An examination of Cyriax's passive motion tests with patients having osteoarthritis of the knee.

The scheme of selective tension testing proposed by Cyriax[1] is a clinical system of diagnosis of painful problems of soft tissues. An anatomical definition of the lesion is based on the patient's response to the application of force (which Cyriax called "tension") in different ways. The diagnosis is rendered based on the patient's report of pain and the amount and direction of available movement.[1(p43)] Physical therapists have adopted this system to determine the cause of patient complaints of pain. The validity of the scheme is grounded in theory and extensive clinical observation, but it has not been studied objectively or empirically.

According to Cyriax,[1] testing is conducted in four ways: active motion, passive motion, resisted contractions, and palpation. The procedures are usually performed in that order. Active motion is designed to assess the patient's willingness to move and his or her range of motion (ROM) and strength. Passive motion is used to assess the amount of motion available and the direction of limitation, if any; the palpable sensation at the end of passive motion (end-feel); and the temporal sequence of pain reported by the patient and resistance felt by the examiner during end-feel testing (pain/resistance sequence). Resisted testing is used to determine the reaction of the muscle, tendon, and bony attachments to contraction. Palpation is used last to confirm the involvement of the structure or structures suggested by the previous portions of the test. A summary of the passive motion component of the system is shown in the Figure, and a full description of the entire system of diagnosis is available in Cyriax's book.[1]

This report addresses only the passive motion part of the examination. The three components of passive motion testing were designed to be used to diagnose a condition based on its pathophysiology. Each of the components is supposed to give additional information. The amount and direction of limitation of motion are examined to determine the presence or absence of a capsular pattern. A capsular pattern is a joint-specific pattern of restriction that indicates involvement of the entire joint capsule.[1(p54)] A noncapsular pattern deviates from the specific pattern and can indicate the presence of ligamentous or partial capsular adhesions, extra-articular involvement, or internal derangements.[1(p57)] The type of end-feel purportedly indicates the anatomical structures that limit passive motion (eg, bone, capsule, muscle contraction, loose bodies in the joint, other parts of the body) or the patient's unwillingness to complete the motion.[1(p53)] The pain/resistance sequence is assessed to guide the vigor of treatment[1(p54)] and is often interpreted as an indicator of the chronicity of inflammation (active, less active, none). According to Cyriax, pain before resistance is felt by the examiner suggests a lesion with active inflammation; pain that he says occurs synchronous with resistance suggests a lesion with less active inflammation, whereas pain after resistance suggests a lesion without inflammation.

The assessment system is designed to differentiate causes of pain stemming from inert structures (capsule, ligament, fascia, bursa, nerve root, dura mater) or contractile structures (muscle, tendon, bony insertions) but is not sufficient for a definitive diagnosis. Other clinical and radiographic tests are necessary to diagnose and discriminate problems arising from tissues such as bone or cartilage or neoplastic disease. Cyriax[1] claims the system can be used to identify patients having osteoarthritis (OA), even though the disease primarily involves articular cartilage. A task force of the American Rheumatism Association defined osteoarthritis as a

... heterogeneous group of conditions that lead to joint symptoms and signs which are associated with defective integrity of the articular cartilage, in addition to related changes in the underlying bone and at the joint margins. Although articular cartilage is poorly innervated and defects in cartilage are not, in themselves, symptomatic, a clinical syndrome of symptoms, which often includes pain, may evolve from these defects.[2(p1039)]

According to Cyriax, as the disease develops and progresses, the capsule and other structures surrounding the joint become involved in predictable ways.[1(p406)]

Cyriax suggested that in knee OA passive motion is restricted in a capsular pattern, with proportionally greater restriction in flexion than in extension.[1(p56)] He contended that a 5-to 10-degree extension loss corresponds to a 90-degree flexion loss (extension loss is 6%-11% of flexion loss).[1(p56)] He suggested that early in the development of the disease, motions end with involuntary muscle contraction (spasm end-feel).[1(pp7,52)] As the disease advances, patients develop capsular end-feels[1(p52)] or hard and painless end-feels in both extension and flexion, purportedly arising from bone hitting bone.[1(pp52,406)] If a loose body were in the joint, a springy block might be anticipated.

End-feel is related to joint motion or pain intensity. For example, tissue approximation is the expected end-feel for knee flexion when the knee has full ROM. Flexion is expected to become limited early in OA, and the flexion end-feel would be expected to become a capsular end-feel as motion is lost. Similarly, patients may be classified as having spasm and empty end-feels, because these types of end-feels are painful during motion.

