Intertester reliability of a modified version of McKenzie's lateral shift assessments obtained on patients with low back pain.

Background and Purpose. McKenzie described a two-step process for assessing patients with low back pain for a lateral shift. The purpose of this study was to determine whether reliable judgments about lateral shifts could be obtained. Subjects. Forty-nine patients with low back pain were each examined separately by two randomly paired physical therapists. Methods. Assessments of the presence and direction of lateral shifts (step 1) were obtained by use of a simple instrument. The relevance of the lateral shifts to the patients' pain complaints (step 1) also was assessed by use of the side-glide test sequence. Results. Generalized kappa coefficients were calculated to determine reliability. The kappa value for the two-step process of lateral shift assessment was .16. The percentage of agreement was 47%. Conclusion and Discussion. Each step in this two-step process was examined separately for possible sources of error. The kappa value for determinations of the presence and direction of lateral shifts was .00, indicating very poor reliability. The kappa value for the determination of the presence of a positive side-glide test sequence was .74, indicating high reliability. The role of lateral shift assessment in the McKenzie system should be reconsidered, given the strong research evidence for poor reliability of determinations of the presence and direction of lateral shifts. [Donahue MS, Riddle DL, Sullivan MS. Intertester reliability of a modified version of McKenzie's lateral shift assessments obtained on patients with low back pain. Phys Ther. 1996;76:706-726.]

The McKenzie system[1] is a commonly used method of examining and treating patients who have low back pain (LBP).[2] Physical therapists using the McKenzie system classify a patient's condition into one of nine syndromes. Classification is based, in part, on data collected during file patient's past medical history and on assessments of the patient's posture, lumbar spine range of motion (ROM), and pain symptom response to repeated movements.

The intertester reliability of McKenzie's classifications of the type of syndrome present was recently examined. Riddle and Rothstein[3] found the reliability of these classifications to be poor. According to the authors, numerous sources of error appeared to be present in the data. One of the sources of error was the assessment of the presence of a lateral shift.

McKenzie[1] has defined a lateral shift as a lateral displacement of the patient's trunk in relation to the pelvis. If the patient's trunk is displaced to the right relative to the pelvis, a right lateral shift is present. If the patient's trunk is displaced to the left relative to the pelvis, a left lateral shift is present. The presence of lateral shifts is determined during the postural examination.

McKenzie[1] stated that there are many causes of a lateral shift and that it is imperative to determine whether the lateral shift is relevant to the patient's current condition. According to McKenzie, the shift is considered to be clinically relevant when a side-glide test (a frontal-plane ROM test of the trunk) alters the location or intensity of the pain reported by the patient. McKenzie therefore recommended the use of a two-step procedure to determine when clinically relevant lateral shifts are present. The first step requires the therapist to observe the patient's standing posture to determine whether a lateral shift is present. The second step requires the therapist to test for the clinical relevance of a lateral shift by using side-glide tests to determine whether the site or the intensity of the pain reported by the patient can be altered. If the lateral shift is determined to be clinically relevant, then, according to McKenzie, the lateral shift must be corrected prior to the use of other treatment procedures or the patient's symptoms may worsen.

Observational analysis for the presence of a lateral shift is the first step in McKenzie's two-step lateral shift assessment. Several researchers[3-5] have investigated the intertester reliability of observational assessments of lateral shifts. The percentage of agreement among clinicians in these studies ranged from 55%[5] to 70%.[4] One of the studies reported a kappa coefficient, a statistic that estimates the degree of agreement after eliminating chance agreement. The kappa was reported to be .96 and was based on measurements taken on 318 subjects.[3] Because treatment in the McKenzie protocol is based, in part, on the observational assessment of lateral shift, the degree of agreement reported in these studies appears to be unacceptably low.

A portion of the error associated with assessments of lateral shift may be due to therapists' inability to determine, based on observation alone, when a lateral shift is present. McKenzie appeared to agree with this notion when he stated that "the lateral shift is sometimes barely discernable, and great care must be taken to ensure that the deformity is not overlooked."[1](p35)

The second step in McKenzie's two-step process of lateral shift assessment is the side-glide test sequence. The side-glide test sequence is a series of movement tests designed, in this situation, to assess the effect of movement on the patient's pain. No studies were found that examined the reliability of assessments made using the side-glide test sequence or using the two-step process of lateral shift assessment described by McKenzie.

