Influence of feedback schedule in motor performance and learning of a lumbar multifidus muscle task using rehabilitative ultrasound imaging: a randomized clinical trial.

Low back pain (LBP) is a major health problem affecting millions and is a leading reason why people seek medical interventions and miss work each year. (1-5) Most initial, acute cases of LBP resolve within 2 to 4 weeks. (1,2,6,7) Recurrence rates of LBP within the first year after the initial, acute episode range from 50% to 86%. (1,6,7) One theory for a mechanism contributing to the recurrence of LBP states that pain may resolve, but local neuromuscular control of lumbar segmental motion remains impaired. (8-11)

Some researchers (9,12-15) classify the multifidus muscle as a primary spine stabilizer because its direct anatomical attachments to the vertebrae suggest that it can directly control (or modulate) intersegmental movements. It has been reported that dysfunction of the multifidus muscle begins within a few weeks of the initial, acute episode. (8,16,17) At that point, the multifidus muscle shows focal impairments in size, timing, amplitude, and co-activation with the abdominal muscles. (8-10,17-19) Impairment of these local factors of neuromuscular control may contribute to a state of potential instability in the lumbar segment, thereby increasing susceptibility to reinjury. (8-10,17,18,20) If impaired multifidus muscle function after an initial injury is a factor leading to recurrent LBP, then improving neuromuscular control at the intersegmental level should be an effective intervention for preventing LBP recurrence.

One intervention to enhance lumbar stabilization in people with LBP incorporates recruitment of the multifidus muscle via the specific "segmental stabilization exercise" (SSE) approach. (21,22) Emerging evidence suggests that SSE training improves short-term and long-term outcomes in patients with acute and chronic LBP. (8-10) However, researchers and clinicians indicate that patients have difficulty learning to recruit the multifidus using direct extrinsic feedback methods, such as palpation and verbal feedback provided by the therapist or a blood pressure cuff placed under the muscle. It may require as long as 10 weeks of supervised in-clinic training plus home programs to achieve proficiency. (8,9,23,24) We speculate that there are at least 2 reasons for this extended time required for learning: (1) an ineffective implementation of feedback for motor learning (25-30) and (2) the low conscious sensory perception of postural control muscles of the spine and trunk. (31)

Research to date has shown the effectiveness of augmenting clinical instruction with visual feedback via real-time rehabilitative ultrasound imaging (RUSI) for enhancing performance of isolated contraction of the transversus abdominis and lumbar multifidus muscles. (32-34) In clinical research studies, Richardson and colleagues (8,22) and Van and colleagues (32) used real-time RUSI as a form of feedback to enhance the motor learning process for isolated recruitment of the multifidus muscle. Although these researchers demonstrated effective rehabilitation outcomes in patients with LBP and subjects who were healthy trained with this method, they did not specifically examine the motor learning process itself. Based on a review of literature, it is unclear whether feedback schedules other than constant feedback in training the multifidus with SSE have been implemented or studied for their effectiveness. (8-10,22,24) Using constant feedback during training could be a reason for the extensive length of time required for learning to recruit the multifidus muscles.

Results from previous research in motor learning indicate that variability in the frequency or schedule of feedback during practice is superior to constant feedback when measuring individuals' retention and ability to transfer skills to novel situations. (26,28,29,35,36) To date, there are no studies exploring variability of feedback schedules using real-time RUSI in learning to recruit the multifidus muscle in isolation. This raises the question of whether using a variable feedback schedule would result in faster skill acquisition and more effective long-term retention of multifidus muscle recruitment. Therefore, the purpose of this study was to investigate the motor learning processes for training patients to isolate the lumbar multifidus muscle under varying feedback schedules. Specifically, we compared the effectiveness of constant feedback or variable feedback delivered with RUSI for skill acquisition and retention of multifidus recruitment. We hypothesized that a variable feedback schedule would result in less success in the subjects' performance of multifidus muscle recruitment during the training phase but would result in greater success in the retention phase, inferring that this practice schedule was better for learning the motor skill. By using a group of subjects who were healthy, this study is a first step in investigating the feasibility and application of motor learning principles in a postural muscle group that is difficult to preferentially isolate during exercise.

Method

Design Overview

This research study was a randomized clinical trial designed to test the hypothesis that a variable feedback schedule would result in superior learning compared with a constant feedback schedule when using RUSI for training the lumbar multifidus muscle. Figure 1 shows the research design to compare the effectiveness of constant to variable feedback provided via RUSI for learning the specific motor skill of isolated isometric lumbar multifidus recruitment.

