Development of a Noninvasive Measure of Pelvic and Hip Angles in Seated Posture

Transcription

1 1597 Development of a Noninvasive Measure of Pelvic and Hip Angles in Seated Posture Stephen Sprigle, PhD, PT, Mary Wootten, MS, BME, Mark Bresler, MEBME, PE, Nanette Flinn, MS, PT ABSTRACT. Sprigle S, Wootten M, Bresler M, Flinn N. Development of a noninvasive measure of pelvic and hip angles in seated posture. Arch Phys Med Rehabil 2002;83: Objective: To develop a valid noninvasive means to measure pelvic tilt and hip angle in seated posture. Design: Validation cohort study using radiographs as a criterion standard for pelvic posture. Setting: Rehabilitation hospital. Participants: Volunteer sample of 10 adult men with no known physical disability. Interventions: Radiographs were taken as subjects sat in erect, anterior, and posterior postures. An electromagnetic tracking device was as a pointer to digitize the anterior superior and posterior superior iliac spines and as a 6 degrees of freedom (df) sensor mounted on the thigh and sacrum. Main Outcome Measures: Variables included pelvic tilt and hip flexion angle. Intra- and interrater reliability of radiographic measures was determined by using intraclass correlation coefficient comparison of the results from 2 investigators. Validity was determined by comparing noninvasive measures of pelvic and hip angles to radiographic measures by using correlation, analysis of variance (ANOVA), and regression. Results: Reliability of radiographic measures of pelvic tilt and hip angles were.98. Pelvic tilt comparison: pointer: r.89, R 2.80; 6-df sensor: r.91, R 2.83; hip angle comparison using 6-df sensor: r.78 with average difference of ANOVA showed that differences between all invasive and noninvasive measures did not differ significantly (P.05). Conclusion: Results indicated excellent reliability of radiographic analysis techniques and represented an improvement over previously published techniques. Noninvasive measures of pelvic tilt and hip angle were shown to be valid. Key Words: Posture, Pelvic; Rehabilitation; Wheelchairs by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation AN OBJECTIVE AND NONINVASIVE measure of pelvic position is an important factor when studying the posture of wheelchair users. Several pelvic and spine posture measurement approaches have been documented including goniometers, 1-3 inclinometers, 1,4 calipers, 5,6 spatial digitization, 7 imaging, 2,8 and video. 3 A problem arises when trying to apply From the Center for Rehabilitation Technology, Helen Hayes Hospital, West Haverstraw, NY (Sprigle, Wootten, Bresler); and Program in Physical Therapy, Mercy College, Dobbs Ferry, NY (Flinn). Supported by the National Institute on Disability and Rehabilitation Research and the New York State Department of Health. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Stephen Sprigle, PhD, PT, Crawford Research Institute, Shepherd Center, 2020 Peachtree Rd NW, Atlanta, GA 30309, gogators@compuserve.com /02/ $35.00/0 doi: /apmr these techniques to the seated posture. The seat and backrest obstruct portions of the pelvis and posterior trunk, making measurement of seated posture difficult. Currently, no instrumentation and techniques have been described and validated that noninvasively measure pelvic and hip angles in a seated posture. Arguably, the seated posture cannot adequately be studied without measuring pelvic tilt. This is especially true during study of wheelchair users who exhibit a wide range of postures. Pelvic tilt is defined as rotation of the pelvis in a sagittal plane around a mediolateral axis. Rearward rotation termed posterior pelvic tilt results in spinal kyphosis, and forward rotation or anterior pelvic tilt tends to result in a forward leaning or hyperlordotic erect posture. Anterior and posterior pelvic tilt directly influence the amount of lordosis and kyphosis, respectively, and can hinder measurement of true hip angle. 