Because the McKenzie system requires therapists to observe posture and perform side-glide tests to assess for the presence of clinically relevant lateral shifts, we believe that it is important to study this two-step procedure. Without an acceptable degree of reliability for assessments of the presence of lateral shifts, however, McKenzie's two-step process for determining when clinically relevant lateral shifts are present also is likely to be unreliable. We therefore saw a need to develop a new instrumented technique for determining when a lateral shift is present. The instrument was designed so that the therapist would not have to be dependent on observation alone to determine when a lateral shift is present. We believed that because observational assessments of lateral shifts are unreliable, the use of a simple instrument might provide more reliable data.

The purpose of this study was to examine the intertester reliability of assessments of the presence of clinically relevant lateral shifts using the two-step process described by McKenzie. We modified McKenzie's procedures for determining when lateral shifts were present (step 1) because the reliability of these procedures has been shown to be inadequate for clinical use. We used a simple instrument to aid in the detection of lateral shifts. We also used the side-glide test sequence (step 2) described by McKenzie.

Method

Pilot study

Because of the poor reliability associated with observational assessments Of lateral shifts, we completed a pilot study designed to determine which of three instruments could be used to most consistently detect the presence, direction, and magnitude of lateral shifts in a group of patients with LBP. We wanted to develop a measurement technique that was both theoretically sound and easy to use.

Ten consecutive patients referred for treatment of their LBP to the Department of Physical Therapy at the Medical College of Virginia Hospital (Richmond, Va) were considered and subsequently accepted for participation in the pilot study. Exclusionary criteria for the pilot study related to patients who may have had a lateral shift for reasons other than LBP. Patients who previously had a radiographically diagnosed structural scoliosis, a leg-length discrepancy, or a spinal fusion were excluded. All patients signed a consent form prior to participation. Four therapists who did not participate in the main study were used to collect data for the pilot study. Two of the four therapists were randomly paired for each patient.

Subjects were required to stand with their feet approximately 50.5 cm (19 in)apart and with their weight evenly distributed for all three techniques. The' therapist first marked the skin overlying each subject's first lumbar (L-I) and first sacral (S-1) spinous processes with 0.6-cm (1/4-in) adhesive markers with a dot in the center. Procedures for identifying these spinous processes were described by Hoppenfeld.[6] The adhesives were removed after each measurement technique was used.

The first measurement technique required the therapist to align a ruler with the dots on the adhesive markers. A gravity-referenced protractor* was then aligned with the ruler to determine whether a lateral shift was present. A lateral shift was judged to be present if the adhesive markers were not aligned vertically. If a lateral shift was present, the magnitude of the lateral shift was measured by determining the angle made by the protractor.

The second measurement technique required the therapist to position the subject so that the adhesive marker overlying the first sacral spinous process was aligned with a plumb line positioned just behind the subject. The therapist determined whether a lateral shift was present by comparing the plumb line with the adhesive marker overlying L-1. If a lateral shift was judged to be present, the therapist used a ruler to measure the distance between the adhesive marker overlying L-1 and the plumb line to determine the magnitude of the lateral shift.

The third measurement technique tested was the one used in this study. Therapists were required to use a level to determine whether the adhesive markers overlying the L-1 and S-1 spinous processes were aligned vertically. If the adhesive markers were not aligned vertically, the therapists determined whether a lateral shift to the right or left was present and also measured the magnitude of the lateral shift with a ruler (see "Method" section for a thorough description of this technique).

The results of the pilot study suggested the data obtained with the third measurement technique was the most reliable. Therapists agreed on whether a lateral shift was present and the direction of the lateral shift in 60% of the subjects. Therapists reported the third measurement technique also was most suitable for clinical use because it was most convenient and easy to use.

Subjects

A sample was selected from the Outpatient Department at Sheltering Arms Rehabilitation Hospital (Richmond, Va). The first 52 patients who were referred for treatment for LBP and who agreed to participate were considered for admission to the study. Exclusionary criteria were the same as for the pilot study. Patients who reported being diagnosed with a structural scoliosis were excluded because the vertebral wedging of scoliosis is a different phenomenon than the lateral shift that was investigated in this study. Although it was our intent to exclude patients with a scoliosis, it is possible that patients with an undiagnosed scoliosis were admitted to the study. Patients with a radiographically diagnosed leg-length discrepancy (LLD) were excluded because the trunk may shift laterally to compensate for the LED. Patients with a lumbar spinal fusion were excluded because file fusion may have resulted in a lateral deviation of the lumbar spinal segments. Lateral shifts caused by abnormally aligned and fused spinal segments also are different from the lateral shifts we were interested in studying. Three patients were excluded from participation because they had a lumbar fusion. A total of 49 patients, therefore, were admitted to the study.