Participants

Approval from the Ohio State University Human Subjects Institutional Review Board (IRB) was received prior to initiation of the study. The first 30 individuals who gave written consent to participate in accordance with the procedures accepted by the IRB, passed a standardized low back and lower-extremity examination, and met the inclusion and exclusion criteria were accepted and randomly assigned to constant feedback (CON) and variable feedback (VAR) treatment groups. Inclusion criteria included adults who were healthy, were between the ages of 18 and 49 years, had a body mass index (BMI) [less than or equal to] 30, had 5/5 muscle strength (force-generating capacity) in bilateral lower extremities and trunk muscles, had 2+ reflexes in the quadriceps femoris and plantar-flexor muscles, were able to accurately detect touch, and had no visible spinal deformities (eg, scoliosis, hyperkyphosis, or spina bifida). (37) The exclusion criteria included a history of back pain within the past year significant enough to receive medical care, abdominal or back surgery that altered the lumbar muscles, or prior training of multifidus muscle recruitment.

Subjects also completed the Habitual Physical Activity Questionnaire (HPAQ), (38,39) modified for Americans by including the word "exercise" with "sport" when referring to physical activity. (40) The modified HPAQ was used to characterize the level of habitual physical activity among the subjects. This questionnaire is short and easy to complete, and its validity and reliability have been studied in various populations. (38-41)

Randomization and Interventions

Envelopes containing the randomized group assignment and feedback schedule for each subject were created a priori to ensure an equal number of subjects per treatment group. Randomization, with stratification by sex, was performed to control for the possibility that differences between males and females would affect learning the motor skill. After all baseline data were collected for a subject, the investigator (WJH) opened the envelope, assigned the subject to a treatment group (CON or VAR), and trained the subject to perform isolated multifidus muscle recruitment through the provision of RUSI feedback in accordance with the predetermined feedback schedule.

[FIGURE 1 OMITTED]

Ultrasound Imaging of the Multifidus Muscle at the Level of S1

Baseline measurements of the multifidus muscle cross-sectional area (CSA) were obtained with the RUSI unit (Model 180/L38 unit) * in accordance with the protocol established by Hides et al (42) and utilized in the principal investigator's (DGH) laboratory. (43) Baseline CSA images of the left and right multifidus muscles were captured and measured to establish baseline comparability and examine for evidence of significant side-to-side asymmetry, which may indicate a history of low back dysfunction. (8,44)

For the training sessions, the subjects were positioned prone on the treatment table with the hips in the neutral position, in accordance with the procedures recommended by Richardson et al (45) for exercising the multifidus muscle. Real-time RUSI of the multifidus muscle at the level of S1 was recorded, transferred to the video recording system, and projected on the television monitor to provide visual feedback. For each subject, the right side was arbitrarily selected for feedback because we did not expect side-to-side differences in the ability to recruit the multifidus muscle in subjects who were healthy. The activation of the muscle was evident as a slight deformation or "bulging" of the muscle belly when viewed as a cross-section. With the live-action video feedback, this bulging was readily visible to the investigator and the subject. Although other researchers (32,46) have used the parasagittal view to detect thickening of the muscle during activation, we found that the transverse view was easier for us to detect more subtle muscle activity.

Training Based on Group Assignment

Subjects attended 15-minute exercise training sessions in the laboratory, twice a week, for a total of 8 training sessions (Fig.1). To orient the subjects to the muscle to be trained, RUSI pictures of the multifidus muscle as well as anatomical illustrations were reviewed with all subjects prior to the start of the initial training session. Subjects were asked to refrain from practice between training sessions. To ensure consistent instruction between groups and training sessions, a written script was read to each of the subjects at the beginning of the study and at the beginning of each training session. This script included the instructions to recruit the multifidus muscle without extraneous movements and to hold each contraction for 3 seconds. It also informed the subjects that the training session would consist of 12 repetitions of the exercise and that a successful performance outcome was visualization of muscle movement on the monitor.

Performance success was defined as isolated isometric recruitment of the first sacral level (S1) multifidus muscle without substitution of extraneous movements such as Valsalva, pelvic tilt, arching the back, lifting the upper trunk, or lifting the lower extremity. (10,45,47) Isometric recruitment of the multifidus muscle was determined by the presence or absence of tissue movement (muscle thickening) of the multifidus muscle as visualized on the RUSI screen.

