Disability and functional status inpatients with low back pain receiving workers' compensation: a descriptive study with implications for the efficacy of physical therapy.

There are a significant number of people who experience low back pain (LBP) as a result of work-related injury.[1-3] Industrial injuries involving the low back often result in a prolonged absence from work.1 In addition, considerable financial cost is incurred through workers' compensation claims.[2]

Work-related low back injuries are influenced by demographic and social variables, as well as psychological and physical factors.[4] To obtain valid assessments of the efficacy of treatment, it is necessary to identify the population characteristics that might influence outcome. There is a commonly held assumption, for example, that patients receiving workers' compensation report pain or exaggerate their symptoms to continue receiving benefits. This assumption is not uniformly supported in the literature. Greenough and Fraser[5] reported that compensated workers were more disturbed compared with noncompensated workers. Nonorganic signs (eg, widespread sensitivity to light touch, disproportionate verbalization of pain symptoms), however, do not appear to be correlated with return to work or with resolution of symptoms.[6] In addition, psychologic variables do not appear to be predictive of the level of disability.[7,8] Leavitt et al[9] found that the number of patients with unconfirmed organic disease or psychologic disturbance was similar for groups receiving workers' compensation compared with those not receiving compensation. These findings were consistent with those of another study[10] showing no difference in pain severity or psychological disturbance for compensated patients compared with noncompensated patients.

Aside from the psychological variables, there are multiple factors that influence the level of disability or the return to work. Frymoyer[7] found that the duration of disability was a strong predictor of the likelihood that a patient would return to work. If a patient was disabled for more than 6 months, then the probability of returning to work was 50%.[7] Lancourt and Kettelhut[4] reported that high Oswestry scores (a high level of self-perceived disability), a history of leg pain, and short tenure on the job were some of the factors that predicted the failure to return to work for patients off work less than 6 months prior to the initial evaluation. Patients' understanding of their medical condition may also be an important variable in predicting return to work. Lacroix et al[8] noted that 94% of their patients with a good understanding of their condition returned to work, compared with 33% of those with a poor understanding of their disorder. Physical impairment, in general, may influence the level of disability, but there is no clear relationship between disability status and a specific physical impairment.[11-13]

Conservative management has been advocated in a number of studies for patients with LBP who receive workers' compensation benefits.[14-18] Some authors,[19-21] however, have reported poor or equivocal outcomes following nonsurgical interventions. The studies showing positive outcomes with physical therapy have reported return-to-work rates on the order of 80%. Saal and Saal[14] found that the return-to-work rate for patients treated with physical therapy was 86% (n=11). This finding was comparable to the return-to-work rates reported in studies of surgical treatment for disk disease.[22,23] Mayer et al[16] presented a descriptive analysis of the return-to-work rates for patients receiving multidisciplinary interventions that included physical therapy, psychological assessment, and counseling. Eighty-six percent of the patients participating in that treatment program (n=62) returned to work, compared with 55% (n=33) of a comparison group (subjects who were denied insurance coverage for treatment).[16] Similar return-to-work rates (approximately 80%) were found in subsequent studies that utilized nonspecific physical therapy interventions.[15,17] Physical therapy in these studies[14-21] was considered nonspecific because (1) multiple interventions were used and (2) the treatments were not designed to treat a specific area of the spine (ie, lifting instructions, trunk strengthening, range of motion, coordination exercises, and endurance training).

In contrast to the studies demonstrating positive clinical outcomes with nonspecific physical therapy, Oland and Tveiten[19] reported that only 23% of their patients treated with this approach (n=15) returned to work. Fredrickson et al[20] found that fewer than 40% of their patients who were not working at the time of the initial evaluation actually returned to work. Wiesel et al[21] reported that only half of their patients receiving workers' compensation and nonspecific physical therapy returned to work and that subsequent attendance at work was intermittent.

The equivocal findings regarding nonspecific physical therapy may have been due to a small therapeutic effect for the patients receiving treatment or to a lack of sensitivity in the outcome measures used to determine a clinically important difference.[24] Bigos et al[25] have suggested that the time lost from work represents a primary outcome for the study of work-related back injuries. Absence from work, however, was not correlated with pain intensity and level of disability in a study by Roland and Morris.[11] They found that self-reports of pain intensity and disability were more sensitive indicators of long-term outcome compared with absence from work. That is, absence from work, or return to work, could be due to factors independent of the level of disability (eg, financial/economic considerations).

The Oswestry disability score[26] has frequently been used to assess the level of perceived disability in patients receiving workers' compensation.[4,5,14,17] In spite of wide use of die Oswestry disability questionnaires, the information reported for studies with positive clinical outcomes is incomplete. Saal and Saal[14] did not report the initial Oswestry scores for their "nonoperative" treatment group. A reference point for evaluating outcome at the termination of their study, therefore, was not available. Mayer et al[16] stated that the Oswestry questionnaire was used only at follow-up, but they reported no data. Hazard et al[17] reported only the initial and discharge Oswestry scores for their intervention group, omitting the Oswestry scores for their comparison groups at discharge. Greenough and Fraser[5] presented a complete Oswestry profile as part of a retrospective study that showed higher levels of disability for compensated patients compared with noncompensated patients. These authors[5] concluded that the payment of compensation delayed recovery from low back injury, but the types of treatment provided to the patients were not described. When considering these studies[5,14,16,17] in aggregate, there is no clear relationship between physical therapy intervention, the level of disability, and return to work.