Cyriax stated that patients with OA are often pain-free,[1(p11)] but pain could stem from impacted loose bodies in joints or from subchondral bone after the cartilage is severely eroded.[1(p406)] Because OA is a condition of a poorly innervated structure and leads to decreased elasticity of the periarticular structures over an extended period of time, most patients would be expected to demonstrate either a painless end-feel or pain developing after the examiner feels resistance. Resisted testing would be strong and painless because muscles are not involved in the disease.[1(p50)] Cyriax claimed that pain could not arise from articular cartilage compression resulting from the contraction because articular cartilage is not innervated.[1(pp11,50)] In addition, compression would relax the ligaments and capsule rather than stressing them. Palpation would reveal osteophytes, coarse crepitus or creaking, and no warmth.[1(pp53,406)]

The primary purpose of our study was to begin the examination of the construct validity of the Cyriax system of soft tissue diagnosis. The process of construct validation of a measure is, by definition, theory dependent. The extent to which a measure performs within a theoretical framework provides evidence for the validity of the underlying construct that is measured by the variable. Many methods are used to examine construct validity of a measure. For example, evidence for validity begins to accumulate if data show that the measure discriminates among groups with and without the attribute being measured, correlates across multiple methods of measuring the same construct, or supports hypotheses incorporating the construct being measured.[3]

In our study, we examined the construct validity of the passive motion portion of the system of selective tension testing from two perspectives. First, we compared the theoretically expected pattern of restriction, end-feel, and pain/resistance sequence with the actual assessments of patients with OA of the knee. The hypotheses were (1) a significant proportion of subjects with OA will demonstrate a capsular pattern (H:1), (2) a significant proportion of subjects with OA will have capsular end-feels for both extension and flexion (H:2), and (3) significantly more subjects with OA will have painless end-feels or pain after resistance than subjects who have pain with resistance or pain before resistance (H:3).

Second, we examined relationships among the components of passive motion testing and joint motion, pain intensity, and chronicity. We hypothesized (1) that subjects with tissue approximation end-feels for knee flexion will have significantly more passive ROM than subjects with spasm and capsular end-feels (H:4), (2) that subjects with spasm or empty end-feels will have significantly higher pain intensity than subjects with other end-feels (H:5), (3) that the pain/resistance sequence win correlate positively with pain intensity (H:6), and (4) that the pain/resistance sequence will correlate positively with chronicity (H:7).

A second purpose of the study was to estimate the reliability of the data generated by each of the components of the passive motion portion of the system. The hypotheses for this portion of the study were (1) there will be no significant differences in passive ROM, end-feel, and pain/resistance sequence between sets of measurements (H:8) and (2) test-retest reliability estimates will exceed .80 for passive ROM (intraclass correlation coefficient [ICC]), end-feel assessments kappa), and pain/resistance sequence (kappa) (H:9).

Method

Subjects

Subjects for the study were 79 patients with OA of the knee who had consented to be screened for a study of the effectiveness of ultrasound on chronic soft tissue tightness.[4] Their OA was diagnosed by radiography or clinical examination by physicians. Among the important criteria for a clinical diagnosis of OA are the presence of osteophytes, morning stiffness for less than 30 minutes, crepitus, bony enlargement, and age.[2] The characteristics of the 20 male and 59 female patients are shown in Table 1. The subjects had a mean age of 68.5 years, an average height of 166.6 cm (65.6 in), and an average weight of 81.1 kg (179.2 lb). Subjects reported feeling stiffness in their knees from the disease for an average of 7 years. On the day previous to screening, subjects had pain in their knees averaging 5.6 cm on a 10-cm visual analogue scale (VAS). All subjects had at least a 10-degree limitation in passive flexion and/or extension ROM.

Table 1. Characteristics of Subjects With Osteoarthritis of the Knee (N = 79)

                                   X        SD       Range

Age (y)                             68.5     13.3     28.0-95.0
Duration of knee stiffness (mo)     83.6    122.4      1.0-612.0
Knee pain (cm)(a)                    5.6      3.1      0.0-10.0
Weight (kg)                         81.1     18.6     49.8-124.9
Height (cm)                        166.6      9.9    149.9-193.0

(a) pain measured by a 10-cm visual analog scale.

Examiners

Four examiners participated in the study. The examiners had practiced physical therapy for 4 to 18 years. All examiners were familiar with the evaluation techniques from their professional and postprofessional education, and they met with each other and the principal investigator (KWH) to review the measurement techniques, specific study procedures, and grading prior to their participation in the study. Each examiner performed all measures on the same set of patients at baseline, after treatment, and after 2 months without active intervention. Evaluators did not have access to previous evaluations.

Procedure

Passive ROM of the knee was measured with a large universal goniometer with the subjects in the supine position with the hip flexed to 90 degrees. According to Cyriax, in OA extension loss is 6% to 11% of flexion loss.[1](p56) In our study, therefore, a capsular pattern was defined as extension loss (with full extension defined as 0[degree]) being [less than or equal to]11% of the flexion loss (with full flexion defined as 150[degrees] to accommodate the maximum flexion ROM of all subjects and to avoid negative loss values). Extension losses greater than 11% of flexion loss were defined as representing a noncapsular pattern. End-feel was assessed at each end of passive ROM using overpressure and assigned to one of six categories. The pain/resistance sequence was also assessed at each end of passive ROM and graded on a four-point scale. These scales are shown in Table 2.

