Relationships Among Selected Measures of Impairment, Functional Limitation, and Disability in Patients With Cervical Spine Disorders.

Cervical spine disorders (CSD) are a common cause of neck pain. Population-based surveys have shown lifetime prevalence rates for neck pain between 67% and 71%,[1,2] whereas between 13% and 22% of the population in industrialized society experiences neck pain at any point in time.[1,3] Cervical spine disorders are not just limited to adults. A recent survey of Finnish high school students showed that 21% of women and 10% of men experienced neck or shoulder pain at least once per week.[4] Cervical spine disorders also account for a substantial number of physical therapy outpatient visits each year. In one report, CSD accounted for 26% of outpatients seeking physical therapy treatment.[5]

When patients with CSD first see a physical therapist, an examination process should occur so that a plan of care can be established and treatment goals can be set.[6] Although there are no research data, some authorities advise assessment of posture, joint range of motion (ROM), and strength among other variables with the assumption that these relate to functional loss.[6,7] Changes in these variables are then supposed to be used to monitor progress and guide treatment. In addition, measures of cervical ROM have been used to evaluate the severity of impairment[8] or disability[9] in patients with work-related CSD.

Many people propose using a disablement model in physical therapy research and practice.[6,10] A disablement model links a disease or pathology to functional consequences. One such model is the International Classification of Impairments, Disabilities, and Handicaps (ICIDH).[11] In this model, there are 4 sequential concepts: disease, impairment, disability, and handicap. This model has been criticized for its definitions of the disability and handicap concepts in that there is some overlap leading to confusion in how to classify an individual with multidimensional limitations.[10] More recently, a modified version, the ICIDH-2,[12] was developed. The ICIDH-2 model classifies information at 3 levels of functioning (ie, body, activity, and society). These levels are known as "dimensions" and are named "Body Functions and Structure," "Activities," and "Participation," respectively. Each dimension consists of "domains" or categories that are used to further classify information. Contextual factors (ie, environmental and personal) that may influence an individual's state are also included.

In this study, we used a well-known disablement model initially proposed by Nagi.[13] This model also contains 4 sequential concepts: pathology, impairments, functional limitations, and disability. In theory, pathology produces impairments, which then lead to functional limitations, which then result in disability. Not all impairments result in functional limitations, however, and not all functional limitations result in disability. In order for the clinician to provide effective care that will ultimately affect function and reduce the potential for disability, therapeutic interventions should, in theory, target only those impairments that are related to (or producing) the functional limitations. Indeed, it has been suggested that, through the examination process, the clinician determines the interrelationships among impairments, functional limitations, and disability for a patient with a given diagnosis and that this information then guides treatment.[6] Examples of cervical spine impairments include deficits in ROM or muscle force, whereas a functional limitation might be a driver's inability to rotate the head and neck to be able to see behind when driving an automobile in reverse. If an individual is then unable to work because his or her occupation requires automobile use, that person could be considered to have a disability.

Little is known about the relationship among impairments, functional limitations, and disability concerning the cervical spine. Schenkman and coworkers[14] have shown a relationship between cervical ROM and "functional axial rotation" (FAR). The FAR test is a measure of combined spinal motion that, in turn, correlates with measures of physical performance, but does not reflect function as it is often defined. In an earlier study, this same group demonstrated that measures of FAR can be used to infer the amount of rotation available for the performance of functional activities.[15] However, the subjects in both of these studies were volunteers without cervical spine pathology. Hagen et al[16] examined the relationship between self-reports of neck pain and limited or painful cervical ROM in 49 male machine operators. They found a weak inverse relationship between cervical ROM and pain experienced within that ROM for extension (r=-.29, P=.04) and left axial rotation (r=-.32, P=.03). Stronger relationships were observed between cervical ROM and the corresponding pain level and self-reports of pain on the Standardized Nordic Questionnaire. More recently, Riddle and Stratford[17] reported weak correlations (r=.12-.40) between cervical ROM and functional status measurements obtained from the Neck Disability Index (NDI) and the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36). Another investigation demonstrated a positive correlation (r=.65, P [is less than] .001) between pain scores on a visual analog scale (VAS) and self-reported disability using the NDI in subjects with neck pain.[18]

Relationships involving cervical muscle force and endurance have also been investigated. A recent report showed low correlations (r [is less than or equal to] .20) between cervical extensor force, endurance, and ROM on the one hand, and pain and disability on the other.[19] Another study documented improved cervical extensor force and endurance in patients with neck pain and also reported a 50% reduction in pain and disability.[20] Subjects in this study did not show improvements in cervical flexor force or extension ROM. A relationship between pastural abnormalities and the incidence of pain in the cervical region has also been demonstrated.[21]

We believe that these findings indicate a disparity in level of association among measures of impairment, functional limitation, and disability in subjects with CSD and suggest the need for further study. According to the disablement model developed by Nagi,[13] impairment is defined as any loss or abnormality of anatomic, physiologic, mental, or emotional structure or function. Functional limitation is defined as a limitation in performance at the level of the whole organism or person.[13] Disability then applies to a limitation in performance of socially defined roles and tasks within a sociocultural and physical environment.[13] If the relationships among physical impairments, functional limitations, and disability in patients with CSD were known, we believe clinicians could examine patients in a more efficient manner, provide a more accurate prognosis, and gauge progress through a therapeutic regimen based on more valid measures. This information could also be useful for providing rationales for treatment to third-party payers.[6] Therefore, the primary purpose of this study was to investigate the relationship among specific cervical spine impairments, functional limitations, and disability in patients with CSD. More specifically, we sought to determine the associations among 3 measures of impairment (pain, ROM, and muscle force), 2 measures of functional limitation (FAR and a test of lifting ability), and 3 self-report measures of disability (NDI and 2 summary scales from the SF-36) in this patient population. A secondary purpose was to determine the influence of payment source and time since onset of symptoms on these same measures.