Subjects in both groups were given primarily knowledge of results (successful or unsuccessful multifidus muscle activation of each exercise repetition) to avoid overwhelming them with extraneous information.

This dichotomous outcome also was chosen to allow subjects to develop their own problem solving during the motor task and to provide a more realistic form of feedback that could be replicated in a busy clinic.

The real-time RUSI for subjects in both treatment groups was recorded on videotape for each of the 12 repetitions. During each repetition of muscle activation, subjects assigned to the CON group received visual feedback of the real-time RUSI of successful or unsuccessful multifidus muscle activation on the monitor, but were not given verbal feedback. A subject assigned to the VAR group received delayed feedback after performing a number of repetitions of the exercise, based on a predetermined schedule. The researcher (WJH) would play the video of all the RUSI and provide summary verbal knowledge of results of the performance up to that point. The predetermined schedules were designed to provide 4 variable intervals of summary feedback during each trial of 12 repetitions, with a minimum of 1 repetition and a maximum of 3 repetitions between the feedback intervals. The variable feedback schedule was randomized between subjects and sessions a priori to help control for an order effect.

Retention Test

After the training phase of the study, each subject returned to the laboratory on 2 separate occasions. The first session was 1 week after completion of training to assess short-term retention (R1). The second session was at least 4 weeks after completion of the training to assess long-term retention (R2). The procedures established for the exercise training sessions were repeated, except that no augmented feedback (visual or verbal) was provided during the exercises. Each subject performed 2 sets of 12 repetitions of isometric recruitment of their right multifidus muscle, with a short rest period between sets. The collection of 2 sets of data allowed for examination of both within session (set 1 versus set 2) and between session (training versus retention) learning effects. In the retention phase of the study, it was decided a priori that, if there was no difference in the performance success between the 2 sets of 12 repetitions within each retention test session, then the data would be collapsed into 1 score.

Testing for Bias in the Ratings

The investigator (WJH) who trained the subjects also determined each subject's success or failure of the exercise attempts during the training and retention testing sessions. These procedures were similar to a published study by Henry and Westervelt. (33) Making a real-time judgment from RUSI of success or failure allowed the investigator to rule out false positive ratings due to movement artifact of the subject or the ultrasound array. Given the possibility of bias in the ratings, we assessed the reliability of the investigator's decisions on a separate occasion after the completion of all data collection. A research assistant randomly selected one training session from the video recordings of each subject. The investigator rated video-recorded activations from subjects while blinded to the subjects' group, training session, and repetition number.

Data Analysis

The Statistical Package for the Social Sciences (SPSS 14.0), ([dagger]) R 2.4.1 for Windows, ([double dagger]) and Microsoft Office Excel 2003 ([section]) were used for all data analyses. A criterion level of P<.05 was considered statistically significant. Subject characteristics between groups were compared using t tests and chi-square analyses. The number of observed successful isometric activations out of the 12 repetitions for each session was calculated and recorded as a percentage. Because a dichotomous variable (success/failure) was used to measure performance, we applied nonparametric statistics to analyze training performance and retention. We elected not to pretest the subjects to assess performance before training. We were concerned that subjects might learn how to recruit the multifidus during a pretest because pressure on the skin over the multifidus muscle belly is a technique utilized to facilitate activation. (45) In the absence of a pretest, performance on the first repetition of the first training session was compared using the Fisher exact test to test for differences between groups at the outset of training.

To test the hypotheses that the CON group would demonstrate greater performance success than the VAR group during the training phase of isolated isometric multifidus muscle recruitment (8 sessions) and that both groups would show improved performance success over the 8 training sessions, the Wilcoxon signed rank test was used to test for differences between the first and eighth training sessions for both the CON and VAR groups. To test the primary hypothesis that the VAR group would demonstrate greater performance success than the CON group at retention testing, a nonparametric Wilcoxon score test (48) was used. This was required because of non-normality of the data. Because the interval between retention tests (R1 and R2) varied from subject to subject, the change from R1 to R2 was defined as the difference between the success rates at the 2 evaluation times divided by the number of weeks between them.

To examine the consistency of the investigator's ratings of success or failure under blinded video-recorded and nonblinded real-time conditions, the pairs of ratings for the selected trials were compared using the Cohen kappa statistic.