Table 2. categories of End-feel Testing and Pain/Resistance Sequence Used in
the Study[1](p53)

End-feel                     Description

Capsular                   A hardish arrest of motion, with some
                             give to it, feeling like
                             leather being stretched or as if two
                             pieces of tough rubber
                             were being squeezed together
Tissue approximation       Motion ends with a sensation suggesting
                             that motion could
                             continue if not stopped by one body
                             part contacting another
Springy block              Noticeable rebound is seen and/or felt
                             at end of motion
Bony                       An abrupt halt to movement as when two
                             hard surfaces meet
Spasm                      A vibrant twang suggesting that muscles
                             have actively or
                             reflexively acted to end motion
Empty                      Pain occurs before the end of motion
                             and patient asks for the
                             motion to stop; examiner feels no
                             resistance
Pain/resistance sequence
  1                        No pain
  2                        Pain occurs after resistance is felt by
                             the examiner
  3                        Pain occurs at the same time that
                             resistance is felt by the
                             examiner
  4                        Pain occurs before resistance is felt
                             by the examiner

The pain/resistance sequence scale was used in three ways. When it was studied as an indicator of OA, subject with no pain and subjects with pain after resistance were combined into one category, and subjects who had pain with resistance and pain before resistance were combined into one category. When the pain/resistance scale was used as a variable for examining the pain relationships, it was considered a four-point scale as described. When the pain/resistance scale was used for analysis of the concept of chronicity, subjects without pain on end-feel testing were dropped from the analysis. Pain in OA does not correlate with stage of disease activity.[5] Patients with early disease may be pain-free, as may patients with very advanced disease.[5] The inclusion of a "no pain" category would abrogate the ordinal nature of the scale as a measure of chronicity.

Pain intensity was measured by asking subjects to mark a VAS(6) representing their pain intensity on the previous day. Chronicity was measured by subject report of the number of months they had felt stiffness in their knees resulting from their disease.

Table 3. Extension and Flexion Ranges of Passive Motion (in Degrees)

Motion and
Time of Measurement     X       SD        Range       n

Extension
  Baseline(a)           9.77    5.68    0.00-30.00    79
  Posttreatment(b)      7.05    5.34    0.00-21.00    61
  Follow-up(b)          7.46    6.76    0.00-30.00    52
Flexion
  Baseline(a)         120.56   18.72   50.00-142.00   78
  Posttreatment(b)    124.25   17.66   48.00-145.00   61
  Follow-up(b)        122.35   18.45   54.00-152.00   52

(a) Used for validity analyses.
(b) Used for reliability analyses.

Test-retest reliability of the passive ROM measurements was estimated using a subset of 52 patients in the ultrasound study who had afl three measurements taken. The data from the posttreatment and follow-up measurement sessions were used for analysis. Although the 2-month interval between measurements is long, subjects received no active intervention during that period. Based on reports from the subjects, nearly afl had continued to do an assigned home exercise program during this period and to be as active as they had been at the end of treatment. Because the condition had been present for a very long time in most of the subjects, we did not expect that passive ROM, end-feel, and pain/resistance sequence would change markedly over 2 months. We acknowledge, however, that change could have occurred in these subjects and consider our reliability estimates as containing this source of error.

The reliability and validity of the VAS and chronicity data were not tested. The VAS has been reported to have test-retest reliability (reported as Pearson correlation coefficients) ranging from .91 to .97[7,8] and correlations (r) ranging from .60 to .90[6,8] with other measures of pain intensity. Chronicity data were gathered by patient self-report. Although no reliability and validity data are available for this particular measure, the reliability of patient reports of other variables, such as activities of daily living, is acceptable, and patient reports correlate very highly with other methods of gathering the same information, such as on-site observation.[9,10]

Data Analysis

One-way chi-square analyses were used to test the first set of hypotheses pertaining to the proportion of subjects with capsular patterns (H:1), capsular end-feels for both extension and flexion (H:2), and painless end-feels or pain after resistance (H:3) at baseline. The hypotheses that the passive ROM of subjects with tissue approximation end-feels would be larger than the passive ROM of subjects with spasm or capsular end-feels (H:4) and that the pain intensity of subjects with spasm or empty end-feels would be greater than the pain intensity of subjects with other end-feels (H:5) were tested with the Kruskal-Wallis analysis of variance (ANOVA) with multiple post hoc pair-wise comparisons.[11] To examine the relationship between the baseline measures of pain/resistance sequence and pain intensity (H:6) or chronicity (H:7), Spearman rank correlation coefficients (rho) were calculated.

