathing suits so that skeletal landmarks are exposed. Based on these findings, the answer to the research question is that physical therapists' visual postural assessment skills may have high interrater reliability (sets B and D) but have low validity as evidenced by less than 10% accurate rating of lumbar lordosis. These conclusions are based on using pictures of clothed subjects rather than live subjects for postural assessment. However, even considering the identified problems with using pictures rather than live subjects, the low overall accuracy of rating lordosis and the negative correlation seen in sets B and D raise some significant questions about physical thera- pists' visual postural assessment skills. Not only is further research on postural evaluation using pictures indicated, this study indicates the need for further research on postural evaluation with live subjects to allow the rater to view the subject from all planes and to palpate bony landmarks. To claim the ability to accurately assess postural curves, we would think that clinicians should be correct at least 50% of the time. Clinically, therapists do not usually directly rate amounts of lordosis between patients but, rather, compare a single patient to the "postural ideal" (7). The results of this study may indicate that therapists have not visualized that "ideal" as well as they may have thought or they may be assessing body contours rather than skeletal alignment. Based on this concept, perhaps classes on postural evaluation should include practical exercises where students rank the amounts of lordosis between subjects and then compare those rankings to actual radiographic measurements. Or, altematively, clinicians can test themselves and improve their evaluation skills by comparing their assessments of patients to the patients' radiographic measurements of lordosis. This practical exercise would help to increase the validity of postural assessments. CONCLUSION The purpose of this study was to evaluate the validity and reliability of visual postural evaluation skills in assessing lumbar lordosis using photographs of clothed subjects. Forty-eight physical therapists (raters) rated subjects from least to greatest amount of lordosis. In 96 trials, only 9 raters correctly rated the amounts of lordosis. A chi-square test showed a high correlation between ratings for two picture sets; however, the observed rankings were not valid when compared with radiographic measures of lordosis. Raters may have been making their decisions based on the gluteal prominence measurement. The results of this study raise some question about physical therapists' ability to accurately assess varying amounts of lumbar lordosis, using photographs of clothed subjects. These conclusions are tentative and subject to more research as they are applying the "picture" model to the clinical setting. 0 The authors acknowledge Ken Finstuen. PhD. David G. Greathouse. PhD, and Pamela Stanton. EdD. REFERENCES 1. Bdrannon RW: Simple clinical measures. Phys Ther 67: Bryan JM. Mosner EA. Shippee R. Stull MA: Investigation of the flex~ble ruler as a noninvasive measure of lumbar lordosis in Mack and white adult female sample populations. J Orthop Sports Phys Ther 11 : Burden RG. Brown KE. Fall MP: Reliability and validity of four 28 BRYAN ET AL JOSPT 12: 1 July 1990

6 instruments for measuring lumbar spine and pelvic positions. Phys Ther Daniel WW: Biostatistcs: A Foundation for Analysis in the Health Sciences. Ed 3. pp New York: John W~ley 8. Sons 5. F~tzgerald GK. Wynveen KJ. Rheault W. Rothsch~ld B: Objective measurement of normal values for lumbar spinal range of motion. Phys Ther 63: Hart DL. Rose SJ: Reliability of a noninvasive method for measuring the lumbar curve. J Orthop Sports Phys Ther 8: Kendall FP. McCreary EK: Muscles and Function. Ed 3. pp Balt~more: Williams 8. Wilkins Mayer TG. Tencer AF. Kr~stoferson S. Mooney V: Use of noninvaslve techniques for quantifcat~on of spinal range of motion in normal subjects and chronic low-back dysfunction patients. Spine 9: Mdl J. Wright V: Measurement of spinal movement. In: Jayson M. The Lumbar Spine and Back Pain. pp New Yo*: G~ne 8. Straton Mosner EA. Bryan JM. Stull MA. Shippee R: A comparison of actual and apparent lumbar lordosis in Mack and white adult females. Spine 14: Stagnara P. DeMauroy JC. Dran G. Gooon GP. Costanzo G. Dimnet J. Pasquet A: Reciprocal angulation of vertebral bodies in a sagittal plane: approach to references for the evaluation of kyphosis and lordosis. Spine 7: Stokes IFA. Bevins TM. Lunn RA: Back surface curvature and measurement of lumbar spinal motion. Spine 12: US Army Regulat~on 600-9: Army Weight Control Program. L)pdate. October Journal of Orthopaedic & Sports Physical Therapy JOSPT 12: 1 July 1990 VALIDITY OF POSTURAL ASSESSMENT