Measurement of accessory motion: critical issues and related concepts.

Manual therapy (MT) encompasses a large number of examination and treatment procedures. Implicit in the term "manual therapy" is that the therapist performs manual procedures to collect data for clinical decision making. Manual therapists, however, sometimes use instruments to collect patient data. For the purposes of this article, therefore, manual therapy is defined as examination and treatment procedures used by therapists on patients with musculoskeletal problems. Because this article focuses on measurement issues, only the examination procedures used by manual therapists will be addressed (see article by Di Fabio in this issue for a discussion of treatment issues in MT).

The examination procedures used by manual therapists on patients with musculoskeletal problems can be classified, for the purposes of this article, into several categories. Those categories of procedures that require the therapist to manually apply forces to the patient (eg, range-of-motion [ROM] tests, manual muscle tests) make up a large part of the examination process. Categories of tests that do not require the therapist to manually apply forces are observational data (eg, gait or postural assessment, active-range-of-motion tests) and instrumented testing. In addition, manual therapists use other data in their decision-making process such as the patient history, verbal and affective responses of the patient to various examination and treatment procedures, and results of invasive diagnostic tests. The examination procedures most manual therapists use and that require the application of manually applied forces can be divided into several broad categories. These broad categories serve to organize the many manually applied examination procedures used by manual therapists into a logical scheme for purposes of discussion. The categories of procedures are ROM tests (passive and resistive), manual muscle tests, palpation tests, accessory motion tests (AMTs), and neurological tests. Some of these categories are self-explanatory, whereas others require some clarification. Palpation tests are included as a separate category because palpation procedures are somewhat similar in terms of what is required of the therapist. The therapist attempts to palpate the skin overlying a specific structure to make some inference about that structure. Neurologic tests such as deep tendon reflex tests or pinprick tests for sensation warrant a separate category because they are part of a series of tests designed to assess the integrity of a patient's nervous system. Accessory motion tests are designed to assess various attributes related to the movement that occurs between joint surfaces. Accessory motion tests are thoroughly discussed later in this article.

The categories of tests that require the therapist to impart forces on the patient have similar attributes. Most tests from these categories require the therapist to closely interact with the patient during the procedure. Most of the observational tests (eg, observational gait analysis), however, do not require a high degree of therapist-patient dialogue.

The patient-therapist interaction appears to be a crucial aspect of clinical decision making in MT. This interaction aids in determining the magnitude and velocity of the forces the examiner imparts to the patient during the examination and subsequently during the treatment. Maitland,[1] for example, claims that changes in a patient's pain response is the most important factor to consider when doing examination procedures designed to assess joint movement. McKenzie[2] recommends the use of assessments of the effect of repeated ROM tests on the location and intensity of a patient's pain when making treatment decisions. The therapist must interact with the patient during the procedures recommended by Maitland and McKenzie in order to make correct assessments. The formidable challenge facing clinical researchers in MT is to attempt to define meaningful measurements obtained during those MT procedures that require responses from the patient. Studies that examine the consistency of examination procedures that necessitate responses from the patient are needed in the MT literature.

Throughout the examination and treatment process, the manual therapist makes clinical decisions based on some or all of these examination data. The processes used to make clinical decisions appear to be very complex and, initially, undefinable. Physical therapists practicing MT, however, must define the methods they use to make clinical decisions. If meaningful treatment effectiveness studies are to be carried out, therapists must be able to define their bases for choosing the appropriate treatment.

Much of the measurement research in MT has focused on the reliability of the different elements (tests) that make up the examination procedure. This research is useful in that it provides data to suggest some tests may be more useful than others. Most manual therapists, however, probably do not make treatment decisions based on information obtained in an isolated fashion from one or two tests. Many measurement studies address the usefulness of one or two examination procedures in isolation. The MT literature lacks measurement studies that combine clusters of tests commonly used by manual therapists to make treatment decisions.[3]

Sources of Error in Manual Therapy

Errors in clinical decision making within and among clinicians are an inevitable consequence of clinical practice. The goal of all clinicians should be to minimize the sources of error that lead to incorrect clinical decisions. Many potential sources of error are likely to affect decisions made based on MT examination procedures. These errors may be due to inconsistency in the obtained measurements (measurement error) or to errors in the clinical decision-making process. Sources of measurement error can be directly examined and identified in clinical research studies. Sources of error in the clinical decision-making process are somewhat more elusive because the processes used to make clinical decisions in MT are not clearly defined for most approaches. Clinical research in the form of case studies is a critical first step to elucidate the methods manual therapists use to make clinical decisions.

Several resources exist that can provide the physical therapist with useful information on measurement issues and clinical decision making.[4-7] Sackett et al,[8] have attempted to provide a comprehensive description of the potential sources of error during clinical examinations. The authors describe the three primary sources of error as being the examiner, the patient, and the examination (the MT examination procedure, in this case).

Error Attributable to the Examiner

The examiner brings many potential sources of error to the examination. Sackett and colleagues[8] claim that a frequent source of examiner error is the particular bias of the clinician doing the examination. For example, because some clinicians may believe spinal joint hypomobilities (decrease in the accessory motion) are an important clinical finding, these clinicians may be biased toward finding hypomobilities more often. Therapists must attempt to eliminate bias when making clinical decisions.

