Sensitivity to change and responsiveness of the Global Physiotherapy examination (GPE-52) in patients with long-lasting musculoskeletal pain.

Background and Purpose. The sensitivity to change and the responsiveness of the Global Physiotherapy Examination (GPE-52) were examined in patients with localized and widespread long-lasting musculoskeletal pain. Subjects. All included patients were on sick leave and constituted the treatment group in a randomized controlled study. Two hundred forty-seven patients (64% women; mean age=43.6 years, SD=10.4) were included. At entry, the patients were categorized into 3 groups according to pain localization. Patients who returned fully to work were compared with those who did not. Methods. The GPE-52 has composite scores in 5 main domains--posture (8 tests), respiration (8 tests), movement (16 tests), muscle (12 tests), and skin (8 tests)--and was administered before and after 4 weeks of multidisciplinary outpatient treatment and at a 6-month follow-up examination. The first 61 patients also were re-examined after 18 months. Results. In this study, the GPE-52 and its 5 main domains were demonstrated to be sensitive to change, measured 6 and 18 months after treatment, in different groups of patients with long-lasting musculoskeletal pain. Responsiveness to important change, defined in this study as return to work, was found only for the total GPE-52 score and within the movement and respiration domains. Responsiveness to important change was greater in patients with localized pain than in patients with widespread pain. Discussion and Conclusion. Although the total GPE-52 score was sensitive to change in patients with long-lasting musculoskeletal pain, a shorter test battery including only the respiration and movement domain scores might be appropriate as an outcome measure in intervention studies for patients with long-lasting musculoskeletal pain. [Kvale A, Skouen JS, Ljunggren AE. Sensitivity to change and responsiveness of the Global Physiotherapy Examination (GPE-52) in patients with long-lasting musculoskeletal pain. Phys Ther. 2005;85:712-726.]

Key Words: Measurement, Movement, Muscle, Posture, Respiration, Responsiveness, Skin, Work status.

Despite our best efforts, the presence of musculoskeletal related pain problems is extensive and still increasing in Western society. Self-reported health complaints account for approximately 50% of all long-term sickness compensation and permanent disability in Northern European countries. (1,2) Moreover, the knowledge of concepts, determinants, and consequences of musculoskeletal health problems and related absence from work is still scanty, and much is yet to be learned about the specific disorders, contributing factors, and prevention strategies. (3) Particularly, patients who are still off work 12 weeks after symptom onset have increasing risk for long-term disability. (4) After 3 months with pain, early chronic pain syndrome seems to set in. This syndrome implies a "vicious circle" of psychological, behavioral, and biologic feedback loops, making successful treatment increasingly difficult. Many studies with return to work as an outcome measure have been based on patients' self-report, not physical assessment. (5) Single physical items used alone may not be sensitive or generic enough to be of extensive value. (6,7) There is still little consensus on what and how to assess, and what characterizes those who respond to treatment and those at risk for chronicity. (8,9)

In the Scandinavian countries, there is a physical therapy approach called Norwegian Psychomotor Physical Therapy (NPPT). (10,11) The NPPT approach is based on the theory that patients with long-lasting problems, physical or psychological, may react with general aberrations related to posture, respiration, and movement, as well as with muscular tension and skin changes. (12-14) From this therapy approach, an examination method called the Global Physiotherapeutic Muscle Examination (GPM) was developed by Sundsvold and coworkers (13-17) in order to document where and to what degree physical changes occur. The GPM-78 has 78 standardized tests that cover examination within 5 main domains: posture, respiration, movement, muscle, and skin. (16,18,19) These main domains are divided into 13 subdomains, each with 6 tests. Within the posture domain, postural items are inspected in both standing and supine positions. Within the respiration domain, inspiration amplitude of abdominal and thoracic areas is inspected in both standing and supine positions. Within the movement domain, passive range of motion of the head and the extremities is measured with a goniometer or in centimeters, flexibility and ability to relax are tested by means of passive movements performed by the therapist, and different active movements are performed by the patient. Included in the examination of the muscle domain are stretch and pressure palpation and the patient's verbal reaction to the stretching maneuver performed by the therapist. The skin domain is examined through pressure and stretch palpation in the same areas as in the examination of muscles.

Each item is scored according to how it relates to a predefined standard, corresponding to the score of 0. Table 1 illustrates the scoring scale. All except 3 items are bipolar; that is, they can be either increased or decreased. These items are scored on a 15-point scale ranging from -2.3 to +2.3 and denote the quantity and direction of deviation away from the predefined standard. The 2 directions give clinical information, whereas absolute values indicate magnitude of problems. Summations of the absolute test scores comprise the basis for a total score, 5 main domain sum scores, and 13 sub-domain sum scores.

The different tests covering examination of the upper-and lower extremities, trunk and head, and the positions of the patient and therapist are defined in a manual. (16,18,19) One fourth of the tests are non--side-specific, whereas the rest are performed on the left side of the body. Sundsvold and coworkers found small differences between the sides in the body in different patient groups with long-lasting problems (personal communication from MO Sundsvold). Assessing just one side of the body can be defended by findings in recent clinical and experimental studies related to posture, movement, and muscles.(20-23) If, however, a major impairment is present on the right side, relevant tests also will be performed on this side, but the data will not be included in the summation of scores.

Development of all GPM versions was based on patients who mainly had psychiatric or psychosomatic problems. Consequently, the construct validity was recently examined on data from people who were healthy and patients with long-lasting musculoskeletal pain. (24-28) In order to meet the demands for a reliable and valid examination method, a modified and shorter version of the GPM-78, called the Global Physiotherapy Examination (GPE-52), was developed. Based on data from 351 people, a form of confirmatory factor analysis called "structural equation modeling" was performed. (29) Confirmatory factor analysis can be used to revise and refine already existing instruments. (30) (It enables the researchers to test whether relationships expected on theoretical grounds actually appear in the data. One domain of the GPM-78 was examined at the time and then modified until the model fitted adequately, while still seeming clinically meaningful. Items that correlated too closely with each other or added little information were identified and omitted. The new version still has 5 main domains, but there are 4 items instead of 6 in each of the 13 subdomains. The items and domains of the GPE-52 are shown in the scoring form (Appendix). An examination takes about 30 minutes. Learning the complete GFM-52 takes 3 days, although clinical practice is recommended before it can be used reliably.