Method

Subjects

We recruited consecutive patients who were referred to a hospital-based outpatient physical therapy clinic and who met the inclusion criteria. All patients were referred by a physician (doctor of medicine or doctor of osteopathy). The inclusion criteria were (1) a primary diagnosis of neck pain or cervical pathology by the referring physician, (2) no contraindications to ROM or muscle force testing or to the performance of a functional lift test, and (3) the ability to read and write English. Subjects were excluded if they had comorbidities that would affect test performance (eg, low back pain, shoulder pathology). A total of 80 subjects (32 men, 48 women) fulfilled these criteria and were included in the study. The subjects' ages ranged from 20 to 88 years ([bar]X=45.7, SD=15.9). Subject demographic data are presented in Table 1.

Table 1. Demographic Characteristics of the Subjects (N=80)

Characteristic                   n    %

Age (y)
  20-30                         12   15
  31-40                         26   33
  41-50                         13   16
  51-60                         17   21
  61-70                          4    5
  >70                            8   10

Sex
  Male                          32   40
  Female                        48   60

Time since onset of symptoms
  0-14 d                        17   21
  15 d-6 mo                     29   36
  >6 mo                         34   43

Payment type
  Managed care                  40   50
  Fee-for-service               40   50

Employment status
  Unaltered                     54   68
  Altered (ie, modified duty,
    off-duty due to health)      8   10
  Retired                       12   15
  Unemployed                     6    7

Workers' compensation            7    9

Motor vehicle accident          22   28

Litigation                       8   10

Testing Procedure

After obtaining written informed consent, scores for 3 measures of impairment, 2 measures of functional limitations, and 3 measures of disability were obtained from each subject during the initial visit (ie, before any treatment intervention occurred). The entire testing procedure took approximately 45 minutes. Each subject was tested by 1 of 3 physical therapists who had 11, 3, and 2 years of experience, respectively.

Impairment Measures

The impairment measures were pain, cervical spine ROM, and cervical muscle force. These 3 variables were chosen because they are commonly assessed in clinical practice and because they have been used in previous studies.[16,18-20,22-25] Pain was assessed through the use of a VAS consisting of a 10-cm vertical line with the anchors "Pain as bad as it could be" and "No pain at all."[26] Subjects were instructed to draw a mark across the line that corresponded to their pain level at that moment. Acceptable reliability of the VAS scores has been shown previously (r=.98, N=39).[27]

Cervical spine ROM was assessed using the Cervical Range of Motion (CROM) device(*) in a manner similar to Youdas et al.[25,28] Subjects were positioned sitting in a metal framed chair with their thoracic spine in contact with the chair's backrest, their feet flat on the floor, and their arms hanging at the sides. To overcome initial apprehension and to increase tissue compliance, subjects performed one movement each of cervical flexion, cervical extension, left lateral flexion, right lateral flexion, left rotation, and right rotation. They then performed 2 subsequent series of movements in the same order. Subjects were instructed to move their head and neck as far as possible until the active ROM was stopped by pain or muscle tightness or until a substitution movement occurred. The mean value for each position during these 2 movement series was then calculated and used for further analysis.

Although acceptable reliability for measuring cervical spine ROM using the CROM device has been shown previously (intraclass correlation coefficient [ICC] (1,1) [is greater than] .80, N = 60),[25] intertester reliability for this same test as well as for muscle force and the 2 functional limitation measures was determined for the purposes of this study. Eight subjects (2 men and 6 women) ranging in age from 17 to 52 years ([bar]X=33.4, SD=11.1) with no history of cervical pathology were tested on one occasion by each tester. A minimum of 4 hours separated the tests. The testers were unaware of each other's results. Intraclass correlation coefficients (ICC [2,1])[29] were calculated for each measure. For the cervical ROM measurements, intertester reliability ranged from .70 for cervical rotation to .88 for cervical flexion.

In our study, the isometric force of the cervical flexor and extensor muscles was measured with a handheld dynamometer.([dagger]) The dynamometer was calibrated prior to the initiation of the study, at the end of the study, and at 6-month intervals according to the manufacturer's instructions. Cervical flexor force was measured in a manner that has been performed previously.[19,23] Subjects were positioned supine with their chin retracted and neck flexed approximately 20 degrees. The dynamometer was placed over the forehead, and resistance was applied by the tester until this position could no longer be maintained (ie, break test). Cervical extensor force was measured with the subjects in a prone position and the cervical spine in approximately 20 degrees of extension. The dynamometer was placed just superior to the occiput, and resistance was applied until the test position could no longer be maintained. For each position, three 5-second contractions with 30- to 60-second rest periods were performed. The recorded force was the force the tester had to apply to break the isometric contraction and move the subject's head in the opposite direction. The mean value of the 3 measurements was then calculated for both test positions and then normalized to body weight. Similar to the findings of others,[19,23] intertester reliability (ICC [2,1][29]) in this study was .93 and .70 for cervical flexor force and cervical extensor force, respectively. We believe we have provided some evidence for the reliability of our measures, but we recognize that one limitation to our study was that reliability was not determined for subjects with CSD.