Subjects included in this study were adults (18-78 years of age) who were being seen in the physical therapy department for LBP with or without lower-extremity pain. New patients and patients being treated at the time of this study were examined. The patient sample is described in Table 1.

Testers

The lateral shift assessments were performed by 10 physical therapists employed full-time in the clinic at the time of the study. Seven of the therapists had been practicing physical therapy for 3 years or less at the time of the study. The therapists reported that patients with LBP made up between 30% and 50% of their caseload. None of the therapists reported that they had received postgraduate training in the McKenzie protocol prior to the study. The therapists were given a written description of the two-step procedure and were asked to practice the procedure on patients and other subjects until the therapists felt their accuracy in using the method was adequate for clinical use.

Procedure

The purpose of the study was explained to patients who met the criteria for admission. If a patient agreed to participate in the study, he or she signed a consent form. The procedure for data collection was as follows.

Two therapists were randomly selected from the pool of therapists. Therapists were paired using a computer-generated random list. If a patient was being treated, the treating therapist did not examine that patient. The recorder (MSD) determined therapist pairings for all patients admitted to the study and was present during all examinations to record data.

The first therapist entered the treatment booth. the subject was asked to stand in a relaxed position with feet positioned approximately 50.5 cm (12 in) apart, with weight evenly distributed over his or her feet and arms at the sides. The therapist marked the skin overlying the spinous processes of the first lumbar vertebrae and first sacral vertebrae with 0.6-cm-diameter (1/4-in-diameter) round adhesive markers. The center of the adhesive markers was previously marked with a dot. The S-1 spinous process was then located in the following manner. The superior aspects of the iliac crests were identified. The therapist placed the radial sides of his or her hands on the superior aspects of the iliac crests. The therapist extended his or her thumbs to find the L4-5 intervertebral space, which is located on an imaginary line drawn between the superior aspects of the iliac crests.6 The therapist then palpated two spinous processes down from this space to locate the S-1 spinous process. The therapist placed an adhesive marker on the skin overlying the center of S-1.

The L-1 spinous process was identified in the following way. The therapist identified the L4-5 intervertebral space as defined previously. The therapist then palpated and counted up four spinous processes from the L4-5 intervertebral space to the L-1 spinous process. The therapist placed an adhesive marker on the skin overlying the center of L-1 (Fig. 1).

Each therapist then used the following method to determine whether a lateral shift was present and, if present, the direction and magnitude of that lateral shift (Fig. 2). A 22-cm bubble level+ and a 30-cm ruler++ were used. The straight edge of the level was aligned with the dot on the S-1 adhesive in such a way that the bubble was midway between the markings on the level, indicating a vertical position. The therapist noted whether a lateral shift was present or absent by determining whether the dots on the two adhesives were aligned vertically. When the dot overlying L4 was aligned to the left of the vertically aligned level, a left lateral shift was judged to be present. When the dot overlying the D1 spinous process was aligned to the right of the vertically aligned level, a right lateral shift was judged to be present. If the dots on the two adhesive markers were aligned vertically, no lateral shift was judged to be present.

If a lateral shift was judged to be present, the direction of the lateral shift was determined by the therapist. This information was reported to the recorder. If a lateral shift was judged to be present, the therapist measured the distance between the straight edge of the level and the dot at the center of the marker overlying L-1. This information was reported to the recorder. The adhesive markers were removed and the patient was asked to walk around the treatment booth until the second therapist entered the booth.

The first therapist exited the treatment booth, and the second therapist entered. The second therapist started the assessment as soon as possible after the first therapist left the treatment booth. The procedure for the second therapist was identical to that used by the first therapist.

Following the second therapist's completion of the lateral shift assessment, the second therapist performed the side-glide test sequence. The side-glide test sequence was performed after both therapists had completed the assessment for the presence of the lateral shift in order to eliminate a possible source of error. The side-glide tests may have altered the lateral shift, which would have contributed error to the lateral shift assessments.