The GPE-52 has been found applicable to patients with all types of long-lasting musculoskeletal problems (eg, neck pain, low back pain, fibromyalgia) and is a useful generic examination instrument, especially for patients with chronic pain. (27) Posture, respiration, movements, and muscles function interactionally; a common pitfall is to overlook this interaction and concentrate only on the part of the body that the patient has reported as being most painful. (11,14) Previous research indicates high scores in patients with extensive physical or psychological problems. (31) Furthermore, it has been demonstrated that the GPE-52 discriminates between patients with long-lasting musculoskeletal problems and people who are healthy and between patients with localized versus widespread pain. (27) Previous studies (27,28) demonstrated mean total GPE-52 scores of around 34 (SD=6.4) in subjects who were healthy as compared with 47 (SD=8.0) in patients with long-lasting musculoskeletal pain problems. Studies including patients with psychiatric disorders have resulted in much higher GPM-78 mean values. (14) The maximum total GPE-52 score is 119.6 (2.3 x 52).

Measures intended to be used as instruments in evaluative studies should be sensitive to change (have the ability to measure any change in health status), as well as show responsiveness (the ability to detect meaningful change). (32, 33) Little seems to have been published regarding these aspects for comprehensive physical assessment batteries. The sensitivity to change and the responsiveness of the GPE-52 have not been formerly studied. The purpose of this study was to examine these aspects by measuring where and to what extent different groups of patients with long-lasting musculoskeletal pain change physical qualities over time. The GPE-52 may be useful for pretreatment examination; however, all domains may not be equally useful as outcome measures in intervention studies.

Method

Sensitivity to change of the GPE-52 and its 5 main domains was examined in patients who received multidisciplinary treatment, and responsiveness was studied by comparing physical findings in patients who returned to work with those who did not.

Subjects

The patients participated in a longitudinal randomized controlled trial with a large study sample. (34) The patients who were assigned randomly to the clinical intervention part constituted our study sample. Two hundred forty-seven consecutive patients were examined by the GPE-52 (159 women, mean age=44.2 years, SD=10.0, range=22-66; 88 men, mean age=42.5 years, SD=ll.1, range=24-64). The patients had been on sick leave from their work for an average of 3.3 months (SD=2, range=1-20) due to long-lasting musculoskeletal problems, but all had had pain for much longer.

Only patients without specific underlying systemic disease or pathology and without substance abuse were included, and a specialist in neurology diagnosed all patients as having problems related to the musculoskeletal system. (34) The patients' primary diagnosis could be grouped into 4 broad categories upon entry into the rehabilitation program: 9.3% had a diagnosis related to the neck or shoulder, 49.4% had a diagnosis related to low back pain, 22.7% had a diagnosis related to general muscular pain, and 18.6% had other diagnoses related to the musculoskeletal system. At the post-treatment examination 4 weeks later, 225 (64% women) were examined. Of those patients, 216 (68% women) were examined 6 months after discharge. The intention was to re-examine all patients 18 months after discharge, but because a new project started at the clinic, there was no capacity for physical re-examinations. Only the first 61 patients (67% women) were examined at 18 months. Work status data at 18 months, however, were available for about 90% of the patients.

The patients' work status was followed by the Norwegian national medical insurance, which registers sickness episodes longer than 2 weeks and all payments from the benefit scheme that covers all economically active individuals, except civil servants (about 8.5%-10%). Work status was registered on all of the 247 patients included in the study, except for 14 patients (civil servants), at both 6- and 18-month follow-up examinations. Subjects were categorized as either working or not working. Thus, people on vocational rehabilitation, partly working and partly on sick leave, or receiving any form of benefit were registered as not working. Patients whose work status data were lacking were excluded from relevant analyses.

Categorization of Patients According to Pain Drawing

Because many patients also had pain in other areas than the location of their main problem, a simple procedure was chosen to categorize the patients into 3 groups. (31,35) In order to map the distribution of pain, the patients marked on a pain drawing the area(s) that had been painful during the last 14 days before starting treatment. The number of squares marked was counted, and patients were categorized according to where they had marked pain. Patients who had pain above a horizontal line in the thoracolumbar region (T12) constituted group 1 (n=34: 79% women). Patients with pain only below the horizontal line constituted group 2 (n=58: 48% women). Those with pain both above and below the line were included in group 3 (n=134: 68% women), and such pain was defined as widespread. Markings either above or below the line were defined as localized. (31,36-39) Patients with localized pain, on average, marked 13.5 (SD=8.4) squares with pain, whereas those with widespread pain marked 30.3 (SD=18.1) squares with pain. For unknown reasons, a pain drawing was not registered for 21 patients.

Physical Findings

In this study, the patients were examined with the GPE-52 by a physical therapist upon admission, and the same physical therapist performed the follow-up examinations. Data from former examinations were not available to the physical therapist at the follow-up examinations, and the physical therapists were unaware of the pain category group to which the patient belonged. Three physical therapists participated. One of them used the measurement tool for about 6 years, whereas the others used it only occasionally for 1 year, after completion of a 1-week course. To ensure that the examiners were equally familiar with the examination procedures and criteria for scoring, they underwent 25 hours of intensive training and then performed an intertester reliability study of 19 people prior to this longitudinal study. Acceptable relative and absolute reliability were achieved. (28) The intraclass correlation coefficient (ICC [2,1]) was .91 for the total GPE-52 score, with a measurement error ([S.sub.w]) of 3.1. The ICC for the posture domain was .65, with an [s.sub.w] of 1.1. Within the respiration domain, the ICC was .60, due to limited range in scores, and the [s.sub.w] was 1.1.(24) Within the movement domain, the ICC was .89, with an [s.sub.w] of 1.6.(25) Within the muscle domain, the ICC was .83, with an [s.sub.w] of 1.4, and within the skin domain the ICC was .76, with an [s.sub.w] of 1.1.(26) The [s.sub.w] was generally [less than or equal to] 0.8 for the subdomains.

Intervention

When the evaluation process was completed, a multidisciplinary cognitive behavioral treatment program was developed, based on individually adapted functional restoration principles. (34) The intensive treatment period lasted 6 hours daily, 5 days a week for 4 weeks, with follow-up at regular intervals for 1 year. The program included treatment of patients in groups of 12, with daily intensive physical and ergonomic training, body awareness and relaxation training, psychological pain management, and teaching and talks about biopsychosocial issues. Workplace-based intervention was offered when deemed necessary.