Another important potential source of error attributable to the examiner occurs when the examiner makes inappropriate inferences based on an examination procedure. An example may help to illustrate this source of error in clinical decision making. An examiner performs an AMT of a spinal segment and concludes the segment is unstable (inference no. 1). The examiner then makes a second inference that the hypothesized joint instability is the cause of the functional deficit (eg, an inability to pick up objects from the floor without pain). On the basis of the AMT, the examiner concludes that the patient has a joint instability (inference no. 1) and chooses to treat the patient with procedures designed to decrease movement of the involved segment in order to increase the patient's function (based on inference no. 2).[9] One or both of the examiner's inferences in the example may have been incorrect. The AMT may not have indicated a joint instability was present, or the patient's functional deficit may have nothing to do with a joint instability. Even if a joint instability was present, the instability may not have caused the functional deficit. The clinical decision-making process was flawed because the therapist made inappropriate inferences based on the AMT.

Much study is needed in MT to examine the criterion-related validity of the measurements obtained using MT procedures. The studies of Bogduk,[10] Cooperman et al,[11] and Jull and Bogduk[12] serve as models for validity studies in MT and will be discussed later in this article. Studies that examine the predictive value of positive and negative tests and the sensitivity and specificity[4] of MT testing procedures should be a high priority in MT research.

The level of prior training and experience of the examiner may also affect the error in MT measurements. For example, Farrell and Jensen (see their article in this issue) suggest that palpatory skills require many hours of training and practice to master. The claim that highly experienced examiners are more skillful and therefore more consistent in their palpatory skills than their less experienced colleagues seems intuitively sound, No studies, however, were found in the MT literature that provide data to support this assertion. Studies that examine the effects of specialized training or experience on the reliability and validity of measurements would help to clarify the importance of training and experience in MT.

The amount of examiner experience and training would appear to affect a therapist's ability to assess accessory motion. Some of the grading scales used to make judgments about the amount of accessory motion present lack operational definitions. Kaltenborn,[13] for example, recommends the use of an ordinal grading scale to describe the amount of accessory motion present (Appendix). Kaltenborn did not operationally define the different grades in the scale. For example, Kaltenborn did not describe how to distinguish between slight decreased movement and considerable decreased movement. Kaltenborn implies the therapist must use an experiential model when determining a grade of movement. That is, the therapist must rely entirely on past experience to decide when normal or abnormal accessory motion is present. Because most grading scales of the amount of motion present require the therapist to make a judgment based on past experience, it could be argued that clinical experience may affect assessments of accessory motion. No data exist, however, to support the contention that highly trained examiners can more reliably perform AMTs than less experienced examiners. Some data exist to suggest highly trained examiners are unable to reliably assess the accessory motion of some joints.[11,14,15]

Examiners may also err in their clinical decisions when they incorrectly perform a test that has been operationally defined adequately. Measurements of the variable of interest (eg, amount of motion, amount of force) could vary depending on whether the examiner performed the test correctly. Errors in hand placement, patient position, and magnitude and direction of force application would likely affect the results of a test.

Error Inherent in the Variable Being Assessed

The variable being assessed may vary randomly and may depend on the activity level of the patient, position of the patient, type of pathology present, level of alertness of the patient, and various other factors. Whether the patient reports pain during the test may also affect measurements. Variations in the severity of a patient's pain may also affect the inherent consistency of the variable being assessed. Research is needed to elucidate factors inherent to the patient that may lead to measurement error.

Error Attributable to the Examination Procedure

A common source of measurement error in MT appears to be the lack of operational definitions for some of the procedures applied to the patient. Operational definitions are instructions that guide the therapist in the process of obtaining a measurement. The instructions must specify how the measurement is taken and the variables that are being measured.5 If therapists are unclear about hand placement, patient position, instructions to the patient, direction or magnitude of force application, or the grading scale to be used, then measurement error is probably unavoidable. Therapists must be able to thoroughly describe the testing or treatment procedure to minimize error and miscommunication among clinicians.

Accessory Motion Tests

During an examination, the manual therapist commonly assesses the movement occurring between those joint surfaces that are thought to be relevant to the patient's problem. Assessments of joint surface movement are commonly used, in conjunction with other tests, to make clinical decisions. This section and those that follow review the literature that provides some of the basis for using AMTs in clinical practice.

Tests of accessory motion are designed to assess problems attributable to joints and the tissues that span the joints. Williams and Warwick[16] define accessory motion as all joint surface movements that cannot be performed actively in the absence of resistance. These authors claim two types of accessory motions occur. Type 1 movements cannot be performed voluntarily unless resistance to active movement is applied (eg, grasping an object). Type 2 movements can only be performed when the subject's muscles are completely relaxed. The accessory movements assessed by physical therapists appear to be the type 2 accessory movements, as defined by Williams and Warwick. For the purposes of this article, accessory motion is defined as movement occurring between joint surfaces that is produced by forces applied by the examiner. Accessory motion tests, therefore, are those examination procedures that are designed to assess some aspect of joint surface movement. The term "joint play" is also frequently used to describe the movement between joint surfaces. Mennell[17] defined joint play as the joint surface movements that are produced by forces applied by the examiner and that are necessary for normal functional movement. implicit in Mennell's definition is the assumption that the joint surface movements assessed by the examiner type 2 movements, as defined by Williams and Warwick) are the same as the movements that occur at the joint during patient-initiated movements (type 1 movements). Data supporting the assertion that examiners can replicate patient-initiated joint surface movements are necessary if Mennell's definition is to be clinically useful. No data exist to support the notion that examiners can replicate the motion between joint surfaces that occurs during voluntary movements. The term "joint play," therefore, will not be used in this article to describe the movements between joint surfaces.