Results

Characteristics of Subjects

Thirty subjects were recruited and randomized into the 2 treatment groups (Fig. 1). Two subjects dropped out for reasons unrelated to the treatment. One subject in the CON group completed 5 training sessions but had to drop out for unrelated health problems. One subject in the VAR group dropped out before beginning the training due to schedule conflicts. These 2 dropouts were not included in the data analysis. The 2 groups were similar (P> .05) in sex, age, BMI, modified HPAQ scores, and CSA measurements at the outset of the study (Tab. 1). The mean CSA difference between the left and right multifidus at the level of S1 was 0.23 [cm.sup.2] (SD=0.8, 95% [confidence interval] CI=0.08, 0.53). Previous work in our laboratory established that these differences were less than 2 standard errors of the measurement, (43) and the muscles could be considered symmetrical.

Training Phase

Fourteen subjects per group completed all 8 training sessions. The 2 groups showed similar (Fisher exact test, P=.26) performance on the first repetition at the outset of the study in their ability to recruit the multifidus muscle, with 11/15 of the CON group and 7/15 of the VAR group being successful. Figure 2 shows the average performance success of the groups at each session over time. The CON group showed greater average performance success early in training (mean=80%) that was maintained at the end of training (mean=84%), with no difference in success rate between sessions 1 and 8 (Wilcoxon signed rank test, P=.19, 95% CI= -9%, 42%). The VAR group showed more gradual improvement across training sessions with an increase in success rate (Wilcoxon signed rank, P=.002, 95% CI=17%, 59%) between sessions 1 and 8.

[FIGURE 2 OMITTED]

Retention Phase

Thirteen subjects from the CON group and 14 subjects from the VAR group completed R1. One subject from the CON group was unavailable for the R1 due to travel. The average time between completion of the training (session 8) and R1 was 7.3 days (SD=1.2). Both groups sustained their level of performance success at completion of the eighth training session when tested again at 1 week for the short-term retention (R1) testing (CON group: Wilcoxon signed rank test, P=.79; VAR group: Wilcoxon signed rank test, P=.36).

Eleven subjects from the CON group and 12 subjects from the VAR group completed R2. Four subjects could not be contacted to return for final testing. The overall average length of time from completion of the training phase and R2 was 13.2 weeks (SD=6.6). There was no difference (P>.05) between the 2 groups in the timing of R2 following training, with the average time of 10.6 weeks (SD=5.3) for the CON group and 15.5 weeks (SD=7.0) for the VAR group.

A last observation carry-forward approach was used to impute a single missing value for a subject who missed R1 and returned later for R2. This approach was considered reasonable as the interval between the final session and R1 was relatively short (1 week). The higher score by the VAR group for R2 indicated better retention of the motor skill of S 1 multifidus recruitment (Wilcoxon score test, P=.04, 95% CI=0.00001, 4.41).

Agreement Between Blinded and Onblinded Conditions

The Cohen kappa statistic testing for percentage of agreement on the success or failure scores for all subjects (N=28) at each training session for the initial (unblinded) ratings compared with blinded ratings showed moderate agreement with kappa=.64 (P=.04) and a 95% CI=0.54 to 0.74 (Tab. 2). (49,50) Because the CON group was highly successful during training, the sample of images randomly selected for testing rater bias was unbalanced and possibly lowered the kappa value for agreement.

Discussion

The most important finding of this study is that the variable schedule of summary RUSI feedback was superior to constant RUSI feedback alone for retention of multifidus muscle recruitment at the level of S1. We demonstrated that a common principle of motor learning in subjects who are healthy, the superiority of variable feedback, resulted in greater long-term retention of the motor skill of specific multifidus muscle recruitment. (25,26,28,29,35,36,51) Most of the motor learning studies focus on movements involving manipulation of objects with the hands, such as throwing a ball, or whole body motor tasks, such as those seen in sports, (52) Studies that investigate the motor learning strategies for local neuromuscular control of lumbar segmental motion via the lumbar multifidus muscle are lacking. The pathways for neural control of movement may be somewhat different for control of the extremities versus control of the trunk and the spine, based on the differences between the lateral systems (corticospinal and rubrospinal tracts) and the ventromedial systems (reticulospinal and vestibulospinal tracts). It is important, therefore, to determine whether principles of motor learning developed through research in the areas of extremity movements also hold true for learning isolated control of trunk muscles such as the multifidus muscle.