Evaluation of the effect of multiple physical therapy treatments requires a serial analysis of the disability profiles for patients receiving workers' compensation. The purposes of this study were (1) to describe the level of disability, physical impairment, and rate of return to work for compensated workers before and after the completion of a physical therapy program with multiple interventions and (2) to evaluate the influence of compliance, chronicity, and pain distribution on the level of disability, impairment, and work status.

Method

Subjects

The subjects in this study were patients attending a private practice clinic in Minneapolis, Minn, between March and October 1993. The protocol for this study did not differ from routine clinical practice. As a standard procedure, all patients seeking treatment at the clinic were asked to sign a consent form. Patients who signed a consent form were included in this analysis if (1) their primary complaint was LBP and (2) they were receiving workers' compensation. All patients meeting these two criteria were evaluated without preselection.

A total of 138 patients (64 male, 54 female), ranging in age from 17 to 63 years (X=38, SD=10), were evaluated. The relevant demographic characteristics of the subjects are summarized in Table 1. Several cases had missing (unrecorded) data, but were analyzed with a "flag" inserted into the data set to indicate a missing entry. Unrecorded data accounted for the fluctuating number of subjects analyzed for any given variable (Tab. 1).

[TABULAR DATA OMITTED]

A diagnosis for each patient was provided by the referring physician, and the diagnostic categories are presented in Table 2. The criteria for each diagnostic category were not solicited from the referring physician, and "diagnosis" was not used as a grouping variable. The diagnostic categories are presented only for descriptive purposes.

Table 2. Number of Patients in Each
Diagnostic Category According to Referring
Physician

Diagnostic Category       N

Herniated disk             52
Spondylosis                 4
Mechanical low back pain   48
Sacroiliac dysfunction      1
Facet joint degeneration    4
Postoperative pain         10
Lumbar strain              16
Sciatica                    1
Lumbar instability          2

Physical Therapists Providing

Care

Fifteen physical therapists (average number of years of experience=13, SD=6, range=5-27) conducted the initial evaluations, treatment, and discharge assessments. Prior to the study, each therapist completed a postgraduate internship program sponsored by the clinic that involved a combination of classroom, laboratory, and supervised clinical evaluations and treatments. The goal of the internship program was to ensure that newly employed therapists had a minimum level of proficiency utilizing manual therapy techniques. The postgraduate internship involved up to 6 months of didactic and supervised clinical work and included specialized training in joint mobilization[27] and muscle energy techniques.[28] In addition, each therapist completed several continuing education courses in manual therapy and was required to pass a practical examination. This examination required a demonstration of assessment and treatment procedures using manual therapy techniques on a nondisabled volunteer. The practical examination was administered and proficiency was rated by a senior therapist at the clinic.

Description of outcomes

The assessment and treatment protocols used in as study were identical to those implemented for the routine care of all patients entering the clinic with a primary complaint of LBP. During the initial assessment, the subject indicated the duration of LBP, pain history, and any cointerventions (eg, medications, chiropractic care). Three disability/impairment outcome measures were evaluated: the Oswestry disability score (expressed as a percentage), fingertip-to-floor distance (FTFD) (in centimeters), and maximum isometric pain-free lift (in pounds).

The Oswestry Low Back Pain Questionnaire is a self-report instrument that evaluates perceived disability in 10 areas (subscales), which include pain intensity and the ability to sit, stand, sleep, and complete activities related to personal hygiene.[26] Each subscale is rated from 0 (no limitation) to 5 (severe limitation). The total maximum score (50) is doubled and reported as a percentage of the patient's perceived disability. Higher scores indicate greater disability.

Serial FTFD measurements have been used previously to assess the efficacy of treatment for LBP.[18,29-31] We conducted a single trial to measure each subject's FTFD. Previous work[32] has shown that patients who have an increase in pain on the first attempt show greater FTFD (less motion) compared with patients who do not report pain during forward bending. The first attempt, therefore, provides a measurement that appears to be sensitive to symptom provocation. Each subject was asked to stand in a comfortable position and to bend forward as far as possible without bending the knees. The distance of the middle finger to the floor was assessed using a tape measure marked in 0.10-cm increments.

Maximum isometric lift (MIL) has been used as one measure of work capacity,[16,33] and normative data have been published by gender for subjects without back injury working in industry.[31] During the MIL assessment, each subject was tested using a spring-loaded dynamometer(*) attached to a level wooden platform. The range of operation was 0 lb (O N) to 660 lb (2,932 N), with 5-lb (22.22-N) increments. The subject was instructed to stand on the platform, pick up the dynamometer handle, and pull upward on the handle as much as possible without causing pain or modifying existing symptoms. The dynamometer handle was fixed at 38.1 cm (15 m) from the platform base in order to standardize lift height. A single trial was collected for the MIL because the therapists were seeking a conservative measure of this variable, which was minimally influenced by pain and repetitive learning.