Descriptions of the motions between joint surfaces should be distinguished from movements of entire bones. Movements of bones are classically described as osteokinematic or physiologic motion.[16] For example, the term "shoulder flexion" implies an osteokinematic motion in which the angle between the humerus and the scapula is increased in the sagittal plane. An example of an accessory motion is anterior glide of the humerus. The examiner imparts a force to the patient's shoulder such that the humeral head translates anteriorly on the glenoid cavity.

Therapists use AMTs to assess joints and the tissues that span joints primarily for four reasons. The first reason is to make inferences about joint function based on the amount of motion between joint surfaces. For example, Maitland[18] claims that joint surface movement assessments can be used to predict the cause of a limitation in a joint's physiological ROM. Others[9,13,19] claim that assessments of the amount of movement between joint surfaces can be used to infer that joint hypermobility is present.

The second basis for AMT assessments is to determine the type of resistance present at the end-range position. Cyriax[20] claims, for example, that judgments of the type of resistance present at an end-range position indicate which tissue or tissues may be causing a limitation in joint motion.

Third, some clinicians use AMTs to determine whether inflammatory processes are present in or around the joint.[21] If the patient's symptoms are reproduced during the procedure, the therapist may infer that one or more structures spanning the joint are inflamed.

The fourth reason some clinicians use AMTs is to test the integrity of ligaments around a joint. For example, the Lachman's test is an AMT designed to determine whether the anterior cruciate ligament (ACL) is injured.[22] Therapists typically assess the amount of movement, the end-feel, and whether pain was produced when making decisions about ligamentous integrity.[11] Hughston et al[23] recommend that a grading scale be used for assessing the severity of knee ligamentous damage. The scale requires the examiner to assess only the amount of motion present during the test. A mild instability is one that allows 5 mm or less of motion, a moderate instability has 5 to 10 mm of joint surface motion, and a severe instability permits greater than 10 mm of motion. Hughston et al did not describe whether these assessments were based on a comparison with the uninvolved side or whether the end-feel or the presence of pain during the test should be considered during AMTs of ligamentous integrity.

Numerous clinical experts[1,9,13,17] have described the AMT procedures they believe to be the most useful. Few studies, however, have examined the reliability or validity of assessments of patients' accessory motions. Therapists must therefore recognize that little is known about the reliability or validity of most AMT measurements. The remainder of this article critically discusses the literature that provides some of the basis for the construct validity of AMTs. The construct validity for a test provides the theoretical knowledge base for the meaningfulness of measurements. Studies examining the criterion-related validity and reliability of judgments made by use of AMTs are also reviewed. Studies of the criterion-related validity and reliability of AMT measurements are necessary to bridge the gap between theory and practice.

Construct Validity for Accessory Motion Tests

Physical therapists perform AMTs because they believe clinically meaningful inferences can be based on the data obtained. An assumption underlying the clinical use of assessments of joint surface motion is that the amount and type of movement at a given joint surface is measurable and predictable. Another commonly held assumption is that a specific AMT can be used to assess the joint surface motion that occurs during a specific physiological joint motion. For example, Wadsworth[24] implies that anterior glide of the head of the humerus on the glenoid cavity mimics the anterior slide that occurs when the patient laterally (externally) rotates the glenohumeral joint.

The theoretical model that serves as the basis for determining the types of accessory motions to assess at a given joint is based, in part, on the work of MacConaill.[25-28] MacConaill theorized that the type of movement occurring between joint surfaces is determined primarily by the shape of the joint surfaces. According to MacConaill, all joint surfaces can be classified as being either sellar (saddlelike) or ovoid (convex or concave) in shape. MacConaill implies that because all joints are either sellar or ovoid, the type of movement at a given joint surface can be predicted for each physiological motion.

These types of movement occurring at these joint surfaces have been conceptually classified by MacConaill: (1) spin, (2) slide, and (3) roll. Spin is defined as rotation of one surface on the opposing joint surface around a longitudinal axis. Slide occurs when the same point on one surface contacts new points on the opposing surface. Roll occurs when equidistant points on the joint surfaces contact one another. MacConaill's model for predicting the types of accessory motions to assess is intuitively logical. Some data exist to support the notion that the type of joint surface motion associated with a given physiological motion can be predicted based on joint surface shape.[29,30]

A theoretical construct for the meaningfulness of AMTs directly relates to whether measurements of a patient's accessory motion reflect the joint surface motion that occurs during voluntary movement. A distinction is made between joint surface motion produced by forces applied by the examiner (accessory motion) and joint surface motion produced during patient-initiated movements. The factors that determine the type and amount of joint surface motion during AMT are different from the factors that determine the type and amount joint surface motion during voluntary movements. For example, one important factor that determines the type and amount of accessory motion perceived by the examiner is the forces applied by the examiner. The magnitude and direction of these forces will have a significant influence on the examiner's perception of the amount of motion present. Another critical factor is the position of the patient's joint during testing. The ability of the soft tissue and joint structures to limit motion will vary, depending on the position of the patient's joint during the AMT.