As expected, the CON group demonstrated greater performance success during training compared with the VAR group's gradual improvement across training sessions. Short-term retention (R1) testing revealed that both groups sustained their level of performance success measured at completion of the training phase. Despite this, there were between-group differences at the long-term retention (R2) test, largely because of the VAR group's greater ability to retain the motor skill. These findings are consistent with motor learning principles. Although constant feedback conditions may appear to be the more appropriate practice strategy for learning because of the high performance success during training, the variable feedback condition enhances the ability to refine motor strategies, resulting in greater long-term retention of the newly learned motor skill. (25-30)

Clinical Significance

Helping individuals to achieve success in lumbar stabilization more effectively and efficiently could potentially reduce the risk of recurrent and chronic LBP. Because RUSI is gaining in popularity as a means to provide biofeedback during exercise, it is important to identify procedures that enhance learning. (33,53-55) The use of variable RUSI feedback during multifidus muscle SSE may be a powerful feedback mechanism for physical therapists to utilize clinically during the acquisition of the motor skill. The subjects in our study were able to learn to recruit the lumbar multifidus muscle at a success rate of about 90% to 95% in as little as 4 weeks with variable RUSI feedback. More importantly, subjects had better long-term retention of this motor skill with variable RUSI feedback than when provided with constant feedback.

Given the high recurrence rate of LBP, long-term retention of multifidus muscle activation may be more important than how fast the isometric exercise is learned. The knowledge of performance provided through this variable augmented feedback appears to help individuals sell-analyze performance strategies in the absence of conscious sensory perception and refine the motor skill more successfully and in less time than previously reported. (9,10,24,56,57) However, without a control group that received only tactile feedback or no augmented feedback, this study cannot generalize the findings to clinical situations where tactile feedback is used in lieu of real-time RUSI. Furthermore, it remains to be seen whether this training approach results in transfer of the learned skill to other postures and functions and whether this muscle activation correlates with meaningful stabilization of spinal motion segments.

Limitations

The small sample size of this study and the high success rate of the CON group during training possibly account for the findings of only moderate agreement between blinded and unblinded ratings of success or failure to recruit the multifidus muscle. The quality of the video recordings of the RUSI images also may have affected the blinded ratings. In order to reduce the potential for rater bias, future studies should use digital recording of the RUSI images and have an expert in RUSI imaging as a second blinded evaluator to independently rate performance success in real time. We also recognize that measuring thickness changes captured on RUSI would have allowed more subtle gradations of success or failure ratings of multifidus muscle recruitment; however, a nominal judgment of success or failure was selected in an attempt to use methods comparable to common clinical practice.

The variable feedback schedule of this study provided subjects with delayed verbal and visual knowledge of results of all repetitions completed up to that point. To further strengthen the generalizability that variable feedback is superior to constant feedback when learning to recruit the multifidus muscle, future studies should investigate whether feedback provided only at randomly selected repetitions without the summary of results enhances learning.

Conclusion

Retention of the motor task of isometric recruitment of the multifidus muscle was best achieved with variable feedback during practice rather than constant feedback in this small sample of subjects who were healthy. The plausible clinical interpretation of these findings suggests that RUSI was a useful biofeedback mechanism during training of multifidus muscle recruitment, and subjects acquired the skill in much less time than expected based on previous pilot work. Future research should determine whether these findings hold true in training and testing individuals with LBP under similar conditions, or if pain, atrophy, or other problems create confounding factors.

All authors provided concept/idea/research design, writing, and data analysis. Dr Heiss and Ms Herbert provided data collection, project management, fund procurement, and subjects. Dr Heiss provided facilities/ equipment. The authors thank John A Buford PT, PhD, for intellectual contributions and writing assistance, Larry Sachs, PhD, for statistical support, and Anand Navalgund for technical support. They also thank all of the individuals who volunteered their time to participate in this research study.

Approval from The Ohio State University Human Subjects Institutional Review Board was received prior to initiation of the study.

An abstract of this research was presented at the Combined Sections Meeting of the American Physical Therapy Association; February 1-5, 2006; San Diego, Calif; and at the Ohio Physical Therapy Association Annual Conference; October 21-23, 2005; Columbus, Ohio.

This study was partially funded by the Rosita Schiller Scholarship from The Ohio State University, School of Allied Medical Professions, Columbus, Ohio.

This article was received October 11, 2006, and was accepted October 2, 2007.

DOI: 10.2522/ptj.20060308

References

(1) Berquist-Ullman M, Larsson U. Acute low back pain in industry: a controlled prospective study with special reference to therapy and confounding factors. Acta Orthop Scand. 1970;170(suppl):1-117.

(2) Von Korff M, Dworkin SF, Le Resche L, Kruger A. An epidemiologic comparison of pain complaints. Pain. 1988;32: 173-183.