The side-glide test sequence was described by McKenzie[1] and was performed in the following order: right side-glide test, left side-glide test, repeated right side-glide test, and repeated left side-glide test. The sequence of side-glide tests was stopped if a single positive test was found. the test was considered positive if the subject reported that the pain changed in location or intensity during or immediately after the test.

The right side-glide test was performed as follows (Fig. 3). Prior to beginning the side-glide test, the therapist asked the subject to describe the location of the pain. The subject also was asked to rate the intensity of the pain using a 0-to-10 verbal rating scale? The subject stood facing the examiner. The therapist placed his or her right hand on the subject's left shoulder and his or her left hand on the subjects's right iliac crest. The subject was asked to move his or her hips to the left and the shoulders to the right, keeping the shoulders parallel to the ground. The therapist pressed both hands toward the midline, assisting the movement of the top half of the subject's body to the right to the end of the range. Following the test, the therapist asked the subject to describe the location and rate the intensity of the pain. If the test altered the location, or increased or decreased the intensity of the pain reported by the subject, a positive right side-glide test was judged to be present. The therapist reported this information to the recorder.

The left side-glide test was then performed as follows. The therapist placed his or her left hand on the subject's right shoulder and his or her right hand on the subject's left iliac crest. The therapist then pressed both hands toward the midline, assisting a movement of the top half of the subject's body to the left to the end of the range. Following the test, the therapist asked the subject to describe the location and to rate the intensity of the pain. If the side-glide test altered the location or increased or decreased the intensity of the reported pain, a positive left side-glide test was judged to be present. the therapist reported this information to the recorder.

If neither the side-glide test to the right nor the sideglide test to the left altered file location or the intensity of the subject's pain, the therapist had the subject perform the repeated side-glide tests. McKenzie[1] stated that if there is no change in a patient's symptoms during or immediately following a test movement, the joints may not have been adequately tested and the process should be repeated by using up to 10 repetitions for each procedure.

The repeated right side-glide test was done first. The therapist asked the subject to describe the location and rate the intensity of the pain. The therapist assisted the subject in performing up to 10 repetitions of the side-glide test to the right using the same techniques as for the single-movement test. The therapist instructed the subject to stop if his or her pain location or intensity changed prior to completing 10 repetitions. The repeated left side-glide test was then performed, if needed.

The second therapist then exited the treatment booth, and the first therapist entered. The side-glide testing procedure was identical to that used by the second therapist. For the purposes of this study, therapists were not required to agree on the direction of the side-glide test that altered the symptoms. The therapists were only required to report whether a positive side-glide test was present. A positive side-glide test is required to determine whether a lateral shift can be judged to be relevant.

The results of the lateral shift assessment and the results of the side-glide test sequence were combined by McKenzie[1] to make judgments about the presence of a clinically relevant lateral shift.

One of three possible decisions could be made about the presence of a lateral shift. the subject could have been determined to have a lateral shift to the right, a lateral shift to the left, or no lateral shift. One of two possible decisions could be made based on the side-glide test sequence. The subject could have been determined to have either a positive or a negative side-glide test sequence. There was, therefore, a total of six possible decisions that could have been made based on the data. We chose to collapse these six possible iterations into four judgments because it is these four judgments that determine, in part, how a patient should be treated using the McKenzie approach. The two most obvious decisions that guide treatment are whether a clinically relevant lateral shift to the right is present or whether a clinically relevant lateral shift to the left is present. The third decision is that the patient has a negative side-glide test sequence. This judgment is made in file presence or absence of a lateral shift. According to McKenzie, when a negative side-glide test sequence is present, a clinically relevant lateral shift is not present and therefore does not require treatment.

The fourth decision is that a positive side-glide test sequence is present but a lateral shift is not judged to be present. McKenzie[1] suggested that patients with a positive side-glide test sequence should be treated differently from patients with a negative side-glide test sequence. For example, McKenzie implied in his text that the side-glide test sequence may be used in lieu of observational assessment to determine when clinically relevant lateral shifts are present but are not readily detectable by observation.

The following are the four possible judgments that were made based on the data collected by each therapist using McKenzie's two-step process:

A clinically relevant lateral shift to the right was present. This decision required the therapist to determine that a lateral shift to the right was present and that the side-glide test was positive.