(*)Baseline back-leg chest dynamometer, Best Priced Products Inc, Box 1174, White Plains, NY 10602.

In addition to the disability/impairment outcomes, two work status measures were evaluated: (1) whether or not the patient was working in some capacity or (2) whether or not the patient was released to work in some capacity at the time of the assessment/reassessment. Working and release-to-work status were determined by the referring physician in consultation with the patient. The decision to return to work was based primarily on the amount of improvement in the patient's symptoms and functional capacity. Release to work might differ from actual work status, because the patient, might have been able to work in some capacity but could not secure employment that would accept personnel with physical limitations.

All outcome measures were recorded together with the diagnosis and symptom characteristics in the patient's chart. The disability/impairment and work-related outcome measures were obtained at the initial assessment (INA), 1 month from the initial assessment (1MO), and at the time of discharge from the clinic (DC). Time to discharge was variable between subjects. The range of days spanning INA to DC was 1 to 162 days (X=32, SD=29; n=132).

Reliability of Outcome

Assessments

The Oswestry questionnaire has been shown to provide a reliable measure of the patient's perception of disability.[26] Fairbank et al[26] tested 22 patients with LBP on consecutive days and reported a test-retest correlation coefficient of r =.99. This finding agreed with the reliability estimates for measurements obtained with similar self-report instruments for patients with LBP.[11,34]

Measurement of FTFD has been shown to be highly reliable when used with patients who have LBP.[18,32] Gauvin et al32 reported intratherapist and intertherapist intraclass correlation coefficients of .98 and .95, respectively, for FTFD measurements. Provocation of symptoms in some patients during forward bending did not influence the reliability of the measurement.

Normative data for MIL have been reported in research sponsored by the National Institute for Occupational Safety and Health.[32] Chaffin[32] calculated the reliability of MIL measurements obtained for 446 male and 105 female subjects using a test-retest coefficient of variation. The range of variation that could be expected with repeated measurements was [+ or -] 13%,[33] but it should be noted that the coefficient of variation is not considered to be a generalizable measure of reliability. Zeh et al[35] found that more than 80% of the between-subject variability ([r.sup.2]) on mean static lift capacity for three trials was explained by the first lift. They concluded that a single lift provided a reasonable and predictable measure of static lift capacity.

Treatment Procedures

Multiple physical therapy interventions were used and included heating modalities, passive stretching, spinal mobilization, active exercise, and lifting instructions. Specific treatment decisions were guided by the summary of significant findings for each patient. The significant findings generally included a description of pain/symptom-provoking motions, an assessment of passive and active motion of the spine, posture, and strength.

Active exercise program were developed based on the pattern of pain/symptom provocation during active motion, the pattern of muscle weakness in the trunk and lower extremities, and the pattern of limitations in passive motion. In general, patients received active exercise programs that were designed to facilitate spinal motion and reduce symptom severity. The extension protocol proposed by McKenzie,[36] for example, was widely used to provide extension mobilization exercises for patients reporting relief of symptoms-with back extension. In addition, extension mobilization exercises were used with patients who had a limitation of motion in this direction.

The decision to provide passive joint mobilization (graded articulation) was based on the finding of segmental hypomobility that was thought to be related to the primary complaint. Individual therapists provided a range of mobilization procedures, but these treatments were generally administered as described by Maitland.[27] Muscle contraction-relaxation techniques were also used to improve motion as described by Greenman.[28] Briefly, patients were positioned by the therapist to the point of resistance to spinal motion and then asked to provide a nonpainful isometric contraction. Each patient was repositioned by the therapist after each contraction so that the posture of the patient could be maintained at the point of resistance. This procedure was designed to facilitate gradual increments in spinal range of motion. The number, magnitude, and duration of the contraction-relaxation cycles varied according to the tolerance of each patient. Therapeutic modalities (ie, ultrasound, electrical stimulation) were provided at the discretion of each therapist and were used to reduce pain and facilitate the patient's response to treatment.

All patients in this study were referred to a functional activities training class. This class consisted of a 1-hour video that reviewed basic spinal anatomy, mechanics, various types of spinal injury, and common spinal surgeries (eg, segmental decompression). The video was followed by a 90-minute group session that reviewed proper lifting and load-carrying techniques. In addition, proper sitting and standing postures were demonstrated as well as the use of back supports and Cushions.[37]

Data Analysis

General analysis for description of outcome. Descriptive data and outcome measures contained in the patient's chart were recorded in a spreadsheet by clinic staff. Patient identifiers (ie, name, address, social security number) were deleted to protect the identity of each patient. Paired t tests were performed to compare disability/impairment outcomes at the INA versus the 1MO assessment and at the INA versus the DC assessment. The "paired" disability/impairment outcome measures were first evaluated without regard to other factors (eg, compliance). Because it is known that static lift capacity differs across gender,[33] an analysis of this variable was done separately for male and female patients. The number of patients working or released to work in some capacity at the INA and the DC assessment were simply described with percentages.