(3) Flicker PL, Fleckenstein JL, Ferry K, et al. Lumbar muscle usage in chronic low back pain magnetic resonance image evaluation. Spine. 1993;18:582-586.

(4) Hart LG, Deyo RA, Cherkin DC. Physician office visits for low back pain: frequency, clinical evaluation, and treatment patterns from a US national survey. Spine. 1995;20:11-19.

(5) Deyo RA, Mriza SK, Martin BI. Back pain prevalence and visit rates: estimates from US national surveys, 2002. Spine. 2006;31: 2724 -2727.

(6) Troup JD, Martin JW, Lloyd DC. Back pain in industry: a prospective survey. Spine. 1981;6:61-69.

(7) Von Kofff M, Deyo RA, Cherkin D, Barlow W. Back pain in primary care: outcomes at 1 year. Spine. 1993;18:855-862.

(8) Hides JA, Richardson CA, Jull GA. Multifidus muscle recovery is not automatic after resolution of acute, first-episode low back pain. Spine. 1996;21:2763-2769.

(9) O'Sullivan PB, Twomey LT, Allison GT. Evaluation of specific stabilizing exercise in the treatment of chronic low back pain with radiologic diagnosis of spondylolysis or spondylolisthesis. Spine. 1997;22: 2959-2967.

(10) Hides JA, Jull GA, Richardson CA. Long-term effects of specific stabilizing exercises for first-episode low back pain. Spine. 2001;26:E243-E248.

(11) Hodges PW, Moseley GL. Pain and motor control of the lumbopelvic region: effect and possible mechanisms. J Electromyogr Kinesiol. 2003;13:361-370.

(12) Panjabi M, Abumi K, Duranceau J, Oxland T. Spinal stability and intersegmental muscle forces: a biomechanical model. Spine. 1989;14:194-200.

(13) Bergmark A. Stability of the lumbar spine: a study in mechanical engineering. Acta Orthop Scand Suppl. 1989;230:1-54.

(14) Bogduk N, Twomey LT. Clinical Anatomy of the Lumbar Spine. 2nd ed. Edinburgh, United Kingdom: Churchill Livingstone; 1991.

(15) Wilke HJ, Wolf S, Claes LE, et al. Stability increase of the lumbar spine with different muscle groups: a biomechanical in vitro study. Spine. 1995;20:192-198.

(16) Rantanen J, Hurme M, Falck B, et al. The lumbar multifidus muscle five years after surgery for a lumbar intervertebral disc herniation. Spine. 1993;18:568-574.

(17) Hides JA, Stokes MJ, Saide M, et al. Evidence of lumbar multifidus muscle wasting ipsilateral to symptoms in patients with acute/subacute low back pain. Spine. 1994;19:165-172.

(18) Sihvonen T, Herno A, Paljarvi L, et al. Local denervation atrophy of paraspinal muscles in postoperative failed back syndrome. Spine. 1993;18:575-581.

(19) Hodges PW, Richardson CA. Inefficient muscular stabilization of the lumbar spine associated with low back pain: a motor control evaluation of transversus abdominis. Spine. 1996;21:2640-2650.

(20) Cholewicki J, McGill SM. Mechanical stability of the in vivo lumbar spine: implications for injury and chronic low back pain. Clin Biomech (Bristol, Avon). 1996; 11:1-15.

(21) Richardson C, Jull G. An historical perspective on the development of clinical techniques to evaluate and treat the active stabilising system of the lumbar spine. Aust J Physiother. 1995;1:5-13.

(22) Richardson C, Hodges PW, Hides JA. Therapeutic Exercise for Lumbopelvic Stabilization: A Motor Control Approach for the Treatment and Prevention of Low Back Pain. 2nd ed. Edinburgh, United Kingdom: Churchill Livingstone; 2004.

(23) Lee TD, Swanson LR, Hall AL. What is repeated in a repetition? Effects of practice conditions on motor skill acquisition. Phys Ther. 1991;71:150-156.

(24) O'Sullivan P, Twomey L, Allison G. Altered abdominal muscle recruitment in patients with chronic back pain following a specific exercise intervention. J Orthop Sports Phys Ther. 1998;27:114-124.

(25) Gentile A. A working model of skill acquisition with application to teaching. Quest. 1972;17:3-23.

(26) Schmidt R. A schema theory of discrete motor skill learning. Psychol Rev. 1975;82: 225-260.

(27) Drowatzky JN. Motor Learning: Principles and Practices. Minneapolis, Minn: Burgess Publishing Co; 1975.