Data Analysis

In this study, parametric statistics were used, as the sample was large, and the assumption of normal distribution was met for the GPE-52 (Kolmogorov-Smirnov test, P>.05, skewness <1). Analyses of variance (ANOVAs), with post hoc contrasts using the Scheffe procedure to determine pair-wise group differences, and t tests were used to test for initial differences among groups and between groups of patients who were and who were not working at the 6- and 18-month follow-up examinations. Sensitivity to change measured within each patient group was measured with paired t tests. At 18 months, the sample size did not allow for categorization between patients with localized versus widespread pain. Responsiveness to important change was examined by comparing those working and not at the 6- and 18-month follow-up examinations.

Logistic regression (LR) was used to predict the risk of not returning to work, where odds ratios (ORs) of >1 indicate an increased risk of not working. Unconditional LR was used with work status as the dependent variable and the GPE scores as the independent variable, (40) where the GPE scores were transformed into 3 categories (33 percentiles), using the lowest category as reference value. Effects were reported as ORs with corresponding confidence intervals (CIs) for the total sample at 6 and 18 months.

This study did not focus on possible ace

and sex differences regarding physical change over time, and these variables therefore were controlled for in the LR analysis. Both factors have been reported to affect complaints and physical findings, especially movement. (41-43) Because age did not affect the results in our study, this factor was not included in the final analysis. The simple subtraction of average posttreatment scores from pretreatment scores assumes that change will be the same, except for random error, for all patients. To control for the effect called "regression to the mean scores," the pretreatment scores also were entered as covariates. (40)

Responsiveness to change was also measured by calculating effect size (ES). Effect size is defined as the mean change from pretest to follow-up found in a variable, divided by the standard deviation of that variable at pretest. (44) Effect size was used to translate magnitude of change into a standard unit of measurement that might provide a clearer understanding of health status results. (32) An ES of 0.2 is considered small, 0.5 medium, and 0.8 represents a large ES. (45,46) An ES of 0.5 indicates that the mean between the 2 examinations differs by half a standard deviation.

Results

Descriptive data regarding sex and age for the whole patient sample and for the 3 groups categorized according to pain distribution are presented in Table 2, as well as mean and standard deviation for total GPE-52 scores at 4 points of time. Of the 247 consecutive patients initially included, 11 dropped out or were excluded after the first examination, and some were not examined either on discharge or at follow-up. The initial GPE-52 scores of these patients did not differ from those of the total sample. Patients with widespread pain (group 3) had higher total GPE-52 scores compared with patients with localized pain (group 1 and 2) both before treatment and at the 6-month follow-up examination (P<.05).

Sensitivity to Change

Figure 1 shows error bars with the mean and 95% CI of the sum score in the 5 main domains of the GPE-52 in the 3 patient groups at pretreatment, posttreatment, and 6-month follow-up examinations. A paired t test showed that the GPE-52 scores for all groups improved (P<.05) from the pretreatment examination to the posttreatment examination within all main domains, except within the skin domain in group 2. From the posttreatment examination to the 6-month follow-up examination, however, the only group that continued to improve was group 2 in the movement domain (t=2.7, df=49, P=.010) and the muscle domain (t=2.6, df=49, P=.014). Conversely, group 1 had an increase in physical aberrations in the skin domain (t=-2.7, df=26, P=.012), and group 3 had an increase in physical aberrations in the respiration domain (t=-3.7, df=115, P<.0001) from the posttreatment examination to the 6-month follow-up examination.

[FIGURE 1 OMITTED]

Responsiveness in Physical Findings Related to Work Status

The total sample that had GPE-52 scores at the pretreatment and 6-month follow-up examinations and work status data at 6 and 18 months constituted 204 patients. At 6 months, 46% of the total sample was registered as working, increasing to 48% at 18 months. At the 6-month follow-up examination, 44% of the patients in group 1 were working, increasing to 59% at 18 months. At 6 months, 54% of the patients in group 2 were working, increasing to 62% at 18 months. At 6 months, 44% in group 3 were working, decreasing to 41% at 18 months.

Total sample. In the total sample at the pretreatment examination, there were no differences in total mean GPE-52 score between the 2 groups categorized according to work status at 6 months, but those subjects who had returned to work at 6 months initially had fewer problems in the movement domain than those who had not returned to work at 6 months (t=-2.1, df=202, P=.034). Logistic regression analysis showed that the risk of not having returned to work at 6 months increased with higher GPE-52 total scores at that time (Tab. 3). Using scores below 37.5 as the reference value, the OR for not having returned to work was 2.2 (95% CI=0.974.80) when GPE-52 total scores at the 6-month follow-up examination were between 37.5 and 45; the OR for not having returned to work was 4.6 (95% CI=1.5-13.9) when scores at the 6-month follow-up examination were higher than 45. An increased risk of not working also was found with increasing scores in the movement domain. Patients with a movement domain score between 9.7 and 14.5 were 2.8 (95% CI=1.3-6.1) times more likely to be sick-listed compared with the reference group (scores <9.7), and the OR was 3.9 (95% CI=1.4-11.0) with scores above 14.5. The P value for high scores in the respiration domain was .10. In the respiration, movement, and muscle domains, the magnitude of change from the pretreatment examination to the 6-month follow-up examination was slightly larger measured with ES in those subjects who were working at 6 months compared with those who were still sick-listed (0.2-0.3 higher ES).

Group 1. Patients who returned fully to work at 6 months had initially higher scores in the muscle domain (P=.03) compared with those who did not return to work, but otherwise there were no differences between those who had and those who had not returned to work at 6 months (Tab. 4). At 6 months, there was no difference in muscular findings when we controlled for age, sex, and pretreatment values. The pretreatment and 6-month follow-up ES values are listed in Table 4 and are illustrated in Figure 2. Although both categories (working versus not working) had an ES of 0.6 in the total GPE-52 score, patients who had returned to work had changed more within the respiration domain (ES=0.9 versus 0.0) and the muscle domain (ES=0.7 versus 0.3). A medium-sized ES also occurred in the nonworking group in the movement domain score (Fig. 2), and at 6 months the nonworking group had about the same mean sum score as the working group (Tab. 4).

[FIGURE 2 OMITTED]

Group 2. Those who later returned to work had initially more aberrations in the movement domain than those who did not return to work (P=.031) (Tab. 4). Patients working at 6 months also improved from initial scores (P<.05). The ES values (Fig. 2) indicated that both patients who returned to work and those who did not return to work had improved from the pretreatment examination to the 6-month follow-up examination, but the working group had a larger magnitude of change in total GPE-52 score (Tab. 4) (ES=1.0 versus 0.5) in the movement domain (ES=1.0 versus 0.4), the muscle domain (ES=0.7 versus 0.3), and the respiration domain (ES=0.5 versus -0.2) compared with the nonworking group.