The differences between passive extension and flexion ROM measurements on the two occasions and test-retest reliability were analyzed individually for three evaluators (one evaluator had only five subjects, and the ICC was unstable) with the ANOVA for repeated measures and the ICC (3,1).[12] The ICC (3,1) was chosen to estimate the reliability of the specific data of each examiner, assuming a single measurement. The differences between measurements of end-feel and pain/resistance sequence on the two occasions were analyzed with the Wilcoxon Matched Pairs Test, and reliability was analyzed with Cohen's kappa coefficients.[13] The alpha level for afl analyses was set at .05. All analyses were performed on a personal computer(*) using the SPSS statistical package.(+)

Results

The descriptive statistics for passive extension and flexion ROM are displayed in Table 3. At baseline, only 8 subjects displayed a capsular pattern and 71 subjects displayed a noncapsular pattern. The frequencies of capsular and noncapsular patterns were significantly different ([chi][sup.2] = 50.24, P <.001), but the hypothesis that a significant proportion of subjects would have a capsular pattern (H:1) was not supported because the results were in the wrong direction.

Table 3. Extension and Flexion Ranges of Passive Motion (in Degrees)

Motion and
Time of Measurement         x       SD       Range       n

Extension
  Baseline[a]              9.77    5.68    0.00-30.00    79
  Posttreatment[b]         7.05    5.34    0.00-21.00    61
  Follow-up[b]             7.46    6.76    0.00-30.00    52
Flexion
  Baseline[a]            120.56   18.72   50.00-142.00   78
  Posttreatment[b]       124.25   17.66   48.00-145.00   61
  Follow-up.sup[b]       122.35   18.45   54.00-152.00   52

[a]Used for validity analyses.
[b]Used for reliability analyses.

The number of subjects demonstrating each type of end-feel is shown in Table 4. The differences in number of subjects with each type of end-feel were significant at baseline for both extension ([chi][sup.2]=193.43, P<.001) and flexion ([chi][sup.2]=80.31, P <.001). Most of the subjects had a capsular end-feel for extension, accounting for 82.0% of the chi-square value. In flexion, most subjects had a tissue approximation end-feel, accounting for 70.0% of the chi-square value. The hypothesis that a significant proportion of subjects would have capsular end-feels (H:2) was supported for extension but not for flexion.

[TABULAR DATA OMITTED]

The number of subjects demonstrating each category of pain/resistance sequence is shown in Table 5. Most of the subjects had no pain, or pain occurred with resistance. There were few subjects in whom pain occurred before or after resistance. The hypothesis that most subjects would have no pain or pain after resistance (H:3) was not supported for either extension or flexion. There was no statistical difference in the number of subjects in the two categories for extension ([chi.sup.2]=2.32). The number of subjects in each of the combined categories of pain/resistance sequence differed from a uniform distribution (50% of the subjects in each of the two cells) for flexion ([chi.sup.2]=5.23, [rho.less than .05), but the majority of the subjects were in the category that combined pain with resistance and pain before resistance.

[TABULAR DATA OMITTED]

Passive ROM of flexion differed depending on type of end-feel (H:4). Passive ROM of flexion for subjects with a tissue approximation end-feel was greater than passive ROM of subjects with capsular end-feels ([chi][sup.2]=28.13, [rho]less than .001). Pain of these subjects did not differ depending on end-feel ([chi][sup.2]=4.90 for extension and [chi][sup.2]=3.35 for flexion). The hypothesis that subjects with spasm and empty end-feels would have greater pain (H:5) was not supported.

The Spearman rank correlation coefficient for pain/resistance sequence and pain intensity was .35 (n=62, [rho]=.003) for extension and .30 (n=62, [rho]=.009) for flexion. The correlation between pain/resistance sequence and the number of months the patient had stiffness was .03 (n=43, [rho]=not significant) for extension and -.01 (n=45, [rho]=not significant) for flexion. These correlations support the hypothesis that pain/resistance sequence would be correlated with another variable representing pain intensity (H:6) but not that pain/resistance sequence would be correlated with another variable representing chronicity (H:7).

The reliability of ROM measurements ranged from .71 to .86 for knee extension and from .95 to .99 for knee flexion. Passive extension and flexion ROM did not differ between test occasions (Tabs. 6 and 7). End-feel also did not differ between test occasions for extension [zeta]=-O.31) or flexion ([zeta]=-1.25). The kappa coefficients for extension end-feel and flexion end-feel were .17 and .48, respectively, indicating slight agreement for extension and moderate agreement for flexion.[14] There were no significant differences between the posttreatment and follow-up measurements of pain/resistance sequence for either extension ([zeta]=-1.61) or flexion ([zeta]=-O.65). The kappa coefficients for the pain/resistance sequence were .36 for extension and .34 for flexion, indicating only fair agreement.[14]

Table 6. Analyses-of-Variance Results for Passive Knee Extension Range of
Motion

Source of Variation     df    SS        MS       F       P

Examiner 1
  Between people        21    1264.91   60.23    5.79    <.01
  Within people         22     229.00   10.41
    Between measures     1      11.00   11.00    1.06    NS[a]
    Residual            21     218.00   10.38
  Total                 43    1493.91   34.74
                              ICC[b](3,1)-.71