9,10 Standard hip flexion measurements are taken with the person supine by using the midline of the trunk as reference. This approach does not transfer well to seated posture. During study of wheelchair users and others who sit for extended durations, pelvic orientation influences the load-bearing structures of the buttock region that impacts stability, comfort, and risk of pressure ulcer formation. For example, rotating into a posterior pelvic tilt is typical of a slouched, kyphotic posture during which the sacro-coccygeal region rotates posteriorly and downward to contact the seat surface. This additional point of contact can increase postural stability but also results in load applied to the tissues over the sacrum and coccyx. These tissues are poorly equipped to bear weight and resulting ischemia will lead to a range of complications, from discomfort to pressure ulcer formation. In addition to seeking a more stable posture, wheelchair users may also adopt a slouched posture for anatomic reasons (eg, tight hamstrings, lack of lumbar extension) or in response to poor-fitting wheelchair configurations (eg, seat depth that is too long, tight seat-to-back angle). Radiographs have been accepted as a means to measure the relationships between bones, which is a significant aspect of posture. For reasons of safety and their limitations to static states, radiographs are not used to study function and posture. They do serve as a means to determine concurrent validity in goniometric studies; however, problems arise when using radiographs to study the relationships between the pelvis, lumbar spine, and femur. These 3 structures are vital for seated posture study but do not appear well in typical radiographic series. Vertebrae, pelvic spines, and femur are structures with different radio-opacity and are located in different depths of field. Gilliam et al 11 used radiographs to determine the relationship between pelvic tilt and sacral angles as a part of a study to determine reliability and validity of goniometric measures of pelvic tilt. Reliability of the goniometric measures was excellent (.98), but validity was not shown because correlation between radiographic and goniometric measures of pelvic tilt by radiologists was lower (.85,.68, respectively). Furthermore, a poor correlation ( 0.6) was noted between radiographic measures of sacral angle and pelvic tilt. Gilliam 11 attributed the

2 1598 NONINVASIVE MEASUREMENT OF SEATED POSTURE, Sprigle low correlation to the difficulty in identifying the iliac spines and sacral plateau in the radiographs. This result was similar to another study 1 that used an inclinometer and a modified standard goniometer to measure pelvic tilt. This study used the sacrum as a measure of pelvic tilt. The reliability of the 2 pelvic tilt instruments was in the good range (.82,.87). However, validity was not shown because the correlation with radiographic measurements was very low (r.10), and differences in pelvic tilt between radiographic and noninvasive techniques were 20 (inclinometer) and 15 (modified goniometer). Analysis of variance (ANOVA) determined that both noninvasive measures differed significantly from the radiographic measures. Validating sacral movement with video was performed in a study of volitional pelvic tilt in standing subjects by using 3-dimensional video and a piezoresistive accelerometer configured as a goniometer to measure posture. 3 The ranges of movement measured by the 2 systems were within 1 of each other and the correlation between them were.84 and.97 on the 2 days of testing. Advanced instrumentation such as a spatial digitizer 7 and an electromagnetic tracking device 8,12 (ETD) have been used to measure posture, but, although their accuracy and reliability was demonstrated, their validity was not determined. Of particular relation to our study is the investigation by Bull and McGregor 8 of lumbar motion during rowing. They affixed ETD sensors to the spine and used magnetic resonance imaging to determine skin movement relative to the underlying bone. Average errors in measuring spinal movement because of softtissue artifact were 1 (standard deviation [SD], 1.0 ). 