Group 3. Before treatment, the group later registered as working had lower mean total GPE-52 scores (P=.037) and fewer aberrations in movement (P=.001) compared with those who were not working (Tab. 4). At the 6-month follow-up examination, there were no longer any differences when we controlled for age, sex, and pretreatment values. Small differences in ES values were registered in group 3 between patients who were working and those who were not working (Fig. 2).

Patients who were examined at 18 months and registered as working at that time (n = 24) had no physical differences at the pretreatment examination compared with those who were later registered as not working (n = 35). At 18 months, patients who were working had fewer aberrations in the respiration and movement domains compared with the nonworking group, and the magnitude of change from the pretreatment examination to 18-month follow-up examination is illustrated in Figure 3. Odds ratios for not working were 4.2 (95% CI = 0.9-18.6) for the middle category of GPE-52 total scores compared with the reference group and 5.5 (95% CI = 0.9-34.5) for the highest category (Tab. 3). In the respiration domain, the risk was 2.9 (95% CI = 0.7 - 11.7)

[FIGURE 3 OMITTED]

for not having returned to work with scores in the middle category and 5.7 (95% CI = 1.4-22.8) in the highest category compared with the reference group (who had scores <5.8). In the movement domain, the OR for not working at 18 months was 5.3 (95% CI = 0.7-38.6) in patients with scores in the highest category (>14) compared with patients with scores in the lowest category (<9.4).

Discussion

The present study showed that the GPE-52 was sensitive to change in patients with long-lasting musculoskeletal pain. Significant change in the physical findings as measured by the GPE-52 and its 5 main domains was found after 4 weeks of treatment. The movement domain was most sensitive, and the skin domain was least sensitive. At all time intervals, patients with localized pain (groups 1 and 2) had fewer physical aberrations, measured with the total GPE-52 sum score, compared with patients with widespread pain (group 3). For patients with localized pain, the total score gradually approached the mean score reported in people who are healthy. (27) The results can be seen as a demonstration of construct validity for the GPE-52, (32) in accordance with psychomotor physical therapy theory (11,14) and other studies of characteristics of patients with localized versus widespread pain. (36,37,47,48) Although me change was statistically significant for all domains, the mean change was larger than the measurement error between testers only in total GPE-52 scores and in the movement domain scores. On average, little change was found in the respiration, posture, and skin domains. Within the muscle domain, the measured mean change was almost as large as the [s.sub.w].

Responsiveness of the GPE-52 was demonstrated to some extent. Patients who returned to work obtained a larger magnitude of physical change compared with those still receiving work benefits; however, as illustrated in Figure 2, the difference in change between those who were working and those who were not working was most obvious in those patients with localized pain. The movement and respiration domains were shown to be responsive to clinically important change, and the findings in our study support the application of these domains as outcome measures in intervention studies. The movement domain showed interesting, although mixed, results. For the total sample, responsiveness was demonstrated, as indicated by the logistic outcome, and the odds of not having returned to work increased with increasing score categories. The findings were significant at 6 months, but at 18 months the sample size was too small, showing the same trend without statistical significance in the movement domain. The muscle domain scores did not show significant responsiveness in the logistic regression analysis, and, measured with ES, the difference between patients who were working and those who were not working was small. The posture domain scores differed little between patients who were working and those who were not working, and the skin domain showed no sign of responsiveness. Patients with long-lasting musculoskeletal pain, however, may not change much regarding posture and skin, as measured in this study.

In patients with pain initially localized only to the upper body (group 1), clinically important change, in this study measured in relation to work status, seemed to have occurred within the respiration and muscle domains, measured from the pretreatment examination to the 6-month follow-up examination (Fig. 2). Patients who later returned to work initially had more muscular problems, but obtained a moderate to large degree of change, indicated by effect size, in contrast to patients who remained on sick leave. Restrained basic respiration has been seen as a cause of recruiting auxiliary musculature, over time perpetuating problems. (49) One cannot establish what comes first, but it is interesting to note that a positive change within the muscle and respiration domains was associated with work status for patients with pain in the upper body. However, patients who did not return to work changed more in the movement domain than patients who did return. The sample size in group 1 was small, and such findings warrant further investigation.

Patients with pain localized to the lower body (group 2) had physical findings that improved, even after treatment, resulting in a quicker return to work compared with patients with pain in the upper body. Furthermore, at 18 months, the return-to-work rate was higher in both groups who initially had localized pain compared with those with widespread pain (61% versus 41%), which conforms to previous findings. (36,50) As in group 1, patients who later returned to work initially had more physical aberrations, but in group 2 this was found in the movement domain. In patients with localized pain, this finding could indicate a greater potential for change and successful outcome. Similar findings have been reported in a study of patients with low back pain, where people who returned to work initially had longer distances from fingertip to floor than people who did not return to work. (51) It must be emphasized, however, that they were not more likely to return to work because they had room for improvement.

Patients with widespread pain (group 3) who later returned to work had fewer physical aberrations at the pretreatment examination, as measured with the GPE-52, mainly within movement domain, compared with patients who remained sick listed. At 6 months, patients who returned to work had still fewer physical aberrations compared with the patients who did not return to work, especially within the movement domain, although not significantly, when we controlled for the pretreatment values. Our findings show that patients with widespread pain had a high degree of physical problems, as measured by the GPE-52, compared with patients with localized pain, and this group had a lower percentage of patients who returned to work, which is in accordance with former studies. (31,36,52)

The effectiveness of treatment, or lack thereof, was not directly measured in this study. This study also did not focus on details of the treatment program or how the patients adhered to the treatment program, although such adherence could have affected the individual patient's response. The control group participating in the randomized controlled trial from which our study sample was taken (34) was not examined with the GPE-52, and it is therefore difficult to infer that the changes seen were related to treatment. The design of this study, however, enables the physical therapist to map magnitude of change in physical findings, both within patients and between the different pain localization groups. (32) The selection biases were minimized by the sampling procedure and the design of the randomized study. (34) The findings reported in this article, therefore, should have some external validity in relation to domains that may be sensitive and responsive in evaluation of change in patients with long-lasting musculoskeletal pain. Except for sex and age, we did not focus on or control for other factors that may influence change such as education, activity level, cognitive factors, negative health beliefs, pain behavior, and psychosocial factors. (5,53,54)