Examiner 2
  Between people         6     369.43   61.57   14.62    <.01
  Within people          7      29.50    4.21
    Between measures     1       0.07    0.21    0.01    NS
    Residual             6      29.43    4.90
  Total                 13     398.93   30.69
                              ICC(3,1)=.85

Examiner 3
  Between people        17    1996.25  117.43   13.51    <.01
  Within people         18     156.50    8.69
    Between measures     1       8.03    8.03    0.92    NS
    Residual            17     148.47    8.73
  Total                 35    2152.75   61.51
                              ICC(3,1)=.86

[a]NS=not significant.
[b]ICC=intraclass correlation coefficient.

Table 7. Analysis-of-Variance Results for Passive Knee Flexion Range of
Motion

Source of Variation    df    SS          MS        F      P

Examiner 1
  Between people       21    28219.73    1343.80   52.51  <.01
  Within people        22      563.00      25.59
    Between measures    1       29.45      29.45    1.16   NS[a]
    Residual           21      533.55      25.41
  Total                43    18782.73     669.37
                                ICC[b](3,1)=.96
Examiner 2
  Between people        6     4735.43      789.24   99.66  <.01
  Within people         7       56.00        8.00
    Between measures    1       23.14       23.14    4.23  NS
    Residual            6       32.86        5.48
  Total                13     4791.43      368.57
                                ICC(3,1)=.99

Examiner 3
  Between people       17     14440.25      849.43  32.63  <.01
  Within people        18       468.50       26.03
    Between measures    1        66.69       66.69  2.82   NS
    Residual           17       401.81       23.64
  Total                35     14908.75      425.96
                               ICC(3,1)=.95

[a]NS=not significant
[b]ICC=intraclass correlation coefficient.

Discussion

Pattern of Restriction, End-feel,

and Pain/Resistance Sequence

as Indicators of Osteoarthritis

Pattern of restriction. A capsular pattern is supposed to indicate involvement of the entire capsule and is expected in OA.[1(p406)] There was a scarcity of patients with OA who had a capsular pattern. Perhaps the majority of these patients had not yet developed the capsular pattern. If the capsular pattern did not develop until very late in the disease, then the system would not be of much assistance in diagnosing OA. Cyriax stated, however, that the capsular pattern would exist regardless of whether the patient is early or late in the course of the disease. He claimed that only the end-feel, not the pattern of restriction, would change with an advancing condition.[1(p53)]

The relative absence of patients with a capsular pattern is more likely a matter of definition. The method of defining a capsular pattern in this study depended on the extension loss/flexion loss percentage defined as the criterion. Cyriax claimed that the loss of extension would be about 11% of the loss of flexion.[1(p56)] In this study, the extension loss represented a larger proportion of the flexion loss than Cyriax suggested (X=40%, SD=27, range=0-130). We have observed that clinicians tend to interpret the capsular pattern as flexion loss greater than extension loss but ignore the proportional relationship between the losses. if this definition of the knee capsular pattern were used, then 76 subjects would have shown a capsular pattern. Most activities of daily living do not require full flexion ROM,[15] so flexion ROM may be lost more easily than extension ROM. Function is affected by only a small loss of extension.[16,17] Patients would be inclined to retain more extension ROM by using their knees in their daily activities.

Cyriax used passive motion testing to indicate the pattern of restriction and as a provocation test, that is, to determine whether the application of force reproduces the patient's pain.[1(p50)] We did not use passive motion assessment as a provocation test, but we believe that this omission did not affect the results substantially. Provocation testing is used primarily to reproduce the patient's symptoms and not to determine the pattern of restriction.[1]

End-feel. According to the examiners in this study, most subjects had the expected capsular end-feel for extension. The end-feel for passive extension in healthy knees is supposed to be capsular. The end of motion might occur earlier in the range when a subject has OA, but the qualitative sensation felt by the examiner would be unchanged. A large number of subjects had tissue approximation end-feels for flexion, which was not expected and may be an overestimate. One likely reason for this result was the obesity of many of the subjects. Because their extremities were large, they could have tissue approximation end-feels along with limitations in passive ROM.

The examiners characterized some subjects as having each of the other end-feels. Each type of end-feel might be expected in OA under specific circumstances, but none of the others would be expected to be common. Spasm end-feels are expected early in the disease as muscles act to protect the joint from motion. Bony end-feels are expected after the joint has deteriorated to the point that osteophyte formation prevents motion. Springy block end-feels might be expected if the patient has an intra-articular derangement, such as an impacted loose body. The small number of patients with bony end-feels for either extension or flexion may have resulted from selection bias; some referring physicians may have screened potential subjects having radiographic evidence of bony blockage.