A clinically relevant lateral shift to the left was present. This decision required the therapist to determine that a lateral shift to the left was present and that the side-glide test was positive.

No lateral shift with a positive side-glide test was present. This decision required the therapist to determine that no lateral shift was present but that the side-glide test was positive.

A negative side-glide test with or without a lateral shift was present. This decision required the therapist to determine that the side-glide test sequence was negative. A lateral shift could have been judged to be either present or absent.

The algorithm used to make decisions based on the two-step process for identifying clinically relevant lateral shifts is shown in Figure 4.

Data Analysis

The nominal-level data generated by this study were analyzed using percentage of agreement and the generalized kappa coefficient.[8] The generalized kappa estimates file degree of agreement after chance agreement has been eliminated.

Results

The kappa coefficient for intertester agreement of the assessment of the presence of clinically relevant lateral shifts obtained on the entire sample was .16. The percentage of agreement was 47% (Tab. 2).

Discussion

The results of this study suggest that McKenzie's two-step process for lateral shift assessment does not produce reliable data even when instruments are used. The low kappa coefficient indicates that the agreement is only slightly better than would be expected by chance alone. Post hoc analyses were therefore done to determine the possible sources of this substantial amount of error. We decided to examine each of the steps in McKenzie's two-step process to determine the likely sources of error in the judgments.

Reliability of Assessments of the Presence and Direction of Lateral Shifts

The intertester reliability of assessments of the presence and direction of lateral shifts using the level and ruler was examined by calculating a generalized kappa coefficient. The kappa was .00, and the percentage of agreement was 43% (Tab. 3). The low kappa value suggests that the extent of agreement for assessments of the presence and direction of lateral shifts is essentially what would be expected by chance. In the only study that has used a kappa coefficient to report the reliability of visual estimates of the presence of lateral shifts, a kappa value of .26 was found.[3] The instrument used in our study provided data that were no more reliable than previously reported data that were obtained by observation. It appears that assessments of lateral shifts are unreliable when obtained with a simple instrument, as used in this study, or when obtained by observation, as described by McKenzie.

Several factors may have contributed to the poor reliability of assessments of the presence and direction of lateral shifts using the method proposed in this study. Kerlinger[9] argues that three major sources of error are present in clinical measurements. Error may be attributable to the instrument, the examiner, or the subject.

One possible source of error related to the instrumentation was the placement of the adhesive markers on the skin overlying the spinous processes. Therapists reported that it was difficult to place the adhesive circles accurately over the L-1 and S-1 spinous processes. Therapists reported that it was difficult to remove the palpating finger and place the adhesive circle on the skin overlying the center of the spinous process. In some cases, the magnitude of the lateral shift was found to be on the order of a few millimeters. Slight misplacement of the adhesive markers could conceivably contribute to errors in judgment, especially when the lateral shift is measured to be only a few millimeters in size. The therapists indicated that palpation and adhesive marker placement was especially difficult on subjects who appeared to be obese. Inaccurate or inconsistent adhesive marker placement may have decreased rater agreement.

The examiners also may have contributed error to the assessments of the presence of lateral shifts. Most of the participating therapists had a limited amount of clinical experience treating patients with LBP. Seven of the 10 therapists involved in this study had been treating patients with LBP for 3 years or less at the time the study was performed. Five of the 10 therapists had been treating patients with LBP for 1 year or less. Inaccurate palpation of the bony landmarks due to inexperience may have contributed to the error.

The therapists in our study had no postgraduate training in the use of the McKenzie system. Data collected in a previous study,s however, suggested that therapists who had attended at least one postgraduate course on the McKenzie system were no better at visually assessing for lateral shifts than were therapists without the training. The percentage of agreement among those therapist with postgraduate training was 52% compared with 61% for file therapists with no training. The kappa coefficient was .12 for therapists with postgraduate training compared with .27 for therapists with no postgraduate training in the use of the McKenzie system. Postgraduate training in the use of the McKenzie system does not appear to improve a therapist's ability to classify, lateral shifts.

Finally, a third source of error may be that the postural deviation known as lateral shift is inherently variable. McKenzie[1] stated that a small number of patients exhibit a lateral shift that may vary in direction. Perhaps the postural deviation that has been recognized as a lateral shift may change as the patient changes positions.