8 ETDs have unique features for postural measurement. The small sensors can be configured as a pointer for spatial digitization of static posture and as a 6 degrees of freedom (df) sensor for dynamic measurements. However, they are susceptible to interference from metal objects, so their applicability in measuring the posture of subjects in wheelchairs and other metallic chairs has been suspect. Part of this limitation can be overcome by use of a nonmetallic test chair, but concerns exist about the effects from spinal fixation instrumentation. Three studies have addressed these concerns and reflect the potential utility of ETDs for this project. Milne et al 13 studied the positional and rotational accuracy of an ETD in the presence of orthopedic implants fabricated from different metals. They found that aluminum and titanium metals used in orthopedic implants and wheelchair construction did not reduce accuracy. The mean positional error was.50mm, and sensitivity was enough to permit a resolution of.25mm. Bull et al 14 found that mild steel has no effect on accuracy when positioned more than 150mm from the sensors and transmitter. Koo et al 12 modified an ETD to measure pelvic orientation of wheelchair users. They housed the ETD sensor in a nylon and acrylic pointer to minimize the effect of a standard crossfold wheelchair (assumed to be low carbon steel) on which subjects were seated. A systematic investigation of accuracy and repeatability found their system to have positional error of 0.7% (1.4mm error in a 20cm measure) and an SD in their pelvic orientation angle measures of The objective of this study was to develop a valid noninvasive means to measure pelvic tilt and hip angle. This objective was met by 2 tasks. The first task developed a reliable radiographic approach to identify the bony landmarks used to define pelvic tilt and hip angle anterior superior iliac spine (ASIS) and posterior superior iliac spine (PSIS) line of the femur and sacral plateau. This task focused on the techniques used to obtain useful radiographs of the regions of interest. The second task used radiographic measurements to determine the validity Fig 1. Sacrum and thigh FOB sensor attachments. of an ETD in measuring pelvic tilt and hip angle in seated posture. METHODS Instrumentation Noninvasive postural measurement was done using a Flock of Birds (FOB) system. a The system uses magnetic field pulses to track the position and orientation of individual sensors relative to a transmitter that is fixed in space. A microprocessor controls the transmitting and sensing signals and converts them into position and orientation information. According to the manufacturer, sensors have a range of approximately 1m, with accuracy within 2%. Two approaches to measure pelvic tilt were investigated. Configuring an FOB sensor as a pointer allowed for digitizing the locations of the ASIS and PSIS, which were then used to calculate pelvic orientation. This FOB pointer approach is limited to measuring static pelvic posture. The second approach involved mounting an FOB sensor to the pad of a sacro-iliac belt. The sensor was configured to measure 6-df positional location and orientation. A second FOB sensor was mounted to the thigh (fig 1). By measuring the angles of the sacrum and thigh in the sagittal plane, this approach permitted dynamic measurement of changes in pelvic orientation and hip flexion angle. Within this study, the 2 sensor configurations are referred to as FOB pointer and 6-df sensor. The 2 approaches were used to determine the relationships and level of agreement between them for applications requiring both static and dynamic monitoring. Additional explanation of these applications can be found later in this study. All radiographs were taken with an Okamoto high-speed cassette b and 800-speed film. The use of high-speed film permitted exposures to be taken by using lower-amperage intensity (mas), resulting in less radiation exposure to the subjects. The same imaging room was used for all radiographs. The cathode tube was aligned to expose the entire pelvis and part of the femur. The radiograph was taken at amperage and voltage settings determined from the subject s hip width using a radiographic technique calculator. c For example, settings of 95kVp and 30mAs would be used for a subject with a 33-cm hip width, with these values adjusted for smaller and larger subjects. These starting values represented the most appropriate combination based on use of the high-speed cassette and film system and need for adequate contrast of the required structures.