Functional Limitation Measures

Two measures of functional limitation were recorded: a measure of combined spinal motion (the FAR test) and a lift test. Combined spinal motion is a component of many routine daily activities (eg, reaching, looking behind oneself). Intuitively then, the FAR test could be a measure of function, although data in support of this are lacking. The FAR test was performed according to the procedure described by Schenkman et al.[15] Subjects sat inside a hoop, which was suspended by 2 tripods at eye level, with their feet flat on the floor and their arms hanging at their sides. Subjects were instructed to rotate the head, neck, and trunk to either side as far as possible without lifting their buttocks from the chair seat. The pelvis was stabilized with a Velcro strap.([double dagger]) The amount of motion was measured by aligning a pointer attached to the subject's head with landmarks on the hoop corresponding to degrees of rotation. Three trials to each side were performed, with the first trial serving as a warm-up. The mean value of the following 2 trials were computed and used for subsequent analysis. Intertester reliability (ICC [2,1][29]) for right and left FAR was .76 and .80, respectively. There is no research to support whether this measure of impairment actually predicts function.

A test of lifting capacity, another impairment measure thought to relate to function, was used in this study.[30,31] The test involved a cycle of lifting a weighted box (3.6 kg for women, 5.9 kg for men) from a 76.2-cm-high shelf to a 137.2-cm-high shelf and back again. Subjects were required to complete 4 cycles within 20 seconds. If all 4 cycles were successfully completed in the 20-second time period, the weight was increased by 2.2 kg for women and 4.5 kg for men. The test continued until 1 of 3 endpoints was reached: (1) 85% of age-adjusted maximal heart rate (220 - age) was achieved, (2) the limit of fatigue or pain was reached as determined by the subject, or (3) subject safety was at risk as evidenced by a substitution pattern. The final weight was recorded and normalized to adjusted body weight as specified by Mayer et al.[30] In this study, intertester reliability (ICC [2,1][29]) for the lift test was found to be .88.

Disability Measures

Both a condition-specific and a generic instrument for self-report of disability and health-related quality of life were used in our study. The NDI[32] is an adaptation of the Oswestry Low Back Pain Questionnaire.[33] It consists of 10 items (subscales) containing 6 statements each that evaluate pain, sleep quality, work ability, and various activities of daily living. The questionnaire was scored as a percentage. High scores indicate a high level of perceived disability, and low scores indicate a low level of perceived disability. Vernon and Mior[32] found test-retest reliability (Pearson product moment correlation coefficient) for the NDI to be .89.

The generic instrument used was the SF-36,[34] which is a self-administered 36-item questionnaire designed to assess general health status. The SF-36 is composed of 8 subscales, each of which measures a distinct component of health status: (1) physical functioning, (2) social functioning, (3) bodily pain, (4) role limitations due to physical problems, (5) role limitations due to emotional problems, (6) general mental health, (7) vitality (energy or fatigue), and (8) general health perceptions. Each component is scored on a scale from 0 to 100, with higher scores representing better health status for that component. In addition, 2 summary scales can be obtained: the physical component summary scale (PCS) and the mental component summary scale (MCS). These scales represent the physical and mental components of health, which are based on weighted aggregates of all 8 subscales, and they were used in this study. Acceptable reliability for the SF-36 has been reported previously.[35] The SF-36 was scored using the statistical software the developers provided with the test.[36]

Order of Testing

The order of testing was the same for all patients, as follows: measurements of pain with the VAS, measurements of cervical ROM and FAR, measurements of force, performance of the lift test, and completion of the NDI and SF-36 questionnaires. Pain measurement was done first in an attempt to eliminate the effects of subsequent test procedures on this variable.

Data Analysis

Subjects were placed in 1 of 3 groups depending on the amount of time since the onset of symptoms (ie, acuity): acute (0-14 days), subacute (15 days-6 months), and chronic ([is greater than] 6 months). Descriptive statistics were used to characterize the 3 groups and their scores on the dependent measures. Before further analysis, a principal component analysis was performed on the cervical ROM, muscle force, and FAR measurements in order to reduce the number of variables entered into the correlation analysis. The main benefit of using a principal component analysis is the ability to represent the variability of several variables by a statistically determined linear combination of those variables. This reduces a multivariate problem into a univariate problem. In our study, we used principal component analysis to find the linear combination of each group of measurements (cervical ROM [6 measurements], muscle force [2 measurements], and FAR [2 measurements]) that accounted for the greatest amount of variation inherent in those measurements, making it the most representative variable to summarize that group of variables. The result is 3 variables (ROM, muscle force, and FAR) rather than the original 10.

Canonical correlations were then calculated to examine relationships among the 3 sets of variables (ie, impairment, functional limitation, disability measures) while controlling for age, sex, and acuity. Canonical correlation is a multivariate technique for assessing relationships between groups of variables. It determines which combination of variables results in the largest correlation. The resulting weights indicate how the variables within a group contribute to the relationship. To assess the relationship between individual variables, a correlation matrix was calculated while again controlling for age, sex, and acuity. Because scores on the VAS and the disability measures cannot be considered ratio levels of measurement, a Spearman rank correlation coefficient was used. In addition, because the analysis for the individual variables involved 28 correlations, we set the time since onset of symptoms at 2 levels: acute (0 days-6 months) and chronic ([is greater than] 6 months). This served to decrease the number of variables used in calculating the correlation matrix for the individual variables. A multivariate analysis of variance (MANOVA) was used to determine whether there were any differences due to time since onset of symptoms (acute, subacute, chronic) or source of payment (managed care or fee-for-service) for the impairment, functional limitation, and disability measures while controlling for age and sex. Because several correlations were calculated, the alpha levels for the canonical and individual correlations were adjusted to maintain an experiment-wise error rate of .05. The canonical correlations were performed at an alpha level of .015, and the individual Spearman correlations were performed at .0024.