(28) Winstein C, Schmidt R. Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psych. 1990;16: 677-691.

(29) Wuff G, Schmidt RA, Deubel H. Reduced feedback frequency enhances generalized motor program learning but not parameterization learning. J Exp Psychol Learn Mem Cogn. 1993;19:1134-1150.

(30) Shea CH, Wuff G. Enhancing motor learning through external-focus instructions and feedback. Hum Mov Sci. 1999;18: 553-571.

(31) Kandel ER, Schwartz JH, Jessell TM, eds. Principles of Neural Science. 4th ed. New York, NY: McGraw-Hill, Health Professions Division; 2000.

(32) Van K, Hides JA, Richardson CA. The use of real-time ultrasound imaging for biofeedback of lumbar multifidus muscle contraction in healthy subjects. J Orthop Sports Phys Ther. 2006;36:920-925.

(33) Henry SM, Westerveh KC. The use of real-time ultrasound feedback in teaching abdominal hollowing exercises to healthy subjects. J Orthop Sports Phys Ther. 2005;35:338-345.

(34) Worth SG, Henry SM, Bunn JY. Real-time ultrasound feedback and abdominal hollowing exercises for people with low back pain. N Z J Physiol. 2007;35(1):4-11.

(35) Pigott R, Shapiro D. Motor schema: the structure of the variability session. Res Q Exerc Sport. 1984;55(1):41-45.

(36) Turnbull S, Dickinson J. Maximizing variability of practice: a test of schema theory and contextual interference theory. J Hum Mov Stud. 1986;12:201-213.

(37) Magee DJ. Orthopedic Physical Assessment. 4th ed. Philadelphia, Pa: Saunders; 2002.

(38) Baecke J, Burema J, Frijters J. Baecke questionnaire of habitual physical activity. J Am Coll Sports Med. 1982;29(6 suppl): S15-S17.

(39) Jacobs DR Jr, Ainsworth BE, Hartman TJ, Leon AS. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc. 1993;25:81-91.

(40) Pressler JM, Heiss DG, Buford JA. Validation of the modified Habitual Physical Activity Questionnaire in young, healthy females aged 17 to 30 years [abstract]. Poster presentation at PT 2004: Annual Conference and Exposition of the American Physical Therapy Association; June 30-July 3, 2004; Chicago, Ill.

(41) Pols M, Peeters P, Bueno-De-Mesquita HB, et al. Validity and repeatability of a modified Baecke questionnaire on physical activity. Int J Epidemiol. 1995;24:381-388.

(42) Hides J, Richardson C, Jull G, Davies S. Ultrasound imaging in rehabilitation. Aust J Physiother. 1995;41:187-193.

(43) Pressler JF, Heiss DG, Buford JA, Chidley JV. Between-day repeatability and symmetry of multifidus cross-sectional area measured using ultrasound imaging. J Orthop Sports Phys Ther. 2006;36:10-18.

(44) Hodges PW, Holm AK, Hansson T, Holm S. Rapid atrophy of the lumbar multifidus follows experimental disc or nerve root injury. Spine. 2006;31:2926-2933.

(45) Richardson C, Jull G, Hodges PW, Hides JA. Therapeutic Exercise for Spinal Segmental Stabilization in Low Back Pain: Scientific Basis and Clinical Approach. Edinburgh, United Kingdom: Churchill Livingstone; 1999.

(46) Kiesel KB, Uhl TL, Underwood FB, et al. Measurement of lumbar multifidus muscle contraction with rehabilitative ultrasound imaging. Man Ther. 2006;12:161-166.

(47) Richardson C, Jull G, Toppcnberg R, Comerford M. Techniques for active lumbar stabilisation for spinal protection: a pilot study. Aust J Physiother. 1995; (monograph no. 1):27-34.

(48) Hollander M, Wolfe DA. Nonparametric Statistical Methods. 2nd ed. New York, NY: John Wiley & Sons Inc; 1999.

(49) Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20:37-46.

(50) Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159-174.

(51) Fitts PM, Posner MI. Human Performance. Belmont, Calif: Brooks & Cole; 1967.

(52) Schmidt RA, Wrisberg CA. Motor Learning and Performance: A Problem-Based Learning Approach. 3rd ed. Champaign, Ill: Human Kinetics; 2004.

(53) Bo K, Sherburn M, Allen T. Transabdomihal ultrasound measurement of pelvic floor muscle activity when activated directly or via a transversus abdominis muscle contraction. Neurourol Urodyn. 2003;22:582-588.