The factors that account for joint surface motion during voluntary movements, however, are much more complex. Muscular, ligamentous, and other external forces probably interact to influence the direction and magnitude of joint surface motion. No studies have demonstrated that a therapist's assessment of the amount of accessory motion can be used to predict joint surface motion or osteokinematic motion produced during patient-initiated movement.

In addition to assessing joint surface motion, therapists commonly assess the nature of the resistance (end-feel) present at a joint's end-range position. Therapists attempt to determine the end-feel because they are interested in determining which tissue or tissues are limiting the motion. The issue of the meaningfulness of assessments of end-feel during AMTs is discussed later in this article.

Why therapists attempt to determine whether a patient reports pain during AMTs is fairly clear. If forces are applied to inflamed tissues in or around the joint, the expected response from the patient would be a report of increased pain. Some data exist to support the usefulness of assessments of the effect of AMTs on the pain reported by the patient.[15] This issue is discussed later in this article.

The construct validity for many of the inferences based on AMT results have not been adequately developed. Some clinicians, however, rely on AMTs to make many important clinical decisions. Therapists should recognize that the construct validity for AMT requires more development and research. Therapists should therefore use the results of AMTs only in conjunction with other more theoretically sound measurements when making clinical decisions. Manual therapists should consider using the Standards for Tests and Measurements in Physical Therapy Practice[4] to determine whether the measurements used in their clinical practices meet the requirements identified in the Standards.

Accessory Motion Testing Procedures--The Dependent Variables

The following discussion of the dependent variables (the measurements) obtained during AMT is not meant to be all-inclusive. Rather, the intent of this section is to discuss some of the more important issues the reader should be aware of when making clinical decisions based on AMTs.

Three of the more common assessments made during AMTs are judgments of the amount of motion present, the type of end-feel, and whether the patient perceived an increase in pain during the procedure. Some authors[1,9,13] report that assessments of the nature of the resistance perceived during the movement (prior to the end-feel) may also be useful. For example, Maitland[1] claims that the nature of the resistance perceived during the movement may indicate whether muscle spasm is present. Maitland provides no evidence to support this assertion.

Assessments of the Amount of Motion Present

A variety of different measurement scales have been proposed for judging the amount of motion present between joint surfaces. Maitland[19] has proposed that the amount of accessory motion found to be present in the segments of the spine be described as "hypomobile," "normal," or "hypermobile." This ordinal-scaled system is similar to Kaltenborn's scale[13] in that it requires the therapist to use an experiential model to make a judgment about the amount of motion present. Therapists must rely on previous clinical experience to determine whether the amount of motion present at a joint is normal or abnormal.

Hertling and Kessler[9] proposed a scale identical to that of Maitland[1] and suggest that this scale be used when assessing the accessory motion of extremity joints. The authors suggest the comparable contralateral joint be used as a basis of comparison when the contralateral joint is "healthy."

The scales described require the therapist to use previous clinical experience as the basis for classifying the amount of accessory motion. Because therapists' clinical experiences vary, an experientially based classification system would probably be interpreted differently by different therapists. A second problem with Hertling and Kessler's grading scale for the extremity joints is the issue of using the "healthy" side as a basis for comparison. There are no data to support the assertion that a peripheral joint's accessory motion is the same bilaterally. For example, the increased usage associated with a dominant extremity as compared with the nondominant side may produce an asymmetry of the accessory motion of the peripheral joints.

The issue of which grading scale should be used to quantify the amount of accessory motion present can only be resolved through sound clinical research. The literature currently does not support the use of one grading scale over another. Because it appears that some accessory motion grading scales have not been operationally defined adequately does not preclude the usefulness of AMT. In addition, because these measurements are typically used in conjunction with other tests to determine treatment, assessments of accessory motion are probably useful in some instances. Research that examines the reliability and validity of measurements of the amount of accessory motion present is needed, especially in the context of how these measurements are used in clinical practice.

Assessments of the Type of End-feel

A variety of classification systems also exist for describing the type of end-feel present during AMT. Cyriax[20] described six different types of end-feels for classifying the type of resistance perceived by the examiner during a physiological passive-range-of-motion test. The end-feels described by Cyriax have since been applied to AMTs. The categories of end-feels described by Cyriax are (1) bone to bone, (2) capsular, 3) empty, (4) spasm, (5) springy block, and (6) tissue approximation. Implicit in the classification system is the assumption that the examiner can determine the structure or structures that limit a joint's accessory motion. For example, the capsular end-feel is defined as the immediate stoppage of movement with some give. The examiner may conclude inappropriately that a taut joint capsule is the only structure that may produce this type of end-feel. No data exist to support the criterion-related validity of the use of the end-feel classification system of Cyriax for inferring the structures that limit a joint's motion.

The alternative to a classification system such as that described by Cyriax[20] is to use well-defined descriptive terms that do not implicate specific structures. Kaltenborn[13] Suggested the use of the terms "soft," "firm," and "hard" to classify the type of end-feel. In his definitions, however, Kaltenborn identifies the tissues he believes are responsible for producing the end-feels. For example, Kaltenborn claims a firm end-feel is present when capsular or ligamentous tissues become taut.