Previous studies[3-5] have shown that observational assessments of lateral shifts result in judgments that are unreliable. Additionally, the method used in our study, did not yield reliable judgments about the presence and direction of lateral shifts. No method bas been demonstrated to yield reliable judgments about the presence and direction of lateral shifts in patients with LBP. In order for lateral shift assessments to be useful, further study is needed to identify a method that can be used to obtain reliable assessments of the presence and direction of lateral shifts.

Effect of Magnitude of Lateral Shift on Reliability

Tenhula et al examined 24 patients with LBP who had "an observable lateral lumbar shift."10(p483) The authors stated that only patients with an observable lateral shift were selected for the study. Apparently, patients who had what appeared to be subtle lateral shifts or patients who did not appear to have a lateral shift were not admitted to the study. One therapist made the judgments about which subjects had observable lateral shifts. A second therapist determined the presence and direction of each lateral shift by viewing a slide image of the patient. Intertester agreement between the therapists was assessed using the kappa statistic. Perfect agreement (kappa=1.00) was found.

The results of the study by Tenhula et al[10] indicate that some lateral shifts can be reliably assessed. Tenhula et al, however, did not operationally define the criteria for determining when an observable lateral shift was present. That is, Tenhula and colleagues did not describe how large lateral shifts must be in order to be readily observable and therefore reliably classified. In addition, the study of Tenhula et al cannot be generalized to all patients with LBP because only a subset of patients with what were reported to be observable lateral shifts were studied. More study is needed to determine whether the magnitude of lateral shifts influences reliability.

The results of the study by Tenhula et al[10] suggest that some lateral shifts that are reportedly judged to be observable can be reliably assessed. We therefore examined the effect of magnitude of lateral shift on reliability in our study.

Post hoc analysis of the magnitude of the lateral shifts as measured by the therapists revealed the following findings. Both therapists reported lateral shifts of 2 mm or greater (range=1-7) in 26 of the 49 subjects. In only 15 of these 26 subjects, or 58% of the tithe, did the therapists agree on the direction of the lateral shift, as compared with a 47% agreement for the entire sample. Both therapists recorded magnitudes of 3 mm or greater in 12 of the 49 subjects examined in our study. In only 7 of these 12 subjects, or 58% of the time, did the therapists agree on file direction of the lateral shift. Reliability did not appear to improve even for the larger lateral shifts measured in this study.

Intertester Reliability of Judgments Made Using the

Side-Glide Test Sequence The side-glide test sequence is routinely used in the McKenzie protocol. The side-glide tests are used to identify limited and painful spinal motions and to determine the relevance of a lateral shift to the patient's pain complaints. The lateral shift is considered relevant when the side-glide test alters the site or intensity of the pain. If the patient reports no change in the intensity or location of the pain during the performance of the side-glide test sequence, the lateral shift, if present, cannot be considered part of the mechanical problem causing file pain.[1] McKenzie[1] recommended performing the sequence of side-glide tests and recording the patient's pain response to each test movement. For the purpose of our study, subjects were determined to have a positive side-glide test sequence when side gliding to either the right or left side altered the location or intensity of the subjects' pain. The therapists were not required to agree on the direction of the side-glide test that altered the symptoms. The therapists only had to determine that a positive side-glide test was present. The test sequence was stopped after the first positive side-glide test because it was established that a side-glide test altered the location or intensity of the subjects' pain.

A generalized kappa was calculated to describe the degree of agreement for the side-glide test sequence. The kappa was .74, and the percentage agreement was 94%. Therapists were able to reliably determine whether the location or intensity of the subjects' pain was altered during the side-glide test sequence.

In our study, both therapists agreed that a positive side-glide test was present in 84% of the subjects. Tenhula et al[10] also found a high proportion of patients exhibiting a positive response to a frontal-plane trunk movement test. Tenhula et al investigated the relationship between direction of lateral shift and patient performance on a side-bending test. The side-bending test was considered positive if the movement produced or increased the patients' pain. Tenhula et al reported that 92% of the patients in their study demonstrated a positive side-bending test. The results of study by Tenhula et al and of our study suggest that a large proportion of patients with LBP report that their pain is altered with side-gliding or side-bending tests.