(54) Whitaker J. Abdominal ultrasound imaging of pelvic floor muscle function in individuals with low back pain. J Man Manip Ther. 2004;12:44-49.

(55) Teyhen DS, Miltenberger CE, Deiters HM, et al. The use of ultrasound imaging of the abdominal drawing-in maneuver in subjects with low back pain. J Orthop Sports Phys Ther. 2005;35:346-355.

(56) Del Rey P. The effects of video-taped feedback on form, accuracy, and latency in an open and closed environment. J Mot Behav. 1971;3:281-287.

(57) Hodges NJ, Franks RC. The role of video in facilitating perception and action of a novel coordination movement. J Mot Behav. 2003;35:247-260.

* SonoSite Inc, 21919 30th Drive SE, Bothell, WA 98021.

([dagger]) sPss Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606-6307.

([double dagger]) The R Foundation for Statistical Computing, c/o Institut fur Statistik und Wahrscheinlichkeitstheorie, Technische Universitat Wien, Wiedner Hauptstrasse 8-10/1071, 1040 Vienna, Austria.

([section]) Microsoft Corp, One Microsoft Way, Redmond, WA 98052-6399.

WJ Herbert, PT, MS, is PhD Candidate, Health and Rehabilitation Sciences Graduate Program, and Graduate Research Associate, Division of Physical Therapy, The Ohio State University, Columbus, Ohio.

DG Heiss, PT, PhD, DPT, OCS, is Associate Professor and Director, Division of Physical Therapy, School of Allied Medical Professions, College of Medicine, and Participating Faculty, Biomedical Engineering Graduate Program, The Ohio State University, 453 W 10th Ave, Columbus, OH 43210-1234 (USA). Address all correspondence to: Deborah.Heiss@ osumc.edu.

DM Basso, PT, EdD, is Associate Professor, Division of Physical Therapy, Assistant Director, School of Allied Medical Professions, and Associate Professor, Department of Neuroscience, College of Medicine, The Ohio State University.

[Herbert WJ, Heiss DG, Basso DM. influence of feedback schedule in motor performance and learning of a lumbar multifidus muscle task using rehabilitative ultrasound imaging: a randomized clinical trial. Phys Ther. 2008;88:261-269.]

Table 1.
Subject Characteristics at the Beginning of the Study (a)

                                           CON

Male                                       5
Female                                     9
Age (y)                                    26.7 (7.8) [18-24]
BMI (kg/[m.sup.2])                         24.74 (3.06) [20.00-26.90]
MHPAQ (total)                               9.1 (1.1) [7.6-10.8]
MHPAQ (sports)                              3.2 (0.4) [2.8-3.8]
Left multifidus muscle CSA ([cm.sup.2])     5.3 (1.3) [3.6-7.4]
Right multifidus muscle CSA ([cm.sup.2])    5.2 (1.1) [3.6-6.9]

                                           VAR

Male                                       4
Female                                     10
Age (y)                                    29.4 (8.3) [19-47]
BMI (kg/[m.sup.2])                         23.75 (1.96) [20.90-30.00]
MHPAQ (total)                               8.4 (1.4) [6.2-11.2]
MHPAQ (sports)                              3.0 (0.6) [1.8-3.8]
Left multifidus muscle CSA ([cm.sup.2])     5.7 (0.96) [4.3-7.8]
Right multifidus muscle CSA ([cm.sup.2])    5.4 (1.1) [3.9-7.6]

                                           Tests (2-tailed)      P

Male                                       [chi square] = 0.16   1.00
Female
Age (y)                                    t = 0.87               .39
BMI (kg/[m.sup.2])                         t = 1.02               .32
MHPAQ (total)                              t = 1.39               .18
MHPAQ (sports)                             t = 1.27               .22
Left multifidus muscle CSA ([cm.sup.2])    t = 0.99               .33
Right multifidus muscle CSA ([cm.sup.2])   t = 0.57               .58

(a) Measurements reported as mean, SD (in parentheses), and range
(in brackets). CON = constant feedback group, VAR = variable
feedback group, BMI = body mass index, MHPAQ = modified Habitual
Physical Activity Questionnaire, CSA = cross-sectional area.

Table 2.
Actual Agreement/Disagreement Between Initial (Unblinded)
and Blinded Ratings

                            Blinded Rating

Initial Rating     Success     No Success     Total

Success            239         20             259
No success          22         55              77
Total              261         75             336