Grouping patients for descriptive analysis. For the purpose of further description, the patients were grouped according to their level of compliance (high, low), chronicity (acute, chronic), and leg symptoms (present, absent). Paired t tests were performed within each of these groups to compare disability/impairment outcomes at the INA versus the 1MO assessment and at the INA versus the DC assessment. The number of patients working or released to work in some capacity at the INA and the Dc assessment was evaluated within groups (ie, high versus low compliance) using a chi-square analysis with a Yates correction for unequal group sizes.[38]

Compliance was calculated as the actual number of treatments divided by the number of scheduled treatments. Patients were deemed non-compliant if they canceled a treatment or failed to arrive for a scheduled treatment. When patients were non-compliant, they were contacted by phone and reminded of their next appointment. High compliance was defined as a score of [greater than or equal to] 80%, whereas low compliance was defined as a score of <80%.

Chronicity was defined by the number of consecutive days with at least some pain. Two groups were formed based on duration of pain: an "acute" group with less than 6 weeks of pain on consecutive days prior to the INA and a "chronic" group with greater than 6 weeks of pain on consecutive days prior to the INA.

Leg symptoms were defined as pain, numbness, and/or paresthesias below the gluteal fold. If subjects had any one or a combination of these symptoms, they, were grouped in a "leg symptom" category. If these symptoms were absent, the subjects were considered not to have leg symptoms.

Description of outcomes in patients lost to follow-up. The extent of initial disability (Oswestry score) and impairment (FTFD and MIL) was evaluated in patients lost to follow-up. Two-group independent t tests were used to compare each disability/impairment outcome measure for those subjects who were followed versus those who were not followed. This approach was necessary to determine whether those subjects lost to follow-up had greater initial involvement than those who were followed.

All data analyses were done using the BMDP statistical computing software (version PC90).[39] The probability level for determining statistical significance was adjusted to account for multiple comparisons. In order to be statistically significant at the .05 level, paired comparisons on each dependent variable (INA versus 1MO and INA versus DC) had to have a probability value of <.025. This significance level was based on a Bonferroni correction factor.[39](pp197-199)

Results

General Description of Outcomes

Clinically meaningful and statistically significant improvements were found for each outcome when comparing the INA with the 1MO reassessment and the INA with the DC evaluation (Tab. 3 and Fig. 1). Between the INA and the 1MO assessment, there was a 7% reduction in mean disability, a 13-cm improvement in mean forward bending, and a 47-lb (209-N) increase in average static lift capacity. Between the INA and the DC evaluation, there was a 10% decrease in mean disability, a 15-cm increase in mean forward bending, and a 69-lb (307-N) improvement in average MIL.

[TABULAR DATA OMITTED]

Normative data for MM were previously established for each gender.[31] The average MIL for both male and female subjects was below normative values at the INA (Fig. 2). At the 1MO and DC assessments, however, each gender group surpassed the normative values for MIL for healthy workers (Fig. 2).

At the INA, 64 out of 137 patients (47%) were working in some capacity and 82 out of 135 patients (61%) were released to work in some capacity. At the time of discharge, 24 out of 42 patients (57%) were working in some capacity and 30 out of 40 patients (75%) were released to work in some capacity.

Description of Outcomes by

Compliance

For the high-compliance group, between the INA and the 1MO assessment, there was a 10% reduction in mean disability, a 13-cm improvement in mean forward bending, and a 50-lb (222-N) increase in average static lift capacity (Tab. 4). Between the INA and the DC evaluation, there was a 12% reduction in disability, a 15-cm increase in forward bending, and a 74-lb (329-N) increase in MIL (Tab. 4, Fig. 3). All improvements in the disability/impairment outcome measures at each reassessment for the high-compliance group were statistically significant (Tab. 4).

For patients with low compliance, there were no significant improvements in mean Oswestry scores at each follow-up compared with the INA (Tab. 4, Fig. 3). The difference between the mean Oswestry scores at the INA and each reassessment were on the order of 4% and were lower, by a factor of 3, compared with the differences in mean scores found for the high-compliance group. The improvements at the 1MO assessment in mean forward bending (14 cm) and average MIL (40 lb [178 N]), however, were similar to the improvements seen in the high-compliance group. Between the INA and the DC evaluation, the increase in mean forward bending for the low-compliance group (17 cm) was comparable to that of the high-compliance group. The low-compliance group, however, showed a much smaller increase in average static lift capacity (30 lb [133 N]) compared with the high-compliance group (Tab. 4). Only the 1MO FTFD and MIL measures for the low-compliance group showed statistically significant improvements compared with the INA (Tab. 4).

There were no significant differences in work status by level of compliance. The percentage of patients working or released to work in some capacity in the high-compliance group was similar to the percentage in the low-compliance group at both the INA and the DC evaluation (Tab. 5).

Description of Outcomes by

Chronicity

For the group with acute symptoms, between the INA and the 1MO assessment, there was an 11% reduction in mean disability, a 17-cm improvement in mean forward bending, and a 54-lb (240-N) increase in average static lift capacity (Tab. 6). Between the INA and the DC evaluation, there was an 18% reduction in mean disability, a 23-cm increase in mean forward bending, and a 93-lb (413-N) increase in average MIL. Each reassessment mean Oswestry score showed a statistically significant reduction in disability compared with the mean Oswestry score at the INA (Tab. 6, Fig. 4). Each follow-up FTFD and MIL value obtained for those patients with acute symptoms showed statistically significant improvements compared with the INA data (Tab. 6).