The sample (N=80) in this study consisted of more women (n=48) than men (n=32), a finding consistent with cross-sectional studies reporting higher prevalence rates of CSD in women.[1,3] All age groups were represented except for those under 20 years. Subject scores on the dependent measures are shown in Tables 2, 3, and 4. The data were variable within each group. Despite the 3 different acuity levels, however, little disparity between groups was noted. The results of the principal component analysis revealed that, for the 6 cervical ROM measurements, 66% of the variability within these measurements could be accounted for by one linear combination of the 6 ROM measurements. The linear combinations determined by principal components analysis for the FAR measurements and muscle force measurements accounted for 92% and 90% of the variability within these measurements, respectively.

[TABULAR DATA 2 NOT REPRODUCIBLE IN ASCII]

Table 3. Subject Scores on the Functional Limitation Measures

                         FAR
Group               ([degrees])(a)   Lift Test

Acute (n = 17)
  [bar]X            105.4            0.17
  SD                 26.5            0.08
  Range              57.5-175.0      0.06-0.34

Subacute (n = 29)
  [bar]X            105.0            0.13
  SD                 21.7            0.07
  Range              45.0-150.0      0.05-0.28

Chronic (n = 34)
  [bar]X            105.2            0.15
  SD                 20.2            0.07
  Range              55.0-140.0      0.06-0.34

(a) FAR = functional axial rotation. The FAR values represents the mean of left and right measurements.

Table 4. Subject Scores on the Disability Measures(a)

Group                  NDI         PCS         MCS

Acute (n=17)

[bar]X                 31.8        36.4        45.2
SD                     13.3        10.7        12.0
Range              2.0-50.0   19.4-56.5   24.5-70.3

Subacute (n=29)

[bar]X                 30.3        38.7        47.1
SD                     14.3         8.1        11.6
Range              4.0-60.0   26.1-55.6   23.2-63.8

Chronic (n=34)

[bar]X                 32.3        38.6        47.5
SD                     18.7         9.8        10.5
Range                0-84.0   22.7-55.7   28.6-63.3

(a) NDI=Neck Disability Index, PCS=physical component summary scale of the 36-Item Short-Form Health Survey (SF-36), MCS=mental component summary scale of the SF-36.

Results

The canonical correlations among the impairment, functional limitation, and disability measures are shown in Table 5. All 3 sets of variables were correlated with each other, with the highest correlation occurring between the impairment measures and the functional limitation measures (r=.82, P [is less than] .0001). The canonical correlation measures not only the correlation between groups of variables, but which variables contribute most to the relationship. For the relationship between impairment and functional limitation measures, the variable that contributed most to the relationship was ROM for the impairment measures (ROM [.69], force [.36], pain [-.17]), and FAR for the functional limitation measures (FAR [.77], lift test [.44]). The impairment measures also correlated with the disability measures (r=.73, P [is less than] .0001). In this relationship, the greatest contribution was pain for the impairment measures (pain [.73], ROM [-.47], force [.03]) and the NDI for the disability measures (NDI [1.02], MCS [.04], PCS [.01]). A weaker relationship was noted between the functional limitation measures and the disability measures (r =.54, P [is less than] .0001). Here, the greatest contribution was the lift test for the functional limitation measures (lift test [.68], FAR [.55]) and the NDI for the disability measures (NDI [-.85], PCS [.21], MCS [.00]). In this relationship, the coefficient for the NDI is negative because of its inverse relationship to the other measures (ie, higher scores for the NDI imply greater limitations, whereas high scores for the other measures imply fewer limitations). The absolute value of the coefficient denotes the relative weight contributed by a given variable to the relationship.

Table 5. Canonical Correlations Between Variable Sets

Variable Sets                             r      P

Impairments and functional limitations   .82   .0001
Impairments and disability               .73   .0001
Functional limitations and disability    .54   .0001

The correlation matrix for the individual variables is presented in Table 6. Of the 28 possible individual correlations, 17 were significant. Six of the 11 correlations that were not significant involved the MCS. Particularly strong inverse relationships were noted between the PCS and the NDI (-.67) and between ROM and the NDI (-.54). Strong relationships were also noted between pain and the NDI (.65), ROM and FAR (.65), and ROM and the lift test (.56). The results of the MANOVA did not show any influence of time since onset of symptoms (P=.83) or payment type (P=.14) on the dependent measures (Tab. 7).

[TABULAR DATA 6 NOT REPRODUCIBLE IN ASCII]

Table 7. Relationship Between Time Since Onset of Symptoms and Payment Type on the Measured Variables

Effect                          F       df      P

Time since onset of symptoms    .66   16,136   .83
Payment type                   1.59      868   .14

Discussion

The results of the canonical correlations lend support to what many clinicians intuitively believe to be correct: that, in patients with CSD, there are links between observed impairments and a person's ability to function and his or her perceived level of disability. The impairment measures correlated strongly with the functional limitation measures (r=.82, P [is less than] .0001). This relationship was driven primarily by the 2 variables addressing axial mobility: ROM and FAR. Cervical rotation is a component of FAR, and, therefore, it is not surprising that a positive correlation was found between the impairment measures and the functional limitation measures. A similar finding was reported by Schenkman et al[14] in a population of subjects without spinal pathology. Cervical ROM is commonly assessed in patients with CSD.[25,37]

The impairment-disability relationship was driven primarily by pain and the NDI, 2 self-reported measures. The relationship between these 2 variables individually was also quite strong (r =.65, P [is less than] .0001). Pain intensity is 1 of the 10 areas addressed on the NDI. A relationship between these 2 variables, therefore, would be expected. The relatively minor contribution to the impairment-disability relationship by the PCS and MCS is consistent with a previous finding that showed small changes (compared with population norms) in many of the health domains of a modified version of the SF-36 in patients with neck sprain and both radiating and nonradiating neck pain.[38] This finding also points out the importance of using a condition-specific instrument for assessment of self-perceived disability in patients with CSD.