Because no study has been performed to demonstrate magnitude of physical change by means of the GPE-52 or similar test batteries, our study can be conceived as a parameter estimation study, which is an important part of the validation process of new instruments. (32) However, a test battery showing statistically significant change may not reflect clinically meaningful change. The results in this study suggest that only the respiration and movement domains, and possibly the muscle domain, can be appropriate as outcome measures in intervention studies for patients with long-lasting musculoskeletal pain. For many physical therapists, a barrier might be the time it takes to administer the complete GPE-52. Clinical practice has shown that it takes little time to train for reliable responses in use of the respiration domain and especially the movement domain, (24,25) and it takes about 10 minutes to use this part of the instrument. The complete GPE-52, however, could be used as a generic screening battery, because it examines body structures and body function and covers a wide range of musculoskeletal disorders. It helps to answers questions such as: "What degree of physical problems does a patient have, and where do these problems occur?" and "Could the physical findings be possible perpetuating factors that should be addressed in treatment?" Generic scales allow comparisons across different disorders, severities of disorders, interventions, and perhaps even demographic and cultural groups. (44)

Conclusion

In this study, the GPE-52 and its 5 main domains were demonstrated to be sensitive to change, measured at 6 and 18 months after treatment, in different groups of patients with long-lasting musculoskeletal pain. Responsiveness to important change, in this study defined as return to work, however, was only found within the movement and respiration domains, although a trend also was seen within the muscle domain, and more in patients with localized pain than in patients with widespread pain. In patients with localized pain, a high degree of physical aberrations at the pretreatment examination, indicated by high GPE scores, seemed to indicate a potential for change and successful outcome. The opposite, however, was found for patients with widespread pain.

The total GPE-52 score used alone only reflects degree of physical aberration, but it does not indicate where the physical aberration exists. The sum scores in the different domains, however, are much more informative than a total score. A shorter version of this examination instrument might still be adequate in future studies, especially in patients with localized pain. However, if some domains were to be omitted, such as the posture and skin domains, the scale might lose some of its validity because possible aberrations as well as unforeseen effects of an intervention may go undetected.

References

(1) Eriksen HR, Svendsrod R, Ursin G, Ursin H. Prevalence of subjective health complaints in the Nordic European countries in 1993. Eur J Pub Health. 1998;8:294-298.

(2) Linaker CH, Walker-Bone K, Palmer K, Cooper C. Frequency and impact of regional musculoskeletal disorders. Baillieres Best Pract Res Clin Rheumatol. 1999;13:197-215.

(3) Brage S. Musculoskeletal Health Problems and Sickness Absence: An Epidemiological Study of Concepts, Determinants, and Consequences [doctoral thesis]. Oslo, Norway: Faculty of Medicine, Institute of General Practice and Community Medicine, University of Oslo; 1998.

(4) Hagen KB, Kvien TK, Bjorndal A. Musculoskeletal pain and quality of life in patients with noninflammatory joint pain compared to rheumatoid arthritis: a population survey. J Rheumatol. 1997;24:1703-1709.

(5) Fishbein DA, Rosomoff HL, Goldberg M, et al. The prediction of return to the workplace after multidisciplinary pain center treatment. Clin J Pain. 1993;9:3-15.

(6) Waddell G, Somerville D, Henderson I, Newton M. Objective clinical evaluation of physical impairment in chronic low back pain. Spine. 1992;17:617-628.

(7) Thomas E, Silman AJ, Papageorgiou AC, et al. Association between measures of spinal mobility and low back pain: an analysis of new attenders in primary care. Spine. 1998;23:343-347.

(8) Frank JW, Brooker AS, DeMaio SE, et al. Disability resulting from occupational low back pain, part II: what do we know about secondary prevention? A review of the scientific evidence on prevention after disability begins. Spine. 1996;21:2918-2929.

(9) Jensen IB, Bodin L, Ljungqvist T, et al. Assessing the needs of patients in pain: a matter of opinion? Spine. 2000;25:2816-2823.

(10) Bulow-Hansen A. Psykomotorisk behandling: historikk. Sjukgymnasten. 1981;12:8-11.

(11) Bunkan BH, Thornquist E. Psychomotor therapy: an approach to the evaluation and treatment of psychosomatic disorders. In: Hegna T, Sveram M, eds. International Perspectives in Physical Therapy. 5: Psychological and Psychosomatic Problems. Vol. 5. London, United Kingdom: Churchill Livingstone; 1990:45-74.

(12) Houge NH. Physiotherapy in certain aspects of psychosomatic medicine. Psychother Psychosom. 1979;32:302-305.

(13) Sundsvold MO, Vaglum P, Ostberg B. Movements, lumbar and tempomandibular pain, and psychopathology. Psychother Psychosom. 1981;35:1-8.

(14) Sundsvold MO, Vaglum P. Muscular pains and psychopathology: evaluation by the GPM method. In: Michel TH, ed. International Perspectives in Physical Therapy. 1: Pain. London, United Kingdom: Churchill Livingstone; 1985:18-47.

(15 Sundsvold MO. Muscular tension and psychopathology. Psychother Psychosom. 1975;26:219-228.

(16) Sundsvold MO, Vaglum P, Denstad K. Global Fysioterapeutisk Muskelundersokelse. Oslo, Norway: private publishing company; 1982.

(17) Kvale A, Sundsvold MO. Global Fysioterapeutisk Muskelmetode: en vurdering av intertester reliabilitet (Evaluation of reliability in the Global Physiotherapeutic Muscle Examination). Fysioterapeuten. 1991; 57:29-34.

(18) Sundsvold MO, Baastad AF. Global Fysioterapeutisk Muskelundersokelse [video]. Oslo, Norway: ACEM; 1994.

(19) Sundsvold MO. Global Fysioterapeutisk Muskelundersokelse. Til Bruk i Klinisk Arbeid og Forskning [manual]. Privately published by author. 1996.

(20) Bunkan BH, Ljunggren AE, Opjordsmoen S, et al. What are the basic dimensions of movements? A psychometric evaluation of the Comprehensive Body Examination, III. Nord J Psychiatry. 2001;55: 33-40.

(21) Friis S, Bunkan BH, Ljunggren AE, et al. What are the basic dimensions of body posture? An empirical evaluation of the Comprehensive Body Examination, I. Nord J Psychiatry. 1998;52:319-326.

(22) Larsson R, Zhang Q, Cai H, et al. Psychosocial problems and chronic neck pain: microcirculation and electromyography of the trapezius muscles at static loads. Journal of Musculoskeletal Pain. 1998;6: 65-76.

(23) Stohler CS, Kowalski CJ, Lund JP. Muscle pain inhibits cutaneous touch perception. Pain. 2001;92:327-333.