For the chronic group, between the INA and the 1MO assessment, there was only a 4% reduction in mean disability and only a 9-cm improvement in mean forward bending. The magnitude of these improvements was much lower in the chronic group than in the group with acute symptoms. The increase in average MIL at the

1MO assessment (40 lb) in the chronic group, however, was similar to that seen in the group with acute symptoms. Between the INA and DC evaluation, there was only a 7% reduction in disability, an 11-cm increase in forward bending, and a 51-lb (227-N) increase in MIL for the chronic group. The magnitudes of all of these improvements for the chronic group at the DC evaluation were smaller than those observed in the acute group (Tab. 6).

[TABULAR DATA OMITTED]

There was a significantly higher percentage of patients in the acute group who were working in some capacity at the time of the INA ([[chi].sup.2]=4.06, df=1, P<.05) and at the time of the DC evaluation ([[chi].sup.2]=3.61, df=1, P<.05) compared with those patients with chronic symptoms (Tab. 5). In addition, a greater percentage of patients with acute symptoms were released to work in some capacity at the INA compared with those patients with Chronic Symptoms ([[chi].sup.2]=6.62, df=1, P<.05). There was no difference in the percentage of patients released to work in some capacity between the acute and chronic groups at the time of the DC evaluation.

[TABULAR DATA OMITTED]

Description of Outcomes by Leg

Symptoms

For those patients with leg symptoms, between the INA and the 1MO assessment, there was only a 2% reduction in disability, a 13-cm improvement in forward bending, and a 41-lb (182-N) increase in static lift capacity (Tab. 7). Between the INA and the DC evaluation, there was an 8% reduction in disability, a 10-cm increase in forward bending, and a 76-lb (338-N) increase in MIL. The reduction in mean disability for the group with leg symptoms at the 1MO or DC assessment compared with the INA was not statistically significant (Tab. 7, Fig. 5). The improvement in mean forward bending at the 1MO assessment compared with the INA, as well as each follow-up mean MIL value, showed significant improvement in static lift capacity compared with the initial MIL values (Tab. 7).

[TABULAR DATA OMITTED]

For patients who did not have leg symptoms, in contrast, each mean reassessment Oswestry score showed a significant improvement in level of disability compared with the INA score. There was a 12% reduction in mean disability between the INA and the 1MO assessment and a 13% reduction in mean disability between the INA and the DC evaluation (Tab. 7). There was a 12-cm improvement in mean forward bending and a 54-lb (240-N) increase in average MIL at the 1MO assessment compared with the INA. These improvements in FTFD and static lift capacity for the non-leg symptom group at the 1MO assessment were similar in magnitude to the improvements seen in the group with leg symptoms. At the DC evaluation, the increase in average MIL (68 lb [302 N]) for the non-leg symptom group was also similar to the increase in the group with leg symptoms. The increase in mean forward bending for the group without symptoms at the DC evaluation (23 cm), however, was over twice as large as the improvement in mean forward bending for the group reporting leg symptoms. There was no significant difference in the frequency of return or release-to-work status between the subjects with leg symptoms compared with the subjects without leg symptoms (Tab. 5).

Description of Outcomes in

Patients Lost to Follow-up

For 66% (n=91) of the original patient sample (n=138), either the patients were lost to follow-up or data were not recorded at the 1MO reassessment. For 80% (n=110) of the original sample, either the patients were lost to follow-up or data were unrecorded in the chart at the time of the DC evaluation. The mean initial Oswestry scores and the mean FTFD and MIL values for those patients lost to follow-up, however, were not significantly different from the mean data of those available for reassessment (fig. 6). The age of the patients lost to follow-up at the 1MO assessment (X=32 years, SD=16) and at the DC evaluation (X=33 years, SD=14) was also not significantly different from that of the patients followed at the 1MO assessment (X=35 years, SD= 15) or at the DC evaluation (X=34 years, SD=20).

Discussion

Patients receiving workers' compensation appeared to benefit from a physical therapy program with multiple interventions (Fig. 1). The magnitude of improvement for those who had limited exposure to treatment (low compliance) was lower than for those who complied with physical therapy (Tab. 4, Fig. 3). Return-to-work status, however, was not different between the low- and high-compliance groups (Tab. 5). This finding might indicate that patients were attempting to return to work regardless of the seventy of their disability. Return to work may be influenced by many factors.[11,16] Mayer et al,[16] for example, noted that many patients settle their insurance claims and make the decision to return to work based on economic considerations, without any change in their physical status.