The accuracy of the number of subjects in each end-feel category is affected by the poor estimates of the reliability of the end-feel data. Because several categories have rather abrupt termination of motion, these categories are difficult to distinguish from each other. Some subjects, therefore, may have been categorized incorrectly. For both motions, most subjects were classified as having end-feels associated with healthy knees. If these classifications were incorrect, using end-feel as a diagnostic indicator would likely lead to frequent underdiagnosis of the condition.

The assumptions underlying the categories of end-feel proposed by Cyriax have not been studied. For example, when the end-feel feels like leather being stretched, Cyriax claimed that capsule or ligament is stopping movement, and when motion ceases abruptly, he assumed that bone is hitting bone.[1(p53)] There is no evidence, however, that structures identified in the end-feel category labels are actually the structures that stop motion. Other practitioners have suggested changing end-feel category labels to more descriptive ones such as "soft", "firm," and "hard."[18,19] Riddle[20] proposed operational definitions for descriptive end-feels. He suggested that a soft end-feel demonstrates a gradual increase in resistance to movement at end-range; a firm end-feel is an abrupt increase in end-range resistance, and a hard end-feel entails an immediate cessation of movement at end-range. This nomenclature avoids the problem of assuming what structure stops motion, but the classification still must be studied to determine whether such assessment appropriately guides diagnosis or treatment selection.

Pain/resistance sequence. The number of subjects with no pain on overpressure supports Cyriax's contention that passive motion is often painless in OA.[1(p11)] The poor reliability for the pain/resistance sequence data makes it difficult to draw conclusions about the use of the pain/resistance sequence as an indicator of OA. The small number of subjects with pain after resistance would suggest that the pain/resistance sequence is not a good indicator of OA. Because the OA of these subjects was long-standing, more of them were expected to demonstrate pain after resistance. The pain/resistance sequence measure may have misclassified subjects as having a moderately acute condition. Alternatively, perhaps some of these subjects had experienced an event, such as an acute flare or unusual activity, that triggered an acute response in their joints.

Cyriax might disagree with the diagnosis of the majority of these subjects, and in some cases, he would probably be correct. In this study, medical diagnosis of OA was used as the "gold standard" for comparison. Although the combination of radiographic evidence and clinical signs can have very good sensitivity and specificity,[2] clinical diagnosis is not flawless. Nonetheless, we advise against intrepreting the variables examined in our study, especially a proportional definition of the pattern of restriction, as sensitive indicators of OA for treatment purposes. According to our data, if the capsular pattern were incorrectly assumed to be highly sensitive and specific for OA, such an assumption would cause many false negative results. As a consequence, patients might be treated as if they had a remediable, local problem, rather than a chronic, degenerative condition. Short-term treatment might be similar to that for OA, but the long-term management might differ in important ways. Patients would not be directed toward self-management, joint protection, and appropriate modifications in activities and lifestyle.

Relationships Among Pattern of

Restriction, End-feel, and

Pain/Resistance Sequence and

Related Constructs Underlying

Joint Motion, Pain intensity,

and Chronicity

Subjects with tissue approximation end-feels had more ROM than subjects with capsular end-feels, supporting a relationship between end-feel and the underlying basis, or construct, for joint motion. Subjects with tissue approximation end-feels were expected to have more ROM than subjects with spasm end-feels, but they did not. In addition, subjects with spasm or empty end-feels were expected to have more pain than subjects with other types of end-feel, and they did not report more pain. Bearing in mind the poor reliability for the end-feel data, these results tentatively support Cyriax's claim that as the disease progresses, the flexion end-feel changes from tissue approximation to capsular,[1(pp52,406)] but refute his idea that pain causes muscles to act to limit motion.

Pain intensity on the previous day is a composite of pain experienced during rest and activity, both weight bearing and non-weight bearing, and may not be related to the level of pain experienced during end-feel testing. The relationship might be stronger if pain intensity had been assessed at the time of end-feel testing, as is commonly done clinically.

The correlation between pain/resistance sequence and pain measured with the VAS was low but significant. The correlation may have been low because of the questionable reliability of measurement of the pain/resistance sequence. To estimate the potential magnitude of the correlations, we corrected them for attenuation due to unreliability.[21] Because no reliability data were available for the pain measure, it was assumed to have been measured without error. The corrected Spearman rank correlation coefficients were .58 for extension and .52 for flexion. This outcome suggests that the pain/resistance sequence is related to pain intensity but is nonredundant, contributing a unique bit of information beyond pain averaged over daily activity.

The correlation between pain/resistance sequence and the number of months of stiffness was extremely I suggesting that the pain/resistance sequence is not a measure of chronicity. Even when corrected for unreliability, assuming that the number of months of stiffness was measured without error, the correlation coefficients were still low (rho=.07 for extension and -.02 for flexion). If the pain/resistance sequence represented the concept of chronicity, then pain after resistance would represent a chronic state; pain with resistance would indicate a subacute state, and pain before resistance would indicate an acute state. The low corrected correlation coefficients suggest that this pattern is not present in these data.