Torg et al[24] recommend classifying the end-feel as either hard or soft when using the Lachman's test to assess the integrity of the ACL. A hard end-feel was defined as a firm sensation with a definite stopping point, and a soft end-feel represented the absence of both a firm sensation and a definite stopping point. The advantage of the definitions for the terms used by Torg et al is that they describe the perceptions of the examiner rather than labeling the tissues that may be producing the end-feel.

I would propose the following definitions for the terms proposed by Kaltenborn[13]: (1) soft end-feel--a gradual increase in resistance at the end-range of motion, (2) firm end-feel--an abrupt increase in resistance at the end-range of motion, and (3) hard end-feel--an immediate stoppage to movement at the end-range of motion. The advantage of this proposed classification system is that neither the terms nor the definitions imply that a specific tissue is producing the end-feel, In addition, I believe the definitions are more operational and therefore are more likely to be reliable. Further research is needed to test this assertion.

Assessments of the Effect of the AMT on Pain Reported by the Patient

One of the more common inferences made based on the use of AMTs is that an inflammatory response is present when the patient reports an increase in pain during the test. Some evidence exists to suggest some AMTs of the sacroiliac joint provide somewhat reliable data about the presence of pain.[15] The study provided no data to support the contention that inflammatory processes were present in the patients with positive test results. Because the symptoms reported by the patient were increased during the AMTs, however, one could argue that inflammatory processes were likely present. More study is needed to determine whether AMTs for other joints provide reliable data about the presence of pain.

Review of the Literature

Accessory Motion Testing Procedures for Spinal Joints

Only one study[12] was found that examined the validity of AMTs of the spine. The authors determined whether AMTs for the cervical spine could be used to identify whether a zygapophyseal joint(s) was responsible for the symptoms reported in a group of patients with neck and headache pain. The criterion used to test the inference was a nerve block administered to the medial branches of the dorsal rami, which reportedly innervated the zygapophyseal joints under study. If the nerve block administered to the involved joint identified by the therapist relieved the patient's symptoms, the judgment that the zygapophyseal joint was inflamed, which was based on the AMT, was determined to be valid.

The 20 subjects admitted to the study[12] reported pain in the neck region for at least 12 months. The examiner was a physical therapist and was also the primary author. All subjects underwent nerve blocks 1 to 4 weeks prior to the examination by the therapist or immediately after the therapist's examination. A positive nerve block was defined as one that produced complete relief of symptoms for at least 3 hours. All nerve blocks were done during a fluoroscopic procedure used to determine the placement of the needle.

The therapist examined several accessory motions and physiological motions of the cervical spine for each patient admitted to the study.[12] The criteria used to identify an involved joint were an abnormal end-feel, an abnormal quality of resistance to motion, and a reproduction of the patient's complaint of pain. All three criteria were necessary to make a judgment of a positive test result. The authors did not define what they meant by abnormal end-feel or abnormal quality of resistance to motion. Interestingly, the authors apparently did not include assessments of the amount of accessory motion present in the criteria for judging that a test was positive.

The therapist was reported to be able to correctly identify the location of the involved joints in all instances.[12] The authors made an argument in their discussion that the nerve blocks were valid criteria for determining the location of an involved zygapophyseal joint. They claimed that the nerve block technique used minimized migration of the anesthetic and produced an anesthesia of the injected joint. The authors, however, acknowledged that the injection may have anesthetized the muscles innervated by the injected nerve. Some of the subjects may have reported pain relief following the block for reasons other than zygapophyseal joint involvement.

The study of Jull et al[12] provides some data to suggest that a therapist is capable of identifying inflamed zygapophyseal joints in the cervical spine. Only one therapist, however, participated in this study. Therefore, the results may not be generalizable beyond the therapist who performed the AMTs.

Matyas and Bach[31] thoroughly summarized the literature and much unpublished work that has examined the reliability of measurements obtained by using many different spinal MT procedures. This well-written article would likely be useful to many manual therapists and may aid the clinician in identifying clinical research questions in MT. Unfortunately, most of the studies that examined the reliability of AMT measurements of the amount of movement, end-feel, and whether pain was produced were taken from unpublished sources.

Perhaps the most frequently referenced article that has examined the reliability of assessments of accessory motions of the spine is the study of Gonnella et al.[32] The authors examined the intratester and intertester reliability for assessments of the accessory motion present during a variety of AMTs designed for the lumbar spine. A total of five experienced therapists participated in the study. The authors used a modified version of the scale described by Kaltenborn[13] to classify the accessory motion. The scale was modified to allow therapists to add a plus or minus ranking to four of the seven grades described by Kaltenborn. The subjects were five physical therapy students without a history of low back pain.

To assess intratester reliability, the therapists were blindfolded when they performed repeated examinations on the same patient.32 Operational definitions of the AMTs tested were not provided. The authors concluded that intratester reliability was acceptable, but that intertester reliability was unacceptable. The authors reported only the means and standard deviations of the measurements, which is not appropriate given the ordinal-level data. Perhaps more importantly, the study examined a small number of healthy subjects in a situation that did not simulate clinical practice. Because the study examined healthy subjects in an artificial setting, the results of this study should not be generalized to a patient population or to clinical practice.