At the completion of the physical therapy program, 24 out of 42 patients (57%) were working in some capacity and 30 out of 40 patients (75%) were released to work in some capacity. The actual return-to-work rate was very close to the total group return-to-work rate reported by Sachs et al[15] at 6 months after discharge (59.3%). These authors utilized a nonspecific physical therapy program, whereas our treatments were based on a clinical evaluation of each patient. These results suggest that nonspecific physical therapy (ie, endurance training or general conditioning) might be as effective in returning patients to work as our multiple-intervention but specific physical therapy program. The return-to-work rates in our study and in the study by Sachs et al, however, were lower than the rates reported for soldier studies[14,16,17] (over 80%). The most likely reason for the discrepancy is the difference in patient selection. In our study, as in the study by Sachs et al,[15] we did not preselect patients. All subjects receiving workers' compensation were eligible to participate. Mayer et al[16] and Hazard et al[17] accepted only patients with chronic symptoms who were not candidates for surgery. Saal and Saal[14] studied only patients with verifiable disk lesions. It is difficult to compare our results with those of Oland and Tveiten[19] because only 5% of their sample were receiving workers' compensation.

A rating of disability severity has been recommended by Fairbank et al.[26] According to their scale, Oswestry scores from 41% to 60% were severe, whereas scores from 21% to 40% indicated moderate disability. The average Oswestry scores in our study at the initial presentation were severe (41% for the INA/1MO pair) to moderate (37% for the INA/DC pair). At the time of the DC evaluation, the average Oswestry scores were below the midpoint of the moderate range (Tab. 3). The reduction of disability in our study was on the order of 10% (Tab. 3). The improvement in disability from the INA to the DC evaluation was comparable to the reduction in mean Oswestry scores reported by Hazard et al[17] for patients receiving a comprehensive rehabilitation program that included psychological and behavioral counseling. Physical therapy in our study included multiple cointerventions such as heating modalities, spinal mobilization, exercise, and lifting instructions. Our program, however, did not include psychological intervention.

There was no attempt to evaluate the efficacy of various cointerventions, because we viewed physical therapy with multiple interventions as the "entity" under investigation. There were two reasons for selecting this approach. First, patients receiving workers' compensation represent a broad range of diagnostic categories, including postsurgical cases (Tab. 2). It was difficult to "preselect" patients for the study of specific interventions when there was no preexisting rationale to limit the types of interventions that are considered as current standards of care. The use of multiple cointerventions is a common aspect of physical therapy practice in our experience. Second, the lack of relationships between specific physical impairments and the level of disability for patients with acute[11] as well as chronic back pain[12,13] suggests that little would be gained from restricting physical therapy to a single intervention. Given the complexities of LBP[4,12] it is unlikely that a single specific intervention would have been effective for all the patients we were required to treat.

It is of interest to compare our multiple-intervention approach with the "category-specific" approach reported by Delitto et al.[40] These authors preselected patients with acute LBP who showed symptom improvement with back extension and symptom magnification with back flexion.[dagger] The treatment (extension exercise and mobilization) was specific to this "category" of patients. Delitto et al[40] found that the treatment group had approximately a 10% improvement in the mean Oswestry score from the initial assessment to the reassessment 5 days later. Although the preselection of patients based on signs and symptoms may lead to a more focused treatment, a broader population of patients may not have symptoms that are compatible with a narrow classification category. Generalization of Delitto and colleagues' approach[40] to other types of patients with LBP, therefore, may be limited.

The similarity between the extent of improvement in disability scores for those patients treated with a "category-specific intervention,"[40] a physical therapy program with multiple interventions (Fig. 1), and a comprehensive transdisciplinary rehabilitation program[17] raises an issue concerning the range of interventions that are necessary to achieve an optimal outcome. Further analysis of this issue is needed because of the differences among studies related to chronicity, duration of follow-up, and patient selection.

Chronicity of symptoms also appeared to influence outcome. Subjects with acute symptoms prior to the initial evaluation had lower disability scores and higher return-to-work rates compared with subjects with chronic pain (Tabs. 5, 6; Fig. 4). These results are consistent with Frymoyer's[7] finding that the duration of disability was a strong predictor of outcome. One possible reason for this finding is that there may be less illness behavior associated with acute versus chronic back pain.[12,13] In addition, this finding supports the notion that early intervention may influence therapeutic outcome in a positive way.

Patients who had lower-extremity symptoms at the initial presentation also had greater disability compared with patients with localized or paraveterbral symptoms (Fig. 5). Leg symptoms have previously been shown to be one of the factors that predict failure to return to work,[4] but our results did not support this finding (Tab. 5). Here, as with the compliance group, return to work was not related to the level of self-perceived disability. The distribution of leg symptoms associated with low back injury, however, is considered an important factor in guiding some types of treatment[36] and may be related to prognosis.[41,42] In addition, pain distribution may be a factor that influences the response to manual therapy.[43]

Various impairment-specific outcome measures have been identified in patients with LBP.[12,44] The amount of trunk motion has been shown to increase during recovery from low back injury.[15-18] Our results replicate this previous finding. Motion of the spine increased significantly at the 1MO and DC assessments compared with the INA (Tab. 3, Fig. 1). Waddell et al[12] reported that lumbar flexion was not reduced in patients with chronic LBP, but this finding contradicts the observations of Mayer et al[45] as well as our findings (Tab. 6; FTFD measure between acute and chronic groups at DC assessment). There seems to be general agreement, however, that a restriction of lumbar flexion is a clinically valid measure of acute LBP[12,46] and provides a test that is difficult to manipulate intentionally.[12] Similarly, the improvement in forward bending for patients who initially presented acute symptoms (Tab. 6) seems to occur in parallel with symptom resolution.