In this study, the measure of chronicity was the length of time the patient felt joint stiffness. In discussing the pain/resistance sequence, Cyriax referred to the activity of the lesion.[1(p54)] Although the two concepts are related, months of stiffness may not reflect the chronicity of the tissue reaction. Nonetheless, the lack of correlation between pain/resistance sequence and months of stiffness diminishes the validity of using the pain/resistance sequence to indicate the chronicity of the lesion.

Based on these data, the validity of some of the assumptions of selective tension testing is questionable. More investigation of the validity of passive motion and the other components of the system is necessary. The diagnostic accuracy of the system must be examined in prospective studies of a wide variety of conditions in differing patient populations. Because results from the knee should not be generalized to other joints, similar studies should examine different joints, particularly their capsular patterns.

Reliability

The reliability estimates for measurements of extension and flexion ROM do not differ markedly from those of other reliability studies of goniometric measurements of knee ROM in which intrarater reliability values of .85 to .98 for extension and .95 to .99 for flexion were found.[22-24] As in these previous studies, reliability was better for flexion than for extension. The lower reliability for knee extension could reflect the difficulty therapists have aligning the goniometer in extension and the inability of a goniometer to account for the rotation of the tibia that occurs as the knee completes extension.[25] This lower reliability may also be a result of the smaller variability in knee extension ROM among subjects compared with the variability of knee flexion.

The reliability estimates of end-feel and pain/resistance sequence assessments may have been low because there was limited variability in the group on both variables. Consequently, chance agreement would be high, decreasing the kappa coefficient.[26] Kappa changes with the probabilities of each of the possible categories and is best when the probabilities are approximately equal. The maximum possible kappa coefficient can be calculated for a given set of marginal probabilities.[13] Given the distributions in this study, the maximum kappa coefficient would be .78 for extension end-feel, .78 for flexion end-feel, .75 for pain/resistance sequence in extension, and .88 for pain/resistance in flexion. For both variables, the reliability estimates are considerably below these values. The reliability of the pain/resistance sequence assessments may be low because the time interval between the onset of pain and the onset of resistance may be too short to determine clinically through manual palpation. The low reliability estimates could represent actual patient change over the 2-month period; however, there were no statistical differences in grades between measurements, and passive ROM reliability estimates were acceptable or nearly acceptable over the same time period. We believe that actual changes in end-feel and pain/resistance sequence are unlikely. The reliability of both end-feel and pain/resistance sequence assessments is probably unacceptable, but should be studied again with less time between measurements and greater variability in the sample.

The low reliability estimates of the end-feel and pain/resistance sequence assessments are similar to those found by other investigators examining tests that rely on physical therapists' judgment of very small motion such as Lachman's Test,[27] tibiofemoral abduction,[28] and tests of sacroiliac mobility.[29] Patla and Paris[30] found the percentage of intrarater agreement of end-feel testing of the elbow to be 75% to 80%, but there was little variability in their sample. Chance agreement, therefore, would be high but was not reported.30 The results of this study underscore the dependence of validity on reliability. It must be possible to classify patients consistently in the same category of end-feel or pain/resistance sequence to have confidence in relationships cited as evidence for or against the validity of Cyriax's system or to make diagnostic and treatment decisions using the system.

Conclusions

The value of studying the validity and reliability of any measurement system is to obtain data that allow refinement of measurements that are potentially informative and to seek new systems if existing systems are inadequate. This study examined the passive motion components of the soft tissue diagnosis system proposed by Cyriax. We examined validity by studying whether the three passive motion components were indicators of subjects with OA of the knee. We also examined relationships among the three indicators of dysfunction and related constructs underlying joint motion, pain intensity, and chronicity. Last, we estimated the test-retest reliability of measurements of each of the three components. The results of this study provide evidence of the need to question and further examine selective tension testing as a diagnostic system. Test-retest reliability estimates were acceptable for passive ROM measurements but not for end-feel and pain/resistance sequence classification. Very few subjects exhibited a capsular pattern by Cyriax's quantitative definition. A proportional definition of a capsular pattern should be abandoned, but the concept of a pattern of ROM loss may be useful. When corrected for unreliability, pain/resistance sequence is an indicator of pain intensity but not chronicity. Poor reliability estimates limit our ability to interpret additional findings. For example, more subjects retained tissue approximation end-feels than predicted; fewer subjects had painless end-feels or pain after resistance during end-feel testing than predicted, and end-feel was related to joint motion but not to pain intensity. More investigation of selective tension testing is needed to improve the reliability and examine other facets of validity, particularly the use of the system to guide treatment decisions.