Kaltenborn and Lindahl[33] examined the intertester reliability of assessments of the accessory motion present in several spinal joints of four healthy subjects. The assessments obtained by Kaltenborn were compared with those obtained by the other nine physical therapist examiners participating in the study. The judgments made by Kaltenborn were the criterion by which the other examiners' measurements were judged. The authors concluded that reliability of the AMTs studied was good. The design of the study of Kaltenborn and Lindahl[33] does not mimic clinical practice, which suggests that the results of the study are probably not generalizable to the clinical setting. The authors also provided no data to support the contention that the Kaltenborn's measurements were valid for any inference.

Potter and Rothstein[15] examined the intertester reliability of assessments of the accessory motion of the sacroiliac joint. Eight experienced therapists from two clinics were randomly paired and examined 17 patients who had unilateral buttock pain. Reliability was assessed for 13 tests of the sacroiliac joint, 2 of which were AMTs. The 2 AMTs were the supine iliac gapping test and the side-lying iliac compression test. Both tests require the therapist to apply forces to the sacroiliac joints by applying forces to the iliac crests. A positive test result was judged to be present when the patient reported an increase in pain during the test. The authors operationally defined each of the tests and the criteria for determining when a test result was judged to be positive.

Limitations of this study[15] were that a relatively small number of patients were examined and that no reliability coefficients were calculated. The study, however, has important implications for clinical practice. The authors found that the two AMTs that showed some evidence of reliability were the tests that required the patient to report whether pain was increased during the procedure. These two tests should probably be done on patients suspected of having sacroiliac joint pathology.

The chiropractic and osteopathic literature contains reports of several studies that have examined the reliability of assessments made using AMTs designed for use on the lumbar spine.[34-40] This literature is briefly reviewed because these studies may be of use to therapists interested in studying the reliability of judgments made based on AMTs. The generalizability of these studies to physical therapists is probably limited because of differences in training between professions. Only one study,[39] which assessed the reliability of AMT judgments made on patients and which used tests that are similar to those used by physical therapists, is reviewed.

Keating et al[39] examined the intertester reliability of a variety of AMT assessments designed for use on the lumbar spine. Judgments were made by three chiropractors who tested 21 subjects who had low back pain. No other description of the subjects was provided.

The spinous processes of the subjects were marked by an examiner prior to participation.[39] These same marks were then used by all three examiners for each subject. The examiner performed several well-defined AMTs and a variety of other procedures for each patient. Examiners were not allowed to observe each other while taking measurements.

The authors39 reported Kappa coefficients for the assessments of whether pain was produced and whether fixations were present during AMT. The Kappa is a reliability index that describes the degree of agreement for categorical data (such as the presence of pain) while eliminating agreement attributable to chance.[41] They operationally defined a fixation as being present when no accessory motion and a hard end-feel was present. A fixation was judged to be absent when a small amount of motion and a springing end-feel was present. The authors did not define what they mean by a springing end-feel.

Kappas ranged from .22 to .65 for assessments of the presence of pain and from .00 to .23 for assessments of the amount of accessory motion present.39 This study suggests that assessments of whether pain is present during an AMT tend to be more reliable than assessments of the amount of motion or type of end-feel. These findings are consistent with the findings of Potter and Rothstein.[15]

Few studies have examined the validity and reliability of assessments based on AMTs designed for use on the spine. Little is known about the validity of AMTs for most inferences. The reliability of judgments based on AMTs performed on spinal joints has not been adequately studied. Some evidence exists to suggest that assessments of the effect of some sacroiliac joint AMTs on pain symptoms may be reliable.

Accessory Motion Testing Procedures for Extremity Joints

Only two studies that examined the reliability of assessments of the accessory motion present in extremity joints of patients were reported in the literature.[11,14] These two studies examined the reliability of measurements from two commonly used tests of the ligamentous integrity of the knee. Tests of ligamentous integrity are slightly different than other AMTs. Ligamentous tests are done for a well-defined purpose (ie, to determine whether a specific ligament is injured), whereas other AMTs are done for more general purposes. Most ligamentous tests, however, require therapists to make judgments of the amount of motion present, the end-feel, and whether pain was produced. It could therefore be argued that ligamentous tests are reflective of other AMTs.

Cooperman et al[11] examined the validity and reliability of judgments made using the Lachman's test, an AMT designed to assess the ACL. They judged whether the test yielded positive or negative results, the end-feel, and the amount of translation. The examiners were two physical therapists and two orthopedic surgeons, all with several years of clinical experience. Examiners used their own AMT techniques. Thirty-two patients with knee problems, 13 of whom had a documented ACL tear, participated in the study.

The predictive value for a positive test result was 47% for all examiners, whereas the predictive value for a negative test result was 70%.[11] The Kappa values for intratester and intertester reliability for judgments of whether the test result was positive or negative ranged from .02 to .69. Kappa values for judgments of the type of end-feel present (soft or hard) and the amount of motion present (0, 1+, 2+, 3+) ranged from .22 to .46.

The results suggest that assessments of whether the test result was positive or negative, the end-feel, and the amount of motion present were, for the most part, unreliable.[11] Judgments of a negative Lachman's test result appear to be fairly useful for predicting whether a patient has an intact ACL. Judgments of a positive Lachman's test result are not as useful for predicting whether a patient has an injured ACL. The study has some limitations because only two physical therapist examiners took part in the study and only patients with chronic injuries were assessed.