3 NONINVASIVE MEASUREMENT OF SEATED POSTURE, Sprigle 1599 Procedures Ten men volunteered from a convenience sample. Inclusion criteria included being older than 18 years of age, and exclusion criteria included radiographs of the pelvic, hip, and lumbar region within the past 12 months. Subjects ranged in age from 25 to 48 years, in height from 170 to 193cm, and a body mass from 70 to 105kg. After procedure, risks, and benefits were explained, subjects signed a consent form that had been approved by the hospital s institutional review board. Subjects donned shorts with a slit along the outer seam to allow placement of sensors along the thigh. Subjects hip width was measured by using calipers so that the settings for the radiograph (amperage and voltage) could be determined. An extensive analysis of the literature and radiographic techniques was done to minimize risk to subjects. Acceptable levels of exposure to x-rays have been a concern for many years, with occupational exposure garnering the most attention. Occupational exposure limits vary across body site and varies somewhat according to which reference body s guidelines are used. The most common value cited is an annual full body exposure limit of 5rads and annual gonadal limits of 5 to 15rads. 15,16 The estimated threshold for temporary male sterility from a single exposure of radiation is 15rads. 17 Diagnostic x-rays, such as those proposed for this study, are not typically included in occupational limits, but lateral pelvic x-rays do expose reproductive organs to radiation that have been determined to have less tolerance than other body sites. The overriding philosophy of diagnostic radiology is to maintain exposure as low as reasonably achievable. Protection of patients by exposure limitations can be achieved by technical decisions. Amperage intensity is the primary influence of x-ray exposure. We used 2 specific technical interventions to minimize exposure: (1) increasing the voltage of the x-ray (kvp increase is often coupled to a decrease in ma) and (2) using a high-speed film (decreasing time [s] will decrease mas). A calculation of radiation absorption was made based on the equipment used, the amperage and voltage of the x-ray, pelvic width, and the distance from the x-ray head to the body and film. The skin entrance dosage was calculated as 1.85rads when using single-speed film and regular cassette but only as.67rads using the high-speed film and cassette purchased for this study. Most subjects received 3 x-rays, whereas a few required 4. Therefore the skin entrance dosage of the latter group was exposed to 2.68rads, which is well within occupational guidelines. Subjects were seated on a wood and plastic test chair whose seat angle was 5 from the horizontal with backrest angle of 95 to the seat. This chair was designed with adjustments common to office chairs and wheelchairs (eg, seat and back angle, seat depth, back height), although only the seat depth feature was needed for our study. One important characteristic of the test chair is a space between the seat and bottom of the backrest upholstery, which allows access to the sacrum. Once seated, the subject was fit with a sacro-iliac belt to which a 6-df sensor was mounted. The sacral pad of the belt was positioned between and slightly distal to the subject s PSIS. The leather iliac band was positioned under the iliac crest because it ran ventral, and a window in the belting was positioned over the ASIS. By ensuring that the belt ran under the iliac crests, the potential for sacral belt slippage was greatly minimized. The ASIS window allowed access for palpation and positioning of an FOB pointer for spatial measurements. A in plastic thigh bar was aligned from the greater trochanter to the lateral femoral condyle and secured to the thigh by a wide neoprene band. A 6-df sensor was attached to the thigh bar to measure thigh angle. A radio-opaque 1 4 2in metal rod was mounted to the distal midline portion of the thigh bar to indicate the line of the femur. A Velcro strap placed around the knees kept a consistent neutral thigh alignment during the radiographs. Immediately after exposure, the cathode tube was raised, and position angles were measured by using the 6-df sensors mounted at the sacrum and thigh. Locations of the ASIS and PSIS were measured by palpation by using the FOB pointer. After the noninvasive measurements were completed, the subject was allowed to relax. Radiographs were examined immediately to ensure the exposure permitted identification of the necessary bony landmarks. If unclear, voltage and amperage settings were adjusted as needed to improve contrast. If another radiograph was needed, the procedure described earlier was repeated for the position in question. If the radiograph was clear, then the procedure was repeated with the subject in the next position. Three postures were measured: (1) an erect posture during which subjects placed their posterior buttocks against a rear pelvic support on the measurement chair and were told to sit comfortably, (2) a forward posture during which the subjects leaned forward and rested their elbows on their knees, and (3) a posterior posture during which the subject s slid their buttocks forward in the seat and adopted a slouched posture. Postural variations were anticipated given the instruction to sit comfortably. The lack of postural control was volitional because variations exist across people and the anterior and posterior postures were relative to each subject s comfortable erect posture. Radiograph Measurement Three radiographic variables sacral angle, hip angle, and pelvic tilt were measured by using a methodology developed analyzing skeleton radiographs. Six radiographs were taken of 2 skeletons in different seated postures. Clear acetate, cut to the size of the radiograph, was mounted over the radiographs to allow for independent identification of anatomic locations. Because the ASIS has a broad protrusion, a technique to repeatedly identify the same portion was devised. A clear aligning square