The lift test and the NDI contributed most to the relationship between the functional limitation and disability measures. The lift test used in this study has been used previously to assess cervical functional ability.[30,31] It involves lifting a weighted box to approximately head height and, as such, would be expected to require some degree of cervical muscle strength. This idea is supported by the individual correlation between cervical muscle force and the lift test (r=.50, P [is less than] .0001). Lifting ability is another area addressed by the NDI and, therefore, we would expect a correlation with this measure.

In this study, FAR and the lift test were considered "functional" measures, implying that they measure some aspect of a person's ability to function. The FAR test measures a person's ability to rotate the head and thorax relative to the pelvis.[15] This action is essential for the performance of many daily activities such as looking over one's shoulder or rolling from a supine position to a side-lying position. The test itself is akin to looking over one's shoulder while in a seated position. In support of its use as a functional measure, Schenkman et al[14] found a strong association between FAR and tests of physical performance (functional reach, moving from a supine to a sitting position, and 360 [degrees] turn). Lifting is another commonly performed daily activity. Lifting ability was assessed using the methods of Mayer et al.[30] The protocol measures the ability to lift a load from approximately waist height to head height as might occur when placing an object on a shelf.

Some additional correlations are noteworthy. A moderate inverse relationship was noted between pain and cervical ROM. Intuitively, one would expect that the more pain a person has, the less he or she would be willing to move the cervical region. Hagen et al[16] found a similar, but weaker, relationship in forest machine operators. The relationship between pain and FAR did not reach statistical significance, which suggests that the subjects were not limited by pain in their ability to rotate the entire spinal column. Perhaps the subjects compensated during FAR by requiring more relative mobility of the thoracic and lumbar regions of the spine. As expected, there was good correlation between the FAR test and cervical ROM (r=.65, P [is less than] .0001) because both measures address axial mobility. However, the correlation between the FAR test and the NDI was not as strong (r=-.41, P [is less than] .001). The fact that the FAR test correlates more strongly with an impairment measure compared with a disability measure suggests that the FAR test is a better measure of impairment than function. Further research is needed to clarify interpretation of FAR test results in this population. A moderate inverse relationship was also noted between pain and the lift test. The most common reason for terminating the lift test was pain that may have limited the subjects' ability to exert force. However, it has also been reported that patients with CSD have weaker cervical musculature.[23,24] Either of these could have contributed to this relationship. In addition, it has been shown that incorporation of strengthening exercises for the cervical musculature in rehabilitation programs is associated with improvements in pain and disability.[24,37,39]

The relationship between the NDI and both the PCS and MCS found in our study is similar to that reported by Riddle and Stratford.[17] In their study, correlations between the NDI and the PCS and MCS were .53 and .47, respectively. In that same study, relatively low correlations were noted between the NDI and cervical ROM measurements, whereas, in our study, higher correlations were noted. This could be because we used principal component analysis for cervical ROM or because we controlled for age, sex, and acuity.

We defined acuity as the amount of time since the onset of symptoms. Others have defined acuity in patients with CSD in a similar manner.[39] A surprising finding, however, was that there was no overall relationship between acuity and the measured variables. The fact that there was no difference between groups suggests that defining acuity solely in this manner may not be appropriate. Other factors may need to be considered as well (eg, pain pattern, amount of limitation in ROM).

There was also no effect of payment type (managed care or fee-for-service) on the variables measured. This finding is consistent with those of other studies that have shown no differences in health status or health outcomes in patients provided with similar interventions covered by either type of insurance.[40,41]

As with any correlational study, statements regarding causality cannot be made. This study was cross-sectional in nature and merely evaluated the relationships between the various measures. Changes in the impairment, functional limitation, or disability measures followed prospectively would be useful and would provide data on the predictive ability of a given measure. Another limitation of our study is the lack of random sampling. Subjects were consecutively referred to one outpatient physical therapy clinic and, therefore, may not be representative of patients with CSD as a whole. Despite these limitations, our data may be useful for developing testable hypotheses. For example, which variables are most useful for predicting outcomes as a result of therapeutic intervention? More specifically, can a given impairment measure predict a functional outcome? Because improving function is often the primary goal of treatment, those impairment measures that are most closely related to the relevant function should be recorded and followed over the course of care. It may also be that the relationships among impairments, functional limitations, and disability are different for different pathologies. The subjects in our study included those with both radiating and nonradiating neck pain. No attempt was made to delineate these 2 groups because of the small sample size. Further study is needed using a larger sample to elucidate the nature of these relationships in patients with different pathologies.

Conclusion

The relationship among selected measures of impairment, functional limitation, and disability in patients with CSD was investigated. Positive correlations were noted between the 3 sets of measures in this population, thereby affirming the assumption that impairments, functional limitations, and disability are related. In addition, acuity (as defined by the amount of time since the onset of symptoms) had no effect on the recorded variables. There was likewise no influence on the variables when comparing patients with managed care or fee-for-service payment arrangements. Limitations of the study were addressed, and suggestions for future research were made.

(*) Performance Attainment Associates, 3550 Labore Rd #8, St Paul, MN 55110.

([dagger]) Chatillon, 8600 Somerset Dr, Largo, FL 33773.

([double dagger]) Velcro USA Inc, 406 Brown Ave, Manchester, NH 03108.