Previous reports[15,16] have also documented an increase in isometric lift capacity following rehabilitation for patients receiving workers' compensation. We found that isometric lift capacity increased over the course of rehabilitation (Figs. 1, 2). In addition, when the MIL values were compared descriptively with normative values for unimpaired workers[33] (Fig. 2), the mean MIL value was below average at the INA and above average at the 1MO and DC assessments for both gender groups. The magnitude of MIL seemed to be affected by compliance and chronicity. Patients with high compliance or acute symptoms showed larger increases in MIL between the INA and the DC assessment compared with patients with low compliance or chronic symptoms. The presence/absence of leg symptoms did not seem to alter the magnitude of improvement in MIL.

Limitations

This study was designed to describe the response to physical therapy provided to a wide variety of patients referred to an outpatient clinic for nonsurgical treatment of work related low back injury. The patients were not preselected and the treatment could not be rigorously standardized because of divergent patient characteristics. The efficacy of physical therapy, therefore, could not be attributed to a specific treatment protocol.[43] In addition, a control group of patients receiving workers' compensation with no treatment could not be justified for two reasons: (1) Patients referred to the clinic came seeking treatment, and (2) patients receiving workers' compensation are required by law to enter a treatment program or risk losing their benefits. The effects due to spontaneous recovery, therefore, could not be fully evaluated.

Therapists experienced in manual therapy provided the treatments described in this study. The results, therefore, might not be generalized to care provided by physical therapists who are not proficient in manual therapy techniques. The relatively large number of patients lost to follow-up might have altered the results in some unpredictable way. We have established, however, that those patients lost to follow-up did not have greater initial disability, more restriction in forward bending, or lower static lift capacity compared with those patients who were followed (Fig. 6).

Conclusions

A physical therapy program with multiple interventions was used to treat patients receiving workers' compensation without preselection. The Oswestry disability score, FTFD and MIL measures, and work status showed significant improvement at the 1MO and DC assessments as compared with the INA. Patients with a high level of compliance had lower disability scores at the 1MO and DC reassessments compared with the INA, whereas patients with low compliance showed minimal improvements in disability status. The return-to-work rates were higher for patients with acute symptoms than for patients with chronic symptoms. The efficacy of each cointervention was not evaluated. Issues concerning the scope of physical therapy that is needed to obtain optimal clinical results will require further study of patient classification systems and specific analysis of cointerventions.

Acknowledgment

We express our sincere appreciation to Kathy Anderson for her assistance with coordinating the implementation of this study. We also thank Diana Rose Di Fabio and Danielle Rae Di Fabio for their support in preparing this manuscript.