was positioned along the iliac crest and the perpendicular arm was drawn toward the ASIS until the square touched the anterior portion of the pelvis. The 90 corner of the aligning square was marked, and its bisector used to identify the point of the ASIS. The PSIS was determined as the most prominent aspect of the posterior iliac spine of the pelvis. Figure 2A shows pelvic and hip angle measures on a photograph, and figure 2B is a radiograph of the same skeleton (the metal attachment screws and wires are prominent), indicating the perpendicular lines that define the ASIS, the selected position of the PSIS, and the sacral angle measure. The line of the femur was indicated by the femoral image in the radiograph or by the line of the radio-opaque marker. Radiographs included a vertical reference line resulting from a chain hanging freely from the x-ray film cartridge. Variables FOB pointer tilt. FOB pointer tilt is the angle formed between the line connecting the ASIS and PSIS and the horizontal; it is calculated from digitized locations of the ASIS and PSIS measured by an FOB pointer. FOB hip angle. FOB hip angle is the angle formed between the thigh and the perpendicular of the line connecting the ASIS and PSIS; it is calculated from the digitized locations

4 1600 NONINVASIVE MEASUREMENT OF SEATED POSTURE, Sprigle Fig 2. Skeleton (A) photograph and (B) radiograph indicating anatomic structures and postural angles. of the ASIS and PSIS (by using an FOB pointer) and the angle of the thigh (by using a 6-df sensor). Sacral belt tilt. Sacral belt tilt is the angle of tilt of the sacrum to the horizontal; it is measured with a 6-df sensor mounted to a sacral belt. Radiograph sacral angle. The radiographic sacral angle is the angle between the line tangent to the plateau of S1 and the horizontal. Radiograph pelvic tilt. Radiographic pelvic tilt is the angle between the line connecting the ASIS and PSIS and the horizontal, measured from radiographs. Radiograph hip angle. The radiographic hip angle is the angle between the line of the femur to the line bisecting the line connecting the ASIS and PSIS. Reliability and Validity Testing Procedures Intraclass correlation coefficients (ICCs) were used to determine intra- and interrater reliability. A 1-way ANOVA was used to determine the mean square values used in the ICC calculations. The clinical significance of the coefficients was determined by the following values reported by Currier 18 : less than.69 as poor;.70 to.79 as fair;.80 to.89 as good, and.90 to.99 as excellent. Intrarater reliability of radiographic measures was calculated by using data from 2 investigators who marked the right and left ASISs and PSISs on 9 radiographs taken on 2 skeletons and 1 human subject. Each investigator marked the radiographs on 2 different days, separated by at least 3 days, by using clean acetates as described earlier. This reliability testing preceded the human subject testing. Interrater reliability of radiographic measures was calculated by using data from 2 investigators who independently marked all 30 human radiographs (10 subjects in 3 postures) with the variables calculated as defined earlier. FOB repeatability and accuracy was determined by a series of measures on the test chair subsequently used for human subject data collection. Grid paper with 1-in spacing was secured to the seat. An FOB pointer and 6-df sensor were placed on a series of 4 points. Accuracy of the FOB pointer requires measuring the location and orientation with respect to the position of the transmitter. Therefore, FOB pointer testing also used different orientations or lines of approach to the measurement point. At least 10 samples were taken at each point, and the series of points were measured 6 times. In addition to spatial location, angle accuracy was tested with the 6-df sensor by using a standard goniometer and rotation around different sensor axes. Validity of the noninvasive measurements was determined by comparison with radiographic measurements. The Pearson correlation coefficient, linear regression, and ANOVA were used for this comparison. Noninvasive measures of pelvic tilt and hip angle should have a linear relationship with radiographic measures. Sacral belt tilt, although not measuring the line formed by the iliac spines, is directly related to that line given the assumption that the sacrum can be considered fixed to the pelvis over the joint ranges and subjects studied. Studies have shown that the sacrum does not move on the pelvis except at extreme ranges of flexion, 19 is insignificant by adulthood, 20 and that a clear relationship exists between sacral and pelvic movement. 21 FOB pointer tilt and FOB hip angle measures mimic the radiographic definitions of these angles so comparisons are straightforward. RESULTS FOB Accuracy and Repeatability The accuracy and repeatability of the electromagnetic sensor was consistent with the manufacturer s documentation and results from published data. 8,12,13 The manufacturer lists accuracy to 2% of the measurement. Given our equipment setup, specifically the relationships between sensors and transmitter, accuracy was expected to be within 1.6mm. By using the FOB pointer, spatial digitization had error less than 1.25mm or approximately 1% over the range studied and the coefficient of variation (CV) from repeatedly measuring the same point was approximately 0.5%. The 6-df sensor had accuracy to.05mm and 0.5 with a CV of approximately 2%. Radiographic Reliability Intrarater reliability was calculated by using 9 radiographs and 2 investigators. Intrarater reliability was determined as excellent with ICC values exceeding.98 collectively and for the individual investigators. Interrater reliability was calculated from 30 radiographs (10 subjects in 3 postures). All radiographic measurements exhibited excellent interrater reliability: ICC of sacral angle,.98; ICC of pelvic tilt,.99; and ICC of hip angle,.97. The mean differences between rater measurements were: sacral angle, 2.7 ; pelvic tilt,.17 ; and hip angle,.82.