References

[1] Cote P, Cassidy JD, Carroll L. The Saskatchewan Health and Back Pain Survey: the prevalence of neck pain and related disability in Saskatchewan adults. Spine. 1998;23:1689-1698.

[2] Makela M, Heliovaara M, Sievers K, et al. Prevalence, determinants, and consequences of chronic neck pain in Finland. Am J Epidemiol. 1991;134:1356-1367.

[3] Bovim G, Schrader H, Sand T. Neck pain in the general population. Spine. 1994;19:1307-1309.

[4] Niemi S, Levoska S, Kemila J, et al. Neck and shoulder symptoms and leisure time activities in high school students. J Orthop Sports Phys Ther. 1996;24:25-29.

[5] Mossberg KA, McFarland C. Initial health status of patients at outpatient physical therapy clinics. Phys Ther. 1995;75:1043-1051.

[6] Rothstein JM, Echternach JL. Hypothesis-oriented algorithm for clinicians. Phys Ther. 1986;66:1388-1394.

[7] Kendall FP, McCreary EK. Muscles: Testing and Function. 4th ed. Baltimore, Md: Lippincott Williams & Wilkins; 1993.

[8] Guides to the Evaluation of Permanent Impairment. 4th ed. Chicago, Ill: American Medical Association; 1993.

[9] Clark W, Haldeman S. The development of guideline factors for the evaluation of disability in neck and back injuries: Division of Industrial Accidents, State of California. Spine. 1993;18:1736-1745.

[10] Jette AM. Physical disablement concepts for physical therapy research and practice. Phys Ther. 1994;74:380-386.

[11] International Classification of Impairments, Disabilities, and Handicaps. Geneva, Switzerland: World Health Organization; 1980.

[12] International Classification of Impairments, Disabilities, and Handicaps-2: International Classification of Functioning and Disability. Geneva, Switzerland: World Health Organization; 1999.

[13] Nagi SZ. Some conceptual issues in disability and rehabilitation. In: Sussman MB, ed. Sociology and Rehabilitation. Washington, DC: American Sociological Association; 1965:100-113.

[14] Schenkman M, Shipp KM, Chandler J, et al. Relationships between mobility of axial structures and physical performance. Phys Ther. 1996;76:276-285.

[15] Schenkman M, Hughes MA, Bowden MG, Studenski SA. A clinical tool for measuring functional axial rotation. Phys Ther. 1995;75: 151-156.

[16] Hagen KB, Harms-Ringdahl K, Enger NO, et al. Relationship between subjective neck disorders and cervical spine mobility and motion-related pain in male machine operators. Spine. 1997;22: 1501-1507.

[17] Riddle DL, Stratford PW. Use of generic versus region-specific functional status measures on patients with cervical spine disorders. Phys Ther. 1998;78:951-963.

[18] Marchiori DM, Henderson CN. A cross-sectional study correlating cervical radiographic degenerative findings to pain and disability. Spine. 1996;21:2747-2751.

[19] Jordan A, Mehlsen J, Ostergaard K. A comparison of physical characteristics between patients seeking treatment for neck pain and age-matched healthy people. J Manipulative Physiol Ther. 1997;20: 468-475.

[20] Jordan A, Bendix T, Nielsen H, et al. Intensive training, physiotherapy, or manipulation for patients with chronic neck pain: a prospective, single-blinded, randomized clinical trial. Spine. 1998; 23:311-318.

[21] Griegel-Morris P, Larson K, Mueller-Klaus K, Oatis CA. Incidence of common postural abnormalities in the cervical, shoulder, and thoracic regions and their association with pain in two age groups of healthy subjects. Phys Ther. 1992;72:425-431.

[22] Mayer T, Gatchel RJ, Keeley J, et al. A male incumbent worker industrial database, part II: cervical spinal physical capacity. Spine. 1994;19:762-764.

[23] Silverman JL, Rodriquez AA, Agre JC. Quantitative cervical flexor strength in healthy subjects and in subjects with mechanical neck pain. Arch Phys Med Rehabil. 1991;72:679-681.

[24] Ylinen J, Ruuska J. Clinical use of neck isometric strength measurement in rehabilitation. Arch Phys Med Rehabil. 1994;75:465-469.

[25] Youdas JW, Carey JR, Garrett TR. Reliability of measurements of cervical spine range of motion: comparison of three methods. Phys Ther. 1991;71:98-104.

[26] Scott J, Huskisson EC. Graphic representation of pain. Pain. 1976; 2:175-184.

[27] Revill SI, Robinson JO, Rosen M, Hogg MI. The reliability of a linear analogue for evaluating pain. Anaesthesia. 1976;31:1191-1198.

[28] Youdas JW, Garrett TR, Suman VJ, et al. Normal range of motion of the cervical spine: an initial goniometric study. Phys Ther. 1992;72: 770-780.

[29] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[30] Mayer TG, Barnes D, Kishino ND et al. Progressive isoinertial lifting evaluation, I: a standardized protocol and normative database. Spine. 1988;13:993-997.

[31] Mayer T, Gatchel R, Keeley J, et al. A male incumbent worker industrial database, part III: lumbar/cervical functional testing. Spine. 1994;19:765-770.

[32] Vernon H, Mior S. The Neck Disability Index: a study of reliability and validity. J Manipulative Physiol Ther. 1991;14:409-415.

[33] Fairbank JCT, Couper J, Davies JB, O'Brien JP. The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy. 1980;66:271-273.

[34] Ware JE Jr, Sherbourne CD. The MOS 36-item Short Form Health Survey (SF-36), I: conceptual framework and item selection. Med Care. 1992;30:473-483.