References

[1] Andersson GBJ. Epidemiologic aspects on low back pain in industry. Spine. 1981;6:53-60. [2] Spengler DM, Bigos SJ, Martin NA. Back injuries in industry: a retrospective study, I: overview and cost analysis. Spine. 1986;11:241-245. [3] Svensson H, Andersson GBJ. Low back pain in forty- to forty-seven-year-old men, 1: frequency of occurrence and impact on medical services. Scand J Rehabil Med. 1982;14:47-53. [4] Lancourt J, Kettelhut M. Predicting return to work for lower back pain patients receiving workers' compensation. Spine. 1992;17:629-640. [5] Greenough CG, Fraser RD. The effects of compensation on recovery from low-back injury. Spine. 1989;14:947-955. [6] Bradish CF, Lloyd GJ, Aldam CF, et al. Do nonorganic signs help to predict the return to activity of patients with low back pain? Spine. 1988;13:557-560. [7] Frymoyer JW. Predicting disability from low back pain. Clin Orthop. 1992;279:101-109. [8] Lacroix MJ, Powell J, Lloyd GJ, et al. Low back pain factors of value in predicting outcome. Spine. 1990;15:495-499. [9] Leavitt F, Garron DC, McNeill TW, Whisler WW. Organic status, psychological disturbance, and pain report characteristics in low-back-pain patients on compensation. Spine. 1982;7:398-402. [10] Mendelson G. Compensation, pain complaints, and psychological disturbance. Pain. 1984;20:169-177. [11] Roland M, Morris R. A study of the natural history of back pain, part II: development of guidelines for trials of treatment in primary care. Spine. 1983;8:145-150. [12] Waddell G, Somerville D, Henderson I, Newton M. Objective clinical evaluation of physical impairment in chronic low back pain. Spine. 1992;17:617-628. [13] Waddell G, Main CJ. Assessment of severity in low-back disorders. Spine. 1984;9:204-208. [14] Saal JA, Saal JS. Nonoperative treatment of herniated lumbar intervertebral disc with radiculopathy: an outcome study. Spine. 1989; 14:431-437. [15] Sachs BL, David JF, Olimpio D, et al. Spinal rehabilitation by work tolerance based on objective physical capacity assessment of dysfunction: a prospective study with control subjects and a twelve-month review. Spine. 1990;15:1325-1332. [16] Mayer TG, Gatchel J, Kishino N, et al. Objective assessment of spine function following industrial injury. Spine. 1985;10:482-493. [17] Hazard RG, Fenwick JW, Kalisch SM, et al. Functional restoration with behavioral support: a one-year prospective study of patients with chronic low back pain. Spine. 1989;14:157-161. [18] Lindstrom I, Ohlund C, Eek C, Wallin L, et al. mobility, strength, and fitness after a graded activity program for patients with sub-acute low back pain: a randomized prospective clinical study with a behavioral therapy approach. Spine. 1992;17:641-652. [19] Oland G, Tveiten G. A trial of modern rehabilitation for chronic low-back pain and disability: vocation outcome and effect of pain modulation. Spine. 1991;16:457-459. [20] Fredrickson BE, Trief PM, VanBeveren P, et al. Rehabilitation of the patient with chronic back pain: a search for outcome predictors. Spine. 1988;13:351-353. [21] Wiesel SW, Feffer HL, Borenstein DG. Evaluation and outcome of low-back pain of unknown etiology. Spine. 1988;13:679-680. [22] Hurme M, Alaranta H. Factors predicting the result of surgery for lumbar intervertebral disc herniation. Spine. 1987;12:933-938. [23] Frymoyer J, Hanley E, Howe J, et al. Disc excision and spine fusion in the management of lumbar disc disease. Spine. 1978;3:1-6. [24] Stratford P, Solomon P, Binkley J, et al. Sensitivity of sickness impact profile items to measure change over time in a low-back pain patient group. Spine. 1993;18:1723-1727. [25] Bigos SJ, Spengler DM, Martin NA, et al. Back injuries in industry: a retrospective study, II: injury factors. Spine. 1986;11:246-251. [26] Fairbank JCT, Davies JB, Mbaot JC, O'Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy. 1980;66:271-273. [27] Maitland GD. Vertebral Manipulation. 5th ed. London, England: Butterworth & Co (Publishers) Ltd; 1986. [28] Greenman R. Principles of Manual Medicine. 1st ed. Baltimore, Md: Williams & Wilkins; 1989. [29] Ponte DJ, Jensen GH, Kent BE. A Preliminary report on the use of the McKenzie protocol versus the Williams protocol. J Orthop Sports Phys Ther. 1984;6:130-139. [30] Lidstrom A, Zachrisson M. Physical therapy on low back pain and sciatica: an attempt at evaluation. Scand J Rehabil Med. 1970;2:37-42. [31] Haldeman S, Gilles J, Haldeman J, et al. Low back pain: a study of 50 patients on a group exercise program. Physiotherapy Canada. 1975;27:71-77. [32] Gauvin MG, Riddle DL, Rothstein JM. Reliability of clinical measurements of forward bending using the modified fingertip-to-floor method. Phys Ther, 1990;70:443-447. [33] Chaffin D. Preemployment strength testing: updated position. J Occup Med. 1978;20:403-408. [34] Rose SJ, Shulman AD, Strube MJ. Functional assessment of patients with low back pain syndrome. Topics in Geriatric Rehabilitation. 1986;1:9-30. [35] Zeh J, Hansson T, Bigos S, et al. Isometric strength testing: recommendations based on statistical analysis of the procedure. Spine. 1986;11:43-46. [36] McKenzie RA. The Lumbar Spine: Mechanical Diagnosis and Therapy. Waikanae, New Zealand: Spinal Publications Ltd; 1981. [37] Nordin M, Frankel VH. Basic Biomechanics of the Musculoskeletal System. 2nd ed. London, England: Lea & Febiger; 1989:183-207. [38] BMDP Statistical Software Manual (Volume 1, Program 4F). Los Angeles, Calif: University of California Press; 1990. [39] BMDP Statistical Software Manual (Volume 1, Software Release). Los Angeles, Calif: University of California Press; 1990. [40] Delitto A, Cibulka MT, Erhard RE, et al. Evidence for use of an extension-mobilization category in acute low back syndrome: a prescriptive validation pilot study. Phys Ther. 1993;73:216-222. [41] Donelson R, Silva G, Murphy K. Centralization phenomenon: its usefulness in evaluating and treating referred pain. Spine. 1990;15:211-213. [42] Donelson R, Grant W, Kamps C, et al. Pain response to sagittal end-range spinal motion: a prospective, randomized, multicentered trial. Spine. 1991;16(suppl):S206-S212. [43] Di Fabio RP. Efficacy of manual therapy. Phys Ther. 1992;72:853-864. [44] Kohles S, Barnes D, Gatchel RJ, Mayer TG. improved physical performance outcomes after functional restoration treatment in patients with chronic low-back pain (early versus recent training results). Spine. 1990;15:1321-1324. [45] Mayer TG, Tencer AT, Kristoferson S, Mooney V. Use of noninvasive techniques for quantification of spinal range of motion in normal subjects and chronic low back dysfunction patients. Spine. 1984;9:588-595. [46] Rae P, Venner RM, Waddell G. A simple clinical technique of measuring lumbar flexion. J R Coll Surg Edinb. 1981;29:281-284.