The finding of a high degree of reliability of assessments obtained from tests designed to alter pain is similar to the findings of studies performed by Potter and Rothstein[11] and Maher and Adams.[12] Potter and Rothstein determined that of 13 commonly used tests for sacroiliac joint dysfunction, only the 2 tests that assessed the response of the patient's pain to the test were shown to yield reliable results. Maher and Adams examined the intertester reliability of judgments of stiffness and pain at the lumbar spine motion segments using posteroanterior central pressure tests. Tests were done on 90 patients with LBP. The results of the study by Maher and Adams indicated that the reliability of judgments based on reproduction of the patients' pain were good, whereas judgments based on the determination of spinal stiffness were shown to be unreliable.

Our study has shown that assessments of the effects of the side-glide test sequence on pain are reliable. Our study, however, only examined the reliability of side-glide testing procedures as they relate to the determination of the relevance of a lateral shift. Specifically, this study only examined whether the side-glide test sequence can be used to alter a patient's reported pain location or intensity. More study is needed to assess other judgments made based on use of the side-glide tests.

The Role of Lateral Shift Assessment in the McKenzie System

McKenzie reported, based on his own survey data, is that 52% of patients with LBP were found to have a clinically relevant lateral shift. No data were reported by McKenzie to support the reliability of his lateral shift assessments. McKenzie[1] claimed that clinically relevant lateral shifts are a common and important examination finding in patients with LBP.

Of the nine classifications proposed by McKenzie,[1] three classifications require the therapist to determine that a clinically relevant lateral shift is present. These are the derangement 4 and derangement 6 classifications and a classification that includes certain dysfunction syndromes. The criteria required to assign a derangement 4 classification to a patient are identical to the derangement 3 criteria except that patients with a derangement 4 classification must be found to have a clinically relevant lateral shift. Derangements 5 and 6 differ in a similar way in that the only difference between them is that a clinically relevant lateral shift must be found in patients with a derangement 6 classification. It would appear, therefore, that lateral shift assessments are critical, especially when differentiating between these two sets of derangements. Because treatment is begun by attempting lateral shift correction for patients having derangement 4 and 6 classifications, reliable identification of clinically relevant lateral shifts is especially important.

Clinically relevant lateral shifts also are found in some patients who are classified as having a dysfunction syndrome. In patients with what McKenzie[1] called a side-gliding dysfunction syndrome, treatment requires the therapist to correct the lateral shift.

Assessments of lateral shifts appear to be an important part of the McKenzie system. Lateral shift assessments are critical not only for classification but also because treatment is determined, to a great extent, by the judgments made based on these procedures. Based on the results of this study and others,[3-5] it appears that McKenzie's method of assessing for the presence and direction of lateral shifts should be changed. We were unable to develop a technique that could be used to obtain reliable assessments of the presence and direction of lateral shifts.

Conclusions

The two-step process McKenzie advocated for identifying clinically relevant lateral shifts is unreliable and therefore potentially misleading. The error associated with McKenzie's two-step process of lateral shift assessment appears to be due primarily to the first step in the process. Assessments of the presence and direction of lateral shifts are error-prone procedures in the McKenzie protocol. The assessments obtained with the instrument used in our study were unreliable, much like observational assessments of lateral shifts. More study of the McKenzie system is needed to determine whether an alternative method can be developed that results in reliable assessments of the presence and direction of lateral shifts. If a reliable method for lateral shift assessment cannot be developed, the McKenzie system should be modified. therapists should base their clinical decisions on procedures with evidence of adequate reliability such as the side-glide test sequence. Assessments made based on the side-glide test sequence provide some reliable information, but it is not clear how the data would be used in the McKenzie system in the absence of assessments of the presence and direction of lateral shifts.

Acknowledgments

We thank the physical therapists in the Outpatient Department at Sheltering Arms Rehabilitation Hospital for participating in this study. We also thank the physical therapists at the Medical College of Virginia Hospital for participating in the pilot study.

References

1 McKenzie R. The Lumbar Spine: Mechanical Diagnosis and Therapy. Waikanae, New Zealand: Spinal Publications Ltd; 1981.

2 Battie MC, Cherkin DC, Dunn R, et al. Managing low back pain: attitudes and treatment preferences of physical therapists. Phys Ther. 1994;74:219-227.