5 NONINVASIVE MEASUREMENT OF SEATED POSTURE, Sprigle 1601 Table 1: Correlation Between Radiographic Tilt and Noninvasive Tilt Sacral Belt Tilt FOB Pointer Tilt Validity of Noninvasive Measurements Validity of the noninvasive measurements (from the FOB) was determined by comparison with radiographic measurements by using the Pearson correlation coefficient (table 1), linear regression (table 2), and ANOVA. An ANOVA comparing radiographic pelvic tilt to the noninvasive measures of tilt found no significant difference (P.25). Absolute differences between radiographic and FOB pointer tilt were between 5 and 10, with the radiographic measures exceeding FOB measures by an average SD of Differences between radiographic tilt and sacral belt tilt cannot be determined because they measure related, yet different constructs. The 2 noninvasive measures of pelvic tilt were highly related. The Pearson correlation between pelvic tilt measures using the 6-df sensor (sacral belt tilt) and FOB pointer was.95 with a regression R 2 equal to.91 and a slope of.93. Correlation between radiographic and a noninvasive measure of hip angle (FOB hip angle) was not as high as the pelvic tilt correlation but did fall within the fair range (r.78). The mean difference between radiographic and noninvasive hip angle measures was An ANOVA comparing radiographic hip angle to the noninvasive measurement found no significant difference (P.28). DISCUSSION The results show excellent reliability of the invasive (radiographic) pelvic postural measure used to validate the noninvasive measures. Both inter- and intrarater reliability of the radiograph measures were considered excellent. This result represents an improvement from that of Gilliam et al 11 who expressed concern over the clarity of radiographs in measuring pelvic tilt. In that study, interrater reliability was.88, with a mean error in pelvic tilt of greater than 3. Our study showed pelvic tilt interrater reliability of.99 and a mean error less than 1. Our results reflect the development of a different procedure to obtain clear views of the iliac spines and sacrum and a standardized procedure for identifying the ASISs. Correct selection of voltage and amperage parameters based on pelvic width improved radiograph clarity, facilitating repeatable identification of bony landmarks. The validity testing showed the usefulness of noninvasive measures. The 2 noninvasive measures of pelvic tilt (sacral belt, FOB pointer) closely followed the radiographic measures as shown by their respective correlations. In addition, the regression results showed that most of the variance of the radiographic tilt measures could be explained by using noninvasive measures of tilt. Finally, ANOVA showed that the noninvasive measures were not distinct from the invasive measures of tilt. The mean errors in pelvic measurement showed that the noninvasive measures underestimated the pelvic tilt, but these differences were acceptable. The range of pelvic tilt within a subject was quite large, often exceeding 45 and the ranges of tilt in the 3 postures also varied greatly across subjects. No relationship was found between the magnitude of radiographic pelvic tilt and the difference with noninvasive measurement. In other words, measurement error was unrelated to amount of pelvic tilt. The validity of the noninvasive measures also represents an improvement over testing reported by Gilliam 11 and Burdett et al. 1 Gilliam, 11 as in our study, used the iliac spines as a measure of pelvic tilt, but found poor correlation between goniometric and radiologic measurements. Our results showed good correlation (r.89) and acceptable linear relationship (R 2.80) (table 2). Gilliam 11 did not mention the techniques used to identify the anterior and posterior iliac spines but did indicate that radiographic clarity was a problem. Burdett 1 used sacral angle as a measure of tilt, which was analogous to our measurements of sacral plateau and using the 6-df sensor mounted on the sacral belt. Burdett 1 