Acknowledgments

We thank the Biostatistical and Data Management Core of the Northwestern University Multipurpose Arthritis Center for their assistance in data processing and data management, especially Ahn Chung and Delilah Jones. We also thank Katie Sirianni, PT, Linda Tieman Roherty, PT, and Babette Sanders, PT, for serving as evaluators in this study and Russell M Woodman, PT, FSOM, OCS, for consulting with us.

(*) Apple Computer Inc, 20525 Mariani Ave, Cupertino, CA 95104. (+) SPSS INC, 444 N Michigan Ave, Chicago, IL 60611.

References

[1] Cyriax J. Textbook of Orthopaedic Medicine, I: Diagnosis of Soft Tissue Lesions. 8th ed. London, England: Bailliere Tindall; 1982. [2] Altman R, Asch E, Bloch D, et al. Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum. 1986; 29:1039-1049. [3] Ghiselli EE, Campbell JP, Zedeck S. Measurement Theory for the Behavioral Sciences. New York, NY: WH Freeman and Co; 1981. [4] Falconer J, Hayes KW, Chang RW. Effect of ultrasound on mobility in osteoarthritis of the knee: a randomized clinical trial. Arthritis Care Res. 1992;5:29-35. [5] Lowman EW. Osteoarthritis. JAMA. 1955;157: 487-488. [6] Scott J, Huskisson EC. Graphic representation of pain. Pain. 1976;2:175-184. [7] Kirby RL. Inspection-palpation-percussion-auscultation and an outcome-oriented alternative approach to the musculoskeletal examination. Med Educ. 1981;15:106-109. [8] Price DD, McGrath PA, Rafii A, Buckingham B. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain. 1983;17:45-56. [9] Kaufert JM, Green S, Dunt DR, et al. Assessing functional status among elderly patients: a comparison of questionnaire and service provider ratings. Med Care. 1979;17:807-817. [10] Shinar D, Grass CR, Bronstein KS, et al. Reliability of the activities of daily living scale and its use in telephone interview. Arch Phys Med Rehabil. 1987;68:723-728. [11] Siegel S, Castellan NJ. Nonparametric Statistics for the Behavioral Sciences. 2nd ed. New York, NY: McGraw-Hill Inc; 1988. [12] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull, 1979;86:420-428. [13] Cohen J. A coefficient of agreement for nominal scales. Educational and Psychological Measurement. 1960;20:37-46. [14] Landis RJ, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159-174. [15] Laubenthal KN, Smidt GL, Kettlekamp DB. A quantitative analysis of knee motion during activities of daily living. Phys Ther. 1972;52:34-43. [16] Cerny K, Walker J, Perry J. Adaptations during the stance phase of gait for simulated flexion contractures at the knee. Phys Ther. 1988; 68:797. Abstract. [17] Potter PJ, Kirby RL, MacLeod DA Effects of simulated knee-flexion contractures on standing balance. Am J Phys Med Rehabil. 1990;69: 144-147. [18] Kaltenborn FM, Evjenth O. Manual Mobilization of the Extremity Joints Basic Examination and Treatment Techniques. 4th ed. Oslo, Norway: Olaf Norlis Bokhandel Universitetsgaten; 1989. [19] Torg JS, Conrad W, Kalen V. Clinical diagnosis of anterior cruciate ligament instability in the athlete. Am J Sports Med. 1976;4:84-93. [20] Riddle DL. Measurement of accessory motion: critical issues and related concepts. Phys Ther 1992;72:865-874. [21] Carmines EG, Zeller RA. Reliability and Validity Assessment. Newbury Park, Calif: Sage Publications Inc; 1979. [22] Clapper MP, Wolf SL. Comparison of the reliability of the Orthoranger[R] and the standard goniometer for assessing active lower extremity ranges of motion. Phys Ther. 1988; 68:214-218. [23] Rothstein JM, Miller PJ, Roettger RF. Goniometric reliability in a clinical setting: elbow and knee measurements. Phys Ther. 1983;63: 1611-1615. [24] Watkins MA, Riddle DL, Lamb RL, Personius WJ. Reliability of goniometric measurements and visual estimates of knee range of motion obtained in a clinical setting. Phys Ther. 1991; 71:90-97. [25] Enwemeka CS. Radiographic verification of knee goniometry. Scand J Rehabil Med. 1986; 18:47-49. [26] Soeken KL, Prescott PA. Issues in the use of kappa to estimate reliability. Med Care. 1986;24:733-741. [27] 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. 199o; 70:225-233. [28] McClure PW, Rothstein JM, Riddle DL. Intertester reliability of clinical judgments of medial knee ligament integrity. Phys Ther. 1989; 69:268-275. [29] Potter NA, Rothstein JM. Intertester reliability for selected clinical tests of the sacroiliac joint. Phys Ther. 1985;65:1671-1675. [30] Patla CE, Paris SV. Reliability of interpretation of the Paris classification of normal end feel for elbow flexion and extension. Journal of Manual and Manipulative Therapy. 1993; 1: 60-66.