[35] McHorney CA, Ware JE Jr, Lu JF, Sherbourne CD. The MOS 36-item Short Form Health Survey (SF-36), III: tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care. 1994;32:40-66.

[36] Ware JE Jr, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual and Interpretation Guide. Boston, Mass: The Health Institute, New England Medical Center; 1993.

[37] Highland TR, Dreisinger TE, Vie LL, Russell GS. Changes in isometric strength and range of motion of the isolated cervical spine after eight weeks of clinical rehabilitation. Spine. 1992;17:S77-S82.

[38] Di Fabio RP, Boissonnault W. Physical therapy and health-related outcomes for patients with common orthopaedic diagnoses. J Orthop Sports Phys Ther. 1998;27:219-230.

[39] Jette DU, Jette AM. Physical therapy and health outcomes in patients with spinal impairments. Phys Ther. 1996;76:930-941.

[40] Ward MM, Lubeck D, Leigh JP. Longterm health outcomes of patients with rheumatoid arthritis treated in managed care and fee-for-service practice settings. J Rheumatol. 1998;25:641-649.

[41] Ware JE Jr, Brook RH, Rogers WH, et al. Comparison of health outcomes at a health maintenance organisation with those of fee-for-service care. Lancet. 1986;1:1017-1022.

Invited Commentary

Dr Hermann and Dr Reese are to be commended for their timely investigation of the widely accepted clinical assumption that variation in cervical impairment is associated with variation in function and disability among patients with cervical spine injuries. I agree with the authors that it is critical that our profession scrutinize the clinical hypotheses and assumptions that guide physical therapists in their diagnosis and treatment planning for patients who seek our services. It is through studies such as theirs that the field of physical therapy will continue to advance toward a more solid foundation of evidence-based practice. The disablement model proposed by Nagi[1] provides a useful framework with which to examine the hypothesized relationships in this investigation, although, as the authors note, other models such as the World Health Organization's International Classification of Impairments, Disabilities, and Handicaps-2 (ICIDH-2) formulation[2] are available to guide such investigations.

I want to focus my comments on 2 methodological aspects of Hermann and Reese's investigation: a statistical assumption and a measurement concern.

A fundamental analytical assumption in Hermann and Reese's article should be acknowledged explicitly because it affects the interpretation of their data. By using canonical correlational analyses, Hermann and Reese, like many clinical investigators, test the strength of linear associations among specific disablement concepts: impairments, functional limitations, and disability. Although this is a reasonable initial analytical step, analysts need to think beyond linear associations to the examination of more complex relationships among disablement concepts. There is growing evidence that many of the hypothesized relationships among disablement concepts can best be described as curvilinear versus simple linear relationships.[3] Several investigators,[4-6] for example, have recently demonstrated that in older adults the relationship between skeletal muscle force and subsequent function and disability can best be characterized as curvilinear, with identifiable clinical thresholds.

The identification of clinical thresholds in the association between specific impairments and function and disability could have important implications for physical therapy practice. Take the example of the relationship between skeletal muscle force and basic physical functions such as walking. If a lower-extremity muscle force threshold could be identified, it would allow for the identification of a minimum level of force needed for improved walking in different patient groups. It could also prove useful in the identification of specific subgroups of patients likely to benefit the most from clinical interventions. Such information could also assist clinicians in identifying evidence on the minimum level of muscle force above which an increase in physiological capacity does not translate into clinically important improvement in a function such as walking. Such data would call into question the common clinical (linear) assumption that improvement in force will logically lead to improvement in function. Such data would guide clinicians into incorporating use of clinical thresholds into their practice. Data such as these will emerge if researchers move beyond the examination of linear models that describe interrelationships among disablement concepts and develop more complex models that describe the precise form of these hypothesized relationships based on our clinical expertise and/or theoretical knowledge of the concepts under investigation.

Some of the measures used by Hermann and Reese illustrate one of the common measurement challenges encountered in conducting conceptually grounded clinical research such as theirs. Simply stated, many clinical measures available to researchers today are not grounded in clearly articulated, distinct clinical concepts such as those included in the disablement model. Without valid measures of distinct clinical concepts such as impairments, functional limitations, and disability, investigators run the risk of using measures that cut across several different clinical concepts, thus confusing the interpretation of their findings.

I believe several of the measures fielded in Hermann and Reese's study displayed confusing conceptual roots and that their questionable validity compromised the authors' ability to interpret their results. Let me illustrate this point with the Neck Disability Index (NDI) used by Hermann and Reese as one of their measures of disability in this study (a similar analysis could be performed on the Medical Outcomes Study 36-Item Short-Form Health Survey [SF-36], which also was used as a measure of disability in this study). No data are presented by the authors to support the validity of the NDI as a distinct measure of disability. If a measure has convincing face validity, this may not be a major flaw. If one examines the face validity of the NDI, however, it appears problematic for use in this study because it appears to encompass concepts other than disability.