found poor correlation and significant differences in radiographic and external measurements. Our results showed an excellent correlation (r.91) and insignificant differences (table 1). The instrumentation and techniques used in our study differ from those reported in the literature, which might account for these differences, but the results indicate that a noninvasive approach to measuring pelvic tilt is valid. Hip angle measures were not as tightly correlated, although the average differences between radiographic and noninvasive measures were less than 5. One of the challenges in determining hip angle by radiographs is the visualization of necessary structures. Pelvic structures ASIS and PSIS as well as the line of the femur are needed. The pelvic structures differ in density from the femur, thus indicating a need to compromise on the x-ray parameters. In this study, radiographic contrast was dictated by the pelvis to ensure the pelvic spines and sacral plateau were evident, sometimes at the expense of femur contrast. This tradeoff probably reduced hip angle repeatability and accuracy. The strong relationship between noninvasive measures of pelvic tilt and sacral angles was similar to the finding of Brumagne et al 3 and supports their combined use in static and dynamic postural assessments. The FOB pointer can be used to digitize the pelvis and thigh, thereby calculating pelvic tilt and hip angle in a static posture. Two 6-df sensors mounted to the sacral belt and thigh band can then be used to measure movement during activity. The 6-df sensors cannot be used to directly measure pelvic tilt because relationships between the sacrum and pelvis differ among people, although these structures move together. In other words, the sacral belt angle is related to but does not match pelvic tilt because the angle of the sacrum on the pelvis varies among people. Therefore, a static pelvic tilt measure by using an FOB pointer can be used to determine the starting pelvic tilt and the 6-df sensor mounted to a sacral belt can track pelvic angle displacements during activity (range of motion [ROM]). The use of sacral angle to measure the ranges of pelvic rotation have been shown in other studies 3,8 and corroborated by our data. The finding of a strong relationship between pelvic tilt (FOB tilt) and sacral angle (sacral belt tilt) extends the ability to measure pelvic ROM by assigning accurate pelvic tilt angles by an FOB pointer. Limitations and sources of error include instrumentation, human factors, and subject sources. FOB data reflect error from several sources, including palpation or landmarks, FOB pointer location, and 6-df sensor positioning and slippage; however, this error would have been reflected in the reliability and Table 2: Regression of Radiographic Tilt Versus Noninvasive Tilt Sacral Belt Tilt FOB Pointer Tilt R Slope.88.88

6 1602 NONINVASIVE MEASUREMENT OF SEATED POSTURE, Sprigle validity analysis. The wood and plastic test chair proved useful for the study and was designed to mimic a wheelchair environment, although it does not replicate it precisely. The use of a single chair to measure all subjects was deemed important, and, because repeated measures within subjects were used for analysis, little experimental error because of the chair was expected. Subjects for this study were limited to nondisabled men. The immobility of the sacrum on the pelvis found in this population might not transfer to other groups, although there is no indication that wheelchair users would have sacroiliac mobility. The requirement of a sacral belt to track pelvic tilt can prove problematic in certain sitting situations if the belt and transducer contact the backrest. This potential limitation needs to be judged on an individual basis because some chairs and wheelchairs have space posterior to the sacrum that would permit a person to sit without restriction, whereas others do not. CONCLUSIONS This project developed a reliable means to use radiographs to study pelvic position and a