McClure et al[14] examined the intertester reliability of judgments of medial knee ligament integrity in patients whose medical history suggested the need for assessment of the medial collateral ligament. Three therapists with extensive clinical experience in treating patients with knee problems served as examiners. The therapists were allowed to use their own techniques in order to mimic clinic practice. Therapists were randomly paired to apply the valgus stress test in the 0- and 30-degree positions to each of the 50 subjects admitted to the study.

The Kappa coefficient was used to describe the degree of agreement for assessments of the end-feel, the amount of motion, and whether pain was perceived by the subject during the test.[14] The authors found that the Kappas were low for the tests done in both the 0-degree position K=.06-.40) and the 30-degree position K=.16-.38). In addition, the percentage of agreement was 72% or less for all assessments except end-feel at 0 and 30 degrees.

The generalizability of the study is limited because only patients with chronic knee injuries were studied.[14] The results might have been different had patients with acute knee injuries been examined. In addition, the skewed distribution of assessments made during end-feel judgments probably artificially deflated the Kappa coefficients for these measurements. Because the sample apparently had a low prevalence of patients with medial collateral ligament injuries, the Kappas may have been artificially deflated.[42] The relatively low percentages of agreement for most measurements, however, suggest that the measurements were still somewhat unreliable.

Despite the limitations, these two studies[11,14] appear to have important implications to other AMTs. The valgus stress test and the Lachman's test of the knee are probably two of the more commonly performed AMTs. Because the knee is one of the largest and more superficial joints, it could also be argued that these tests are two of the simpler AMTs to administer. Less commonly administered tests done on smaller and less accessible joints are likely to also have questionable reliability, but research is needed.

Only two studies that examined the reliability of AMTs performed on patients with problems attributable to extremity joints were reported in the MT literature.[11,14] Results of these studies bring into question the usefulness of AMTs done on patients with extremity joint problems.

Summary

Perhaps the greatest research need in MT is thorough descriptions of the methods manual therapists use to make clinical decisions. Measurements from many clinical tests and other data are collected by the manual therapist during an examination. Case studies describing the processes used to arrive at clinical decisions will provide clinical researchers with the information necessary to design generalizable effectiveness studies.

Many studies have been published in the MT literature that have examined the reliability of measurements taken during one or a few selected clinical tests. Although these studies provide the clinician with some insights into the usefulness of these procedures, the studies have limited usefulness. Many times manual therapists appear to make decisions based on a battery of tests rather than a few tests in isolation.

Studies examining the usefulness of MT procedures in a more relevant clinical context are also needed. Studies are also needed that examine the criterion-related validity of MT measurement procedures. The predictive value of positive and negative test results as well as the sensitivity and specificity of the many tests manual therapists use should be studied.

The theoretical arguments that provide the basis for the inferences made based on AMT assessments appear to be weak. Research is needed that is directed toward the anatomical and mechanical bases of AMTs. Research must also be directed toward assessing the reliability of assessments made using spinal and peripheral joint AMTs. Studies examining the reliability of AMT assessments obtained in a clinical context are essential because these studies are prerequisite to meaningful MT effectiveness studies that require the use of AMTs.

Appendix. Kaltenborn's Grading Scale for Describing the Amount of Accessory Motion Present at a joint[13] 0=No movement (ankylosis) 1=Considerable decreased movement 2=Slight decreased movement 3=Normal 4=Slight increased movement 5=Considerable increased movement 6=Complete instability

Several study design issues were also discussed in this article and should be addressed in future studies.

Acknowledgments

I wish to thank G Kelley Fitzgerald, PT, Janet Kues, PT, Robert L Lamb, PhD, PT, and Scott Sullivan, PT, for their helpful comments during the preparation of this manuscript.