(24) Kvale A, Johnsen TB, Ljunggren AE. Examination of respiration in patients with long-lasting musculoskeletal pain: reliability and validity. Advances in Physiotherapy. 2002;4:169-181.

(25) Kvale A, Johnsen TB, Ljunggren AE. Examination of movement in patients with long-lasting musculoskeletal pain: reliability and validity. Physiotherapy Research International. 2003;8:36-52.

(26) Kvale A, Ljunggren AE, Johnsen TB. Palpation of muscle and skin: is this a reliable and valid procedure in assessment of patients with long-lasting musculoskeletal pain? Advances in Physiotherapy. 2003;5: 122-136.

(27) Kvale A, Skouen JS, Ljunggren AE. Discriminative validity of the Global Physiotherapy Examination-52 in patients with long-lasting musculoskeletal pain versus healthy persons. Journal of Musculoskeletal Pain. 2003;11:23-36.

(28) Kvale A. Measurement Properties of a Global Physiotherapy Examination in Patients With Long-Lasting Musculoskeletal Pain [doctoral thesis]. Bergen, Norway: Section for Physiotherapy Science, Department of Public Health and Primary Health Care, Faculty of Medicine, University of Bergen; 2003.

(29) Kline RB. Principles and Practice of Structural Equation Modeling. New York, NY: Guilford Press; 1998.

(30) Floyd FJ, Widaman KF. Factor analysis in the development and refinement of clinical assessment instruments. Psychol Assess. 1995;7: 286-299.

(31) Kvale A, Ellertsen B, Skouen JS. Relationships between physical findings (GPE-78) and psychological profiles (MMPI-2) in patients with long-lasting musculoskeletal pain. Nord J Psychiatry. 2001;55: 177-184.

(32) Stratford PW, Spadoni G, Kennedy D, et al. Seven points to consider when investigating a measure's ability to detect change. Physiother Can. 2002;54:16-24.

(33) Kirshner B, Guyatt G. A methodological framework for assessing health indices. J Chron Dis. 1985;38:27-36.

(34) Haldorsen EM, Kronholm K, Skouen JS, Ursin H. Multimodal cognitive behavioral treatment of patients sicklisted for musculoskeletal pain: a randomized controlled study. Scand J Rheumatol. 1998;27: 16-25.

(35) Skouen JS, Vilsvik D, Ellertsen B. A new and simple classification system of chronic low back pain patients. In: Proceedings of the 24th Annual Meeting of the International Society for the Study of the Lumbar Spine. 1997:108-109.

(36) Andersson HI, Ejlertsson G, Leden I, Rosenberg C. Characteristics of subjects with chronic pain, in relation to local and widespread pain report: a prospective study of symptoms, clinical findings and blood tests in subgroups of a geographically defined population. Scand J Rheumatol. 1996;25:146-154.

(37) Natvig B, Bruusgaard D, Eriksen W. Localized low back pain and low back pain as part of widespread musculoskeletal pain: two different disorders? A cross-sectional population study. J Rehabil Med. 2001;33: 21-25.

(38) Trygg T, Lundberg G, Rosenlund E, et al. Personality characteristics of women with fibromyalgia and of women with chronic neck, shoulder, or low back complaints in terms of Minnesota Multiphasic Personality Inventory and Defense Mechanism Technique modified. Journal of Musculoskeletal Pain. 2002;10:33-55.

(39) Natvig B, Rutle O, Bruusgaard D, Eriksen WB. The association between functional status and the number of areas in the body with musculoskeletal symptoms. Int J Rehabil Res. 2000;23:49-53.

(40) Streiner DL, Norman GR. Health Measurement Scales: A Practical Guide to Their Development and Use. New York, NY: Oxford University Press; 1995:163-180.

(41) Battie MC, Bigos SJ, Sheehy A, Wortley M. Spinal flexibility and individual factors that influence it. Phys Ther. 1987;67:653-658.

(42) Moll J, Wright V. Measurement of spinal movement. In: Jayson MI, ed. The Lumbar Spine and Back Pain. 2nd ed. London, United Kingdom: Pitman Publishing Ltd; 1980:157-183.

(43) Sullivan MS, Dickinson CE, Troup JD. The influence of age and gender on lumbar spine sagittal plane range of motion: a study of 1,126 healthy subjects. Spine. 1994;19:682-686.

(44) Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status. Med Care. 1989;27:S178-S189.

(45) Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. London, United Kingdom: Lawrence Erlbaum Associates Publishers; 1988:1-27.

(46) Tingey RC, Lambert MJ, Burlingame GM, Hansen NB. Developments in the assessment of clinically significant change: assessing clinical significance, proposed extensions to method. Psychotherapy Research. 1996;6:109-123.

(47) Croft P, Burt J, Schollum J, et al. More pain, more tender points: is fibromyalgia just one end of a continuous spectrum? Ann Rheum Dis. 1996;55:482-485.

(48) White KP, Harth M. The occurrence and impact of generalized pain. Baillieres Best Pract Res Clin Rheumatol. 1999;13:379-389.

(49) Krogstad BS. Myofunctional Aspects of Craniomandibular Disorders [thesis]. Oslo, Norway: Department of Prosthetic Dentistry and Stomatognathic Physiology, Dental Faculty, University of Oslo; 1992.

(50) Hagen KB, Thune O. Work incapacity from low back pain in the general population. Spine. 1998;23:2091-2095.

(51) Strand LI, Ljunggren AE, Haldorsen EMH, Espehaug B. The impact of physical function and pain on work status at 1-year follow-up in patients with back pain. Spine. 2001;26:800-808.

(52) Hunt IM, Silman AJ, Benjamin S, et al. The prevalence and associated features of chronic widespread pain in the community using the "Manchester" definition of chronic widespread pain. Rheumatology (Oxford). 1999;38:275-279.

(53) Ohlund C, Lindstrom I, Eek C, et al. The causality field (extrinsic and intrinsic factors) in industrial subacute low back pain. Scand J Med Sci Sports. 1996;6:98-111.

(54) Haldorsen EM, Kronholm K, Skouen JS, Ursin H. Predictors for outcome of a multi-modal cognitive behavioural treatment program for low back pain patients-a 12-month follow-up study. Eur J Pain. 1998;2:293-307.

A Kvale, PT, Phi), is Assistant Professor, Section for Physiotherapy Science, Department of Public Health and Primary Health Care, Faculty of Medicine. University of Bergen, Bergen, Norway. Address all correspondence Io Dr Kvale al Section fin- Physiotherapy Science, Faculty of Medicine, University of Bergen, Kalfarveien 31, 5018 Bergen, Norway ([e-mail]).