Let me illustrate the concern directly. Five of the NDI items--personal care, work, driving, sleeping, and recreation--appear to represent examples of the concept of disability, defined by the authors as performance of socially defined roles accomplished within a cultural and physical environment. The conceptual roots for 5 other NDI items, however, are not so easy to classify and appear to represent concepts other than disability. Items such as lifting, reading, and concentration, for example, appear to better fit the concept of functional limitations, defined in the article as limitation in performance at the level of the whole person. Neck Disability Index items such as pain intensity and headaches may reflect the underlying concept of impairment, defined by the authors as abnormalities of structure or function at the level of organs and/or body systems. Thus, in this study, the use of the NDI as a distinct measure of disability separate from the concepts of functional limitations and impairment is likely flawed. The summary scores would appear to reflect variation in all 3 disablement concepts and not solely disability. This conceptual confusion may explain why the authors found a counterintuitive result. Canonical correlations between impairments and disability (.73) were stronger than observed correlations between the variable sets of functional limitations and disability (.54). The disablement model would clearly suggest that functional limitations should be more strongly related to disability than to impairments.[1,7]

One interpretation for this finding can be found in the authors' subanalyses. They report that the greatest contributing variable in the correlation between impairments and disability was found in the correlation between pain intensity and NDI scores. This comes as little surprise when one understands that pain intensity is a component of the NDI. At the very least, the choice of the NDI (or any other instrument) as a measure of disability requires data validating its conceptual foundation. Better yet, researchers need to develop and subsequently use clinical measures designed with a clear underlying conceptual foundation. Such measures are still lacking in disability research; thus, it is understandable why Hermann and Reese made the selections they did in this investigation.

Alan M Jette, PT, PhD Professor and Dean Sargent College of Health & Rehabilitation Sciences Boston University 635 Commonwealth Ave Boston, MA 02215 ([e-mail])

References

[1] Nagi SZ. Some conceptual issues is disability and rehabilitation. In: Sussman MB, ed. Sociology and Rehabilitation. Washington, DC: American Sociological Association; 1965:100-113.

[2] International Classification of Impairments, Disabilities, and Handicaps-2: International Classification of Functioning and Disability. Geneva, Switzerland: World Health Organization; 1999.

[3] Buchner D, deLateur B. The importance of skeletal muscle strength to physical function in older adults. Ann Behav Med. 1991;13:95-98.

[4] Ferrucci L, Guralnik JM, Buchner D, et al. Departures from linearity in the relationship between measures of muscular strength and physical performance of the lower extremities: the Women's Health and Aging Study. J Gerontol A Biol Sci Med Sci. 1997;52:M275-M285.

[5] Jette AM, Jette DU. Functional and behavioral consequences of sarcopenia. Muscle Nerve Suppl. 1997;5:S39-S41.

[6] Jette AM, Assmann SF, Rooks D, et al. Interrelationships among disablement concepts. J Gerontol A Biol Sci Med Sci. 1998;53:M395-M404.

[7] Verbrugge LM, Jette AM. The disablement process. Soc Sci Med. 1994;38:1-14.

Author Response

We would like to thank Dr Jette for his thoughtful commentary on our study, and we appreciate the opportunity to respond. It is through dialogue such as this that the science and practice of physical therapy can be advanced. Important points were raised regarding our statistical approach and choice of measures used.

Our primary goal was to determine the relationships among 3 sets of variables (impairment measures, functional limitation measures, and disability measures) in patients with cervical spine disorders. Our objective was to look at these variables in a multivariate sense and not simply examine pair-wise comparisons. Dr Jette is correct in pointing out that canonical correlations test the strength of linear associations among sets of variables. A statistical approach that tests for nonlinear associations among sets of variables would be difficult due to the multivariate nature of the problem. Testing for nonlinear (eg, curvilinear) relationships is more easily done using univariate correlations, where nonlinearities are more straightforwardly addressed. Linearity is always a simplification of the truth, but often a fairly good one. Our multivariate approach, although linear, may yield insights that would not otherwise be found using univariate correlations. However, we agree with Dr Jette that other statistical approaches should be used to further elucidate the nature of relationships among disablement concepts, especially in light of recent findings.[1,2] We also agree that determination of clinical thresholds from such studies could greatly affect practice.

We chose the measures used in this investigation because they are commonly used in daily practice and they have been studied previously. Dr Jette raises a very important issue in that many commonly used clinical measures purported to represent a single disablement concept actually represent more than one concept, thus making their interpretation problematic. We recognize that some of the measures we used (eg, Neck Disability Index) do not reflect distinct disablement concepts and that there is some overlap. The reader should be mindful of this as the results of our investigation are considered. We, too, believe that more valid measures representing clear and distinct disablement concepts must be developed in order for the science and practice of physical therapy to be advanced.

It is our hope that this investigation will stimulate further research and testing of clinical hypotheses as well as development of valid clinical measures that, as Dr Jette mentioned, will build a more solid foundation of evidence-based practice.

Karl M Hermann, PT, PhD C Shane Reese, PhD

References

[1] Ferrucci L, Guralnik JM, Buchner D, et al. Departures from linearity in the relationship between measures of muscular strength and physical performance of the lower extremities: the Women's Health and Aging Study. J Gerontol A Biol Sci Med Sci. 1997;52:M275-M285.

[2] Jette AM, Assmann SF, Rooks D, et al. Interrelationships among disablement concepts. J Gerontol A Biol Sci Med Sci. 1998;53:M395-M404.

KM Hermann, PT, PhD, is Director, The Therapy Center, College Station Medical Center, 1605 Rock Praire Rd, Ste 5, College Station, TX 77845 (USA) ([e-mail]). Address all correspondence to Dr Hermann.

CS Reese, PhD, is Technical Staff Member, Statistical Sciences, Los Alamos National Laboratory, Los Alamos, NM. Dr Reese was a doctoral candidate in the Department of Statistics, Texas A&M University, College Station, Tex, at the time of the study.

Concept/research design and writing were provided by Dr Hermann and Dr Reese; data analysis was provided by Dr Reese; and data collection, project management, fund procurement, and subjects were provided by Dr Hermann.

This study was approved by the Institutional Review Board for the Use of Human Subjects in Research at Texas A&M University.

This study was funded, in part, by a grant from the Orthopaedic Section, American Physical Therapy Association.

This article was submitted February 10, 2000, and was accepted August 31, 2000.