References

[1] Maitland GD. Peripheral Manipulation. 2nd ed. Boston, Mass: Butterworth; 1977. [2] McKenzie RA. The Lumbar Spine: Mechanical Diagnosis and Therapy. Waikanae, New Zealand: Spinal Publications; 1981. [3] Cibulka MT, Delitto A, Koldehoff RM. Changes in innominate tilt after manipulation of the sacroiliac joint in patients with low back pain: an experimental study. Phys Ther. 1988; 68:1359-1363. [4] Task Force on Standards for Measurement in Physical Therapy. Standards for tests and measurements in physical therapy practice. Phys Ther. 1991;71:589-622. [5] Kerlinger FN. Foundations of Behavioral Research. 3rd ed. New York, NY: Harcourt Brace Jovanovich Inc; 1986. [6] Rothstein JM. Measurement in Physical Therapy. New York, NY: Churchill Livingstone Inc; 1985. [7] Clinical Decision Making. Alexandria, Va: American Physical Therapy Association; 1989. [8] Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2nd ed. Boston, Mass: Little, Brown and Co Inc; 1991. [9] Hertling D, Kessler RM. Management of Common Musculoskeletal Disorders. 2nd ed. Philadelphia, Pa: JB Lippincott Co; 1990, [10] Bogduk N. The reliability of diagnosis. New Zealand Journal of Physiotherapy. 1986;14:25-27. [11] Cooperman JM, Riddle DL, Rothstein JM. Reliability and validity of judgments of the integrity of the anterior cruciate ligament of the knee using the Lachman's test. Phys Ther. 1990; 70:225-233. [12] Jull G, Bogduk N, Marsland A. The accuracy of manual diagnosis for cervical zygapophysial joint pain syndromes. Med J Aust. 1988; 148:233-236. [13] Kaltenborn FM, Evjenth O. Manual Mobilization of the Extremity Joints Basic Examination and Treatment Techniques. 4th ed. Oslo, Norway: Olaf Norlis Bokhandel Universitetsgaten; 1989. [14] McClure PW, Rothstein JM, Riddle DL. Intertester reliability of clinical judgments of medial knee ligament integrity. Phys Ther. 1989; 69:268-275. [15] Potter NA, Rothstein JM. Intertester reliability for selected clinical tests of the sacroiliac joint. Phys Ther. 1985;65:1671-1675. [16] Williams PL, Warwick R. Gray's Anatomy. 36th ed. Philadelphia, Pa: WB Saunders Co; 1980. [17] Mennell JM. Joint Pain. Boston, Mass: Little, Brown & Co Inc; 1964. [18] Maitland GD. Passive movement techniques for intra-articular and periarticular disorders. Australian Journal of Physiotherapy. 1985;31:3-8. [19] Maitland GD. Vertebral Manipulation. 5th ed. Boston, Mass: Butterworth; 1986. [20] Cyriax JH. Textbook of Orthopaedic Medicine, Volume I: Diagnosis of Soft Tissue Lesions. 7th ed. London, England: Bailliere Tindall; 1978. [21] Grieve GP. Common Vertebral Joint Problems. New York, NY: Churchill Livingstone Inc; 1981. [22] Torg JS, Conrad W, Kalen V. Clinical diagnosis of anterior cruciate ligament instability in the athlete. Am J Sports Med. 1976;4:84-93. [23] Hughston JC, Andrews JR, Cross MJ, Moschi A. Classification of knee ligament instabilities, part 1: the medial compartment and cruciate ligaments. J Bone Joint Surg [AM]. 1976;58: 159-172. [24] Wadsworth CT. Manual Examination and Treatment of the Spine and Extremities. Baltimore, Md: Williams & Wilkins; 1988. [25] MacConaill MA. The movements of bones and joints, V: the significance of shape. J Bone Joint Surg [Br]. 1953;35:290-297. [26] MacConaill MA. The geometry and algebra of articular kinematics. Bio-Med Eng. 1966; 1: 205-212. [27] MacConaill MA, Basmajian JV. Muscles and Movements A Basis for Human Kinesiology. Baltimore, Md: Williams & Wilkins; 1969. [28] MacConaill MA. Joint movement. Physiotherapy. 1964;50:359-367. [29] Frankel VH, Burstein AH, Brooks DB. Biomechanics of internal derangement of the knee: pathomechanics as determined by analysis of the instant centers of motion. J Bone Joint Surg [AM]. 1971;53:945-962. [30] Sammarco GJ, Burstein AH, Frankel VH. Biomechanics of the ankle: a kinematic study. Orthop Clin North Am. 1973;4:75-96. [31] Matyas TA, Bach TM. Reliability of selected techniques in clinical arthrometrics. Australian Journal of Physiotherapy 1985;31:175-199. [32] Gonnella C, Paris SV, Kutner M. Reliability in evaluating passive intervertebral motion. Phys Ther. 1982;62:436-444. [33] Kaltenborn FM, Lindahl O. Reproducibility of the results of manual mobility testing of specific intervertebral segments. Swed Med J. 1969;66:962-965. [34] Nansel DD, Peneff AL, Jansen RD, et al. Interexaminer concordance in detecting joint play asymmetries in the cervical spines of otherwise asymptomatic subjects. J Manipulative Physiol Ther. 1989;12:428-433. [35] Johnston WL, Allan BR, Hendra JL, et al. Interexaminer study of palpation in detecting location of spinal segmental dysfunction. J Am Osteopath Assoc. 1983;82:839-845. [36] Love RM, Brodeur RR. Inter- and Intra-examiner reliability of motion palpation for the thoracolumbar spine. J Manipulative Physiol Ther. 1987;10:1-4. [37] Herzog W, Read LJ, Conway JW, et al. Reliability of motion palpation procedures to detect sacroiliac joint fixations. J Manipulative Physiol Ther. 1989;12:86-92. [38] Cassidy JD, Potter GE. Motion examination of the lumbar spine. J Manipulative Physiol Ther. 1979;2:151-158. [39] Keating JC, Bergmann TF, Jacobs GE, et al. Interexaminer reliability of eight evaluative dimensions of lumbar segmental abnormality. J Manipulative Physiol Ther. 1990; 13:463-470. [40] Youngquist MW, Fuhr AW, Osterbauer PJ. Interexaminer reliability of an isolation test for the identification of upper cervical subluxation. J Manipulative Physiol Ther. 1989; 12:93-97. [41] Cohen J. A coefficient of agreement for nominal scales. Educational and Psychological Measurement. 1960;20:37-46. [42] Spitznagel EL, Helzer JE. A proposed solution to the base rate problem in the Kappa statistic. Arch Gen Psychiatry. 1985;42:725-728.