JS Skouen, MD, PhD, is Clinical Director, The Outpatient Spine Clinic, Department of Physical Medicine and Rehabilitation, Haukeland University Hospital, Bergen, Norway, and Assistant Professor, Section for Physiotherapy Science, Department of Public Health and Primary Health Care, Faculty of Medicine, University of Bergen.

AE Ljunggren, PT, PhD, is Professor, Section for Physiotherapy Science, Department of Public Health and Primary Health Care, Faculty of Medicine, University of Bergen.

All authors provided concept/research design, writing, and data collection and analysis. Dr Skouen provided project management, subject recruitment, facilities, and consultation. Dr Ljunggren provided institutional liaisons. The authors acknowledge Tove Dragesund, PT, MSc, and Hildegunn Lygren, PT, MSc, for assistance with data collection.

This study was approved by file Regional Ethics Committee and the National Data Inspectorate and was performed according to the Helsinki Declaration.

A poster and abstract of this work were presented at the 10th World Congress on Pain; August 17-22, 2002; San Diego, Calif.

This article was received December 19, 2003, and was accepted February 22, 2005.

Table 1.

Schematic View of the Application of the Scoring Scale in the Global
Physiotherapy Examination (GPE-52)(a)

Domain                  -
                        -2 (Much)                 - 1 (Somewhat)
                        -2.3 -2.0 -1.7            -1.3 -1.0 -0.7

Posture (8 tests)       Extended

Respiration (8 tests)   Increased movement
                        Too much movement

Movement (16 tests)     Too large range
                        Too flexible

Muscle (12 tests)       Too stretchable
                        No reaction/indifferent
                        reaction
                        Too compressible

Skin (8 tests)          Too stretchable
                        Too compressible

Domain                  Scale                     +
                        0                         + 1 (Somewhat)
                        -0.3 0 +0.3               +0.7 +1.0 +1.3

Posture (8 tests)                                 Flexed

Respiration (8 tests)                             Decreased movement
                                                  Paradoxical movement

Movement (16 tests)     Good                      Reduced range
                                                  Too restrained

Muscle (12 tests)                                 Too short
                                                  Painful

                                                  Too dense, too solid

Skin (8 tests)                                    Too short
                                                  Too distended, too
                                                  compact

Domain                  + 2 (Much)
                        +1.7 +2.0 +2.3

Posture (8 tests)

Respiration (8 tests)

Movement (16 tests)

Muscle (12 tests)

Skin (8 tests)

(a) Positive and negative scores give clinical information, whereas
absolute values indicate magnitude of aberrations away from the
standard. The scoring scale has 2 main groups on both sides of the main
group 0. The standard, also called "ideal" or "good", lies in the
middle lies in the middle of main group 0. Each main group is
subdivided into 3 subgroups. To avoid use of several decimals (eg,
0.333 and 0.6(57), the scores have been rounded off (eg, 0.3 and 0.7).
Thus, the maximum possible score for each test is [+ or -]2.3, and t
and the total possible sum score with 4 tests is 9.2 (maximum possible
score is 18.4 with 8 tests, 27.6 with 12 tests, and 36.8 with 16
tests). The manual(19) contains standardized definitions for each item
of how to transfer the quality and quantity (in centimeters or degrees)
into a score.

Table 2.

Descriptive Data for the Entire Sample of Patients With Long-Lasting
Musculoskeletal Pain (N=247) and When Categorized into 3 Groups
According to Pain Distribution (a)

                         Total
                         Sample          Group 1

Age pretreatment (y)
  X (SD)                 43.6            41.5
  Range                  22-66           27-64
N (women, men)
  Pretreatment           247 (159, 88)   34 (27, 7)
  Posttreatment          225 (145, 80)   30 (24, 6)
  6-mo follow-up         216 (147,69)    30 (24, 6)
  18-mo follow-up         61 (41, 20)     7 (5, 2)
Global Physiotherapy
  Examination (GPE-52)
  X (SD)
  Pretreatment           46.9 (8.0)      44.5 (6.7)
  Posttreatment          41.7 (8.3)      38.5 (7.1)
  6-mo follow-up         41.3 (8.4)      39.9 (6.8)
  18-mo follow-up        40.5 (10.9)     39.5 (9.5)

                         Group 2         Group 3
Age pretreatment (y)
  X (SD)                 41.2            45.1
  Range                  25-66           22-64
N (women, men)
  Pretreatment           58 (28, 30)     134 (91, 43)
  Posttreatment          56 (27, 29)     125 (84, 41)
  6-mo follow-up         53 (27, 26)     121 (87, 34)
  18-mo follow-up        13 (6, 7)        34 (24, 9)
Global Physiotherapy
  Examination (GPE-52)
  X (SD)
  Pretreatment           44.5 (7.4)      48.1 (8.1)
  Posttreatment          40.0 (7.7)      43.3 (8.5)
  6-mo follow-up         38.9 (7.6)      42.8 (9.0)
  18-mo follow-up        34.9 (9.6)      44.2 (11.0)

(a) Group 1-pain marked above a horizontal line in the thoracoluntbar
region, group 2-pain marked below the horizontal line in the
thoracolumhar region, group 3-pain marked above and below the
horizontal line. There was no pain drawing for 21 of the patients.

Table 3.

Logistic Regression Analysis (a) : Odds Ratios (ORs)b With 95%
Confidence Intervals (95% CI) and P Values for Not Having
Returned to Work at 6-Month and 18-Month Follow-up Examinations

Tests                    Score

6-mo follow-up (n=204)
 Total GPE-52 score
  Category 1             [less than or equal to] 37.5
  Category 2             37.6-45.0
  Category 3             [greater than or equal to] 45.1
Respiration
  Category 1             [less than or equal to] 5.6
  Category 2             5.7-6.8
  Category 3             [greater than or equal to] 6.9
Movement
  Category 1             [less than or equal to] 9.6
  Category 2             9.7-14.5
  Category 3             [greater than or equal to] 14.6
18-mo follow-up (n=59)
 Total GPE-52 score
  Category 1             [less than or equal to]36
  Category 2             36.1-45
  Category 3             [greater than or equal to] 45.1
Respiration
  Category 1             <5.8
  Category 2             5.8-7.0
  Category 3             [greater than or equal to] 7.1
Movement
  Category 1             [less than or equal to] 9.4
  Category 2             9.5-14
  Category 3             [greater than or equal to] 14.1