3 Riddle DL, Rothstein JM. Intertester reliability of McKenzie's classifications of the syndrome types present in patients with low back pain. Spine. 1993;18:1333-1344.

4 Nelson MA, Allen MA, Clamp SE, DeDombal FT. Reliability and reproducibility of clinical findings in low back pain. Spine. 1979;4:97101.

5 Kilby J Stigant M, Roberts A. The reliability of back pain assessment by physiotherapists using a "McKenzie algorithm." Physiotherapy. 1990; 76:579-583.

6 Hoppenfeld S. Physical Examination of the Spine and Extremities. New York, NY: Appleton-Century-Crofts; 1976.

7 Murphy DF, McDonald A, Power C, et al. Measurement of pain: a comparison of the verbal analogue with a nonvisual analogue scale. Clin J Pain. 1988;3:197-199.

8 Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull. 197l;76:378-382.

9 Kerlinger FN. Foundations of Behavioral Research. Philadelphia, Pa: Harcourt, Brace, Jovanovich Publishers; 1986.

10 Tenhula JA, Rose SJ, Delitto A. Association between direction of lateral lumbar shift, movement tests, and side of symptoms in patients with low back pain syndrome. Phys Ther. 1990;70:480-486.

11 Potter NA, Rothstein JM. Intertester reliability of selected clinical tests of the sacroiliac joint. Phys Ther. 1985;65:1671-1675.

12 Maher C, Adams R. Reliability of pain and stiffness assessments in clinical manual lumbar spine examination. Phys Ther. 1994;74:801809.

13 McKenzie RA. Prophylaxis in recurrent low back pain. N Z Med J. 1979;89:22-23.

* Sears Craftsman Universal Protractor, Sears, Roebuck & Co, Sears Tower, Chicago, IL 60684.

Key Words: Nick and trunk, back; Pain; Tests and measurements, functional

MS Donahue, PT, was a candidate for the Master of Science degree, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia Campus, Virginia Commonwealth University, Richmond, VA, when this research was conducted.

DL Riddle, PT, is Associate Professor, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia Campus, Virginia Commonwealth University, Box 980224, MCV Station, Richmond, VA 23298 (USA) ([e-mail]). Address all correspondence to Mr Riddle.

MS Sullivan, PT, is Assistant Professor, Department of Physical Therapy, School of Allied Health Professions, Medical College of Virginia Campus, Virginia Commonwealth University.

This study was approved by the Institutional Review Board of Virginia Commonwealth University.

  Table 1.
    Characteristics of Patient Sample (N=49)

Variable

Gender

 Female                                              34
 Male                                                15
Age (y)
                                                     39.2
SD                                                   13.9
Range                                                (18-78)
Duration of symptoms (mo)
                                                     9.8
SD                                                   3
Range                                                (0.75-120)

Pain distribution
Bilateral, unilateral, or midline
LBP,[sup.a] with no buttock or
lower-extremity pain                                 34
Bilateral, unilateral, or midline LBP, with
unilateral buttock or lower-extremity pain           12
Bilateral LBP, bilateral lower-extremity pain        1
Patients not reporting pain distribution             2

Number of previous episodes of low back pain
No previous episodes                                 39

1 episode                                            6
2 episodes                                           2
3 episodes                                           2

"LBP=low back pain.

Exact 27020, LS Starrett Co, Athol, MA 01331.

Westcott no. R590-12, Acme United Corp, Fairfield, CT 06430.

Table 2.

Assessments of the Presence, Direction, and Relevance of Lateral Shifts"

  Therapist 2

                 Relevant   Relevant   Positive   Negative
                 Right      Left       Side       Side Glide
                 Shift      Shift      Glide,
                                       No Shift

Relevant right shift   2        7        3        0
Relevant left shift    4        16       5        1
Positive side
glide, no shift        1        3        0        0
Negative side glide    1        1        0        5

"Generalized kappa=.16, percentage of agreement=47%.

  Table 3.
    Assessments of the Presence and Direction of Lateral Shifts"

Therapist 1

Therapist 2

                 Left Lateral    Right Lateral      No Lateral
                    Shift            Shift             Shift

Left lateral shift    18               8                 4
Right lateral shift   5                3                 2
No lateral shift      6                3                 0

"Generalized kappa=.00, percentage of agreement=43%.

[FIGURES HAVE BEEN OMITTED]