valid approach to measure noninvasively pelvic tilt and hip angle of seated subjects. The noninvasive approach can be used in research addressing postural issues of wheelchair users, office workers, and other groups who sit for extended durations. The measurement of the seated posture is problematic because of the lack of access to important structures. Electromagnetic sensors were shown to ably measure both static posture and dynamic movement. The use of electromagnetic sensors can be affected by environmental metal. This project used a wood and plastic test chair, but our experience with a parallel project and the study by Koo et al 12 indicate that judicious placement of the transmitter permitted measurement of posture as people sat in wheelchairs. Acknowledgments: taking the x-rays. We thank Maura Ryan for discussing and References 1. Burdett RG, Brown KE, Fall MP. Reliability and validity of four instruments for measuring lumbar spine and pelvic positions. Phys Ther 1986;66: Lee Y-H, Chen C-Y. Belt effects on lumbar sagittal angles. Clin Biomech 2000;15: Brumagne S, Lysens R, Spaepen A. Lumbosacral repositioning accuracy in standing posture: a combined electrogoniometric and videographic evaluation. Clin Biomech 1999;14: Youdas JW, Garrett TR, Egan KS, Therneau TM. Lumbar lordosis and pelvic inclination in adults with chronic low back pain. Phys Ther 2000;80: Alviso DJ, Gordon DT, Lintell JL. Intertester reliability for measuring pelvic tilt in standing. Phys Ther 1988;68: Gajodosik R, Simpson R, Smith R, Dontigny RL. Pelvic tilt intratester reliability angle. Arch Phys Med Rehabil 1976;12: Smidt GL, McQuade KJ, Wei S-H. Evaluation of the Metrocom and its use in quantifying skeletal landmark locations. J Orthop Sports Phys Ther 1992;16: Bull AM, McGregor AH. Measuring spinal motion in rowers: the use of an electromagnetic device. Clin Biomech 2000;15: Nwaobi OM, Hobson DA, Taylor SJ. Mechanical and anatomical hip flexion angles on seating children with cerebral palsy. Arch Phys Med Rehabil 1988;69: Mundale MO, Hislop HJ, Rabideau RJ, Kottke FJ. Evaluation of extension of the hip. Arch Phys Med Rehabil 1956;37: Gilliam J, Brunt D, MacMillan M, Kinard RE, Montgomery WJ. Relationship of the pelvic angle to the sacral angle: measurement of clinical reliability and validity. J Orthop Sports Phys Ther 1994;20: Koo TK, Mak AF, Lee YL. Posture effect on seating interface biomechanics: comparison between two seating cushions. Arch Phys Med Rehabil 1996;77: Milne AD, Chess DG, Johnson JA, King GJ. Accuracy of an electromagnetic tracking device: a study of the optimal operating range and metal interference; J Biomech 1996;29: Bull AM, Berkshire FH, Amis AA. Accuracy of an electromagnetic measurement device and application to the measurement and description of knee joint motion. Proc Inst Mech Eng [H] 1998; 212: ICRP Publication 60, 1990 Recommendations of the International Commission on Radiological Protection. Ann ICRP 1991;21:No National Council on Radiation Protection and Measurement. Limitation of exposure to lowering radiation. Bethesda (MD): NCRP; NCRP Report No Committee on Biological Effects of Ionizing Radiations, Board on Radiation Effects Research, Commission on Life Sciences, National Research Council. Health effects of exposure to low levels of ionizing radiation: BEIR V. Washington (DC): Natl Acad Pr; Currier DP. Elements of research in physical therapy. Baltimore: Williams & Wilkins; Calliet R. Low back pain syndrome. Philadelphia: FA Davis; Walker JM. Age-related differences in the human sacro-iliac joint: a histological study; implications for therapy. J Orthop Sports Phys Ther 1986;7: Farfan HF, Huberdeau RM, DuBow HI. Lumbar intervertebral disc degeneration: the influence of geometrical features on the pattern of disc degeneration a post mortem study. J Bone Joint Surg Am 1972;54: Suppliers a. Ascension Technologies Corp, PO Box 527, Burlington, VT b. Okamoto Manufacturing Co, Sagisu, Fukushima-Ku Osaka , Japan. c. Supertech Inc, PO Box 186, Elkhart, IN