Tests                    OR    95% CI      P

6-mo follow-up (n=204)
 Total GPE-52 score
  Category 1             1.0
  Category 2             2.2   0.97, 4.8   .059
  Category 3             4.6   1.5, 13.9   .007
Respiration
  Category 1             1.0
  Category 2             1.2   0.6, 2.6    .616
  Category 3             1.9   0.9, 3.9    .101
Movement
  Category 1             1.0
  Category 2             2.8   1.3, 6.1    .008
  Category 3             3.9   1.4, 11.0   .009
18-mo follow-up (n=59)
 Total GPE-52 score
  Category 1             1.0
  Category 2             4.2   0.9, 18.6   .063
  Category 3             5.5   0.9, 34.5   .067
Respiration
  Category 1             1.0
  Category 2             2.9   0.7, 11.7   .131
  Category 3             5.7   1.4, 22.8   .013
Movement
  Category 1             1.0
  Category 2             1.3   0.3, 5.6    .713
  Category 3             5.3   0.7, 38.6   .100

(a) Logistic regression was used with work status as the dependent
variable and the Global Physiotherapy Examination (GPE) scores as the
independent variable, where the GPE scores were transformed to 3
categories (33 percentiles), using the lowest category as reference
value. Sex and pretreatment values were controlled for. Only the
domains that tended to be significant are listed.

(b) ORs >1 indicates risk of not returning to work.

Table 4.

Means and Standard Deviations for the Global Physiotherapy Examination
(GPE-52) and Its 5 Main Domains, P Values From Logistic Regression
Analysis, and Effect Sizes (a)

                              Before Treatment

                                           Not
                              Working      Working
Patient Groups                [bar.X]      [bar.X]      P
n=All(Working)                (SD)         (SD)

Group 1: n=27 (12 working)
Total GPE-52 score            45.7 (6.7)   43.1 (6.7)   .174
  Posture                      5.9 (1.2)    5.8 (1.6)   .277
  Respiration                  7.3 (1.0)    7.0 (1.8)   .559
  Movement                    12.8 (4.5)   13.2 (3.5)   .990
  Muscle                      13.9 (2.1)   11.9 (2.6)   .030 (b)
  Skin                         5.8 (2.0)    5.3 (1.9)   .477
Group 2: n=50 (27 working)
Total GPE-52 score            46.3 (7.2)   43.0 (6.0)   .078
  Posture                      5.6 (1.9)    6.1 (1.5)   .155
  Respiration                  6.9 (2.0)    6.3 (1.3)   .230
  Movement                    15.5 (3.9)   13.0 (3.7)   .031 (b)
  Muscle                      12.2 (2.3)   12.1 (2.1)   .559
  Skin                         6.0 (2.7)    5.6 (2.0)   .230
Group 3: n=116 (51 working)
Total GPE-52 score            46.0 (7.9)   49.6 (8.2)   .037 (b)
  Posture                      6.1 (1.7)    5.8 (1.9)   .314
  Respiration                  6.4 (1.6)    6.9 (1.5)   .226
  Movement                    14.4 (4.8)   17.6 (4.7)   .001 (b)
  Muscle                      13.4 (2.4)   13.3 (2.7)   .890
  Skin                         5.6 (2.2)    6.0 (2.4)   .458

                              6-Month Follow-up

                                           Not
                              Working      Working
Patient Groups                [bar.X]      [bar.X]
n=All(Working)                (SD)         (SD)          P

Group 1: n=27 (12 working)    41.4 (8.4)    39.1 (6.0)   .444
Total GPE-52 score             5.2 (2.1)     5.1 (1.9)   .940
  Posture                      6.5 (1.2)     7.0 (1.3)   .238
  Respiration                 10.9 (5.1)    10.7 (2.6)   .230
  Movement                    12.5 (2.2)    11.0 (2.2)   .344
  Muscle                       6.3 (1.8)     5.3 (1.8)   .164
  Skin
Group 2: n=50 (27 working)    38.9 (7.2)    40.1 (7.1)   .024 (b)
Total GPE-52 score             5.5 (2.0)     5.4 (l.7)   .461
  Posture                      5.9 (1.4)     6.5 (1.4)   .011 (b)
  Respiration                 11.5 (3.8)    11.4 (4.0)   .011 (b)
  Movement                    10.6 (2.5)    11.5 (2.7)   .133
  Muscle                       5.4 (2.5)     5.4 (2.4)   .836
  Skin
Group 3: n=116 (51 working)   40.4 (9.4)    44.8 (8.3)   .235
Total GPE-52 score             5.2 (1.9)     5.3 (1.8)   .495
  Posture                      6.5 (1.4)     6.6 (1.3)   .584
  Respiration                 11.5 (4.9)    15.0 (4.7)   .096
  Movement                    12.0 (2.9)    12.3 (2.5)   .309
  Muscle                       5.1 (1.8)     5.6 (2.3)   .682
  Skin

                              Effect Size

Patient Groups                             Not
n=All(Working)                Working      Working

Group 1: n=27 (12 working)     0.6         0.6
Total GPE-52 score             0.5         0.4
  Posture                      0.9         0.0
  Respiration                  0.4         0.7
  Movement                     0.7         0.3
  Muscle                      -0.3         0.0
  Skin
Group 2: n=50 (27 working)     1.0         0.5
Total GPE-52 score             0.1         0.5
  Posture                      0.5        -0.2
  Respiration                  1.0         0.4
  Movement                     0.7         0.3
  Muscle                       0.2         0.1
  Skin
Group 3: n=116 (51 working)    0.7         0.6
Total GPE-52 score             0.5         0.3
  Posture                      0.0         0.1
  Respiration                  0.6         0.5
  Movement                     0.6         0.4
  Muscle                       0.2         0.2
  Skin

(a) Patients included were examined by GPE-52, at both pretreatment and
6-month posttreatment examinations, and were classified according to
work status registered at the 6-month follow-up examination. Age and
sex were controlled for at both the pretreatment examination and the
6-month follow-up examination; at 6 months, the pretreatment value of
the GPE-52 also was entered as a covariate. Effect size indicates
magnitude of change from the pretreatment examination to the 6-month
follow-up examination (calculated as the mean change from pretreatment
examination to the 6-month follow-up examination, divided by the
standard deviation of the pretreatment scores).

(b) P<.05 (P value at the pretreatment and 6-month follow-tip
examinations between patients who
were working and those who were not working at five months after
treatment).