Physician referral to physical therapy in a cohort of workers compensated for low back pain.

Workers with back injuries have made up a large proportion of the physical therapist's work load. In a cohort of 6,581 compensated workers in California, 1,257 workers (19.1%) had musculoskeletal injuries (including 767 back injuries), of which 761 workers (60.5%) were referred for physical therapy services.[1] Haig et al[2] addressed whether length of time off work was altered by consultation with a physician specialist in physical medicine. They found a reduction in absence time from work in the group that saw the specialist, although the study did not take into account physical therapy and the group sizes were small. Other studies have focused on physicians' knowledge of physical therapy and their attitude toward referral. Stanton et al[3] studied medical residents and found that they scored low in their knowledge of physical therapy assessment and treatment techniques. Nevertheless, 98% of the respondents did refer patients for physical therapy. Ritchey et al[4] claimed that physicians referred patients for physical therapy based on their knowledge of services that can be provided for their patients. These authors, however, believed that physical therapy services were underutilized by physicians. They theorized that this situation could be reversed if physicians were more sensitive to what physical therapists do and better informed about the benefits of physical therapy.

Two studies examined the referral of patients (in the general population, not exclusively workers) for physical therapy by general practitioners. In a study done in the Netherlands,[5] one third of the 6,397 new referrals (in the year of study) were for back problems (excluding the cervical spine and curvatures of the spine). In addition, that study did not demonstrate a difference between high-, medium-, and low-referring general practitioners with respect to diagnostic categories. Akpala et al[6] described the referral for physical therapy from one group practice to one physical therapy facility in Scotland. They found that 717 (22%) out of the 32,372 patients served by this particular group practice were sent to the physical therapy facility under study. This number, however, did not reflect the true referral rate of the entire population because other facilities were not taken into account. Other trends described were that patients in the 45-to 54-year-old age group represented the highest proportion of referrals, female patients tended to be referred more than male patients (58% versus 42%), physicians within the group practice varied in their referral patterns (range=8%-41%), and 39% of all referrals were for vertebral column syndromes. In a study of physical therapy facilities in the United States, 25% of the patients were treated for low back pain.[7] Delays in starting physical therapy were associated with increased duration and cost of treatment. In addition, workers' compensation cases cost 36% more than those charged to other insurance carriers.

To our knowledge, there have not been any studies that have compared profiles of subjects who were referred or not referred for physical therapy for back pain problems. Additionally, the impact of physical therapy (in general, as opposed to looking at specific techniques) on absence from work has not been described.

The objectives of our study were (1) to describe the variables associated with the referral of a cohort of workers for physical therapy services; (2) to characterize the physical therapy by duration of treatment and choice of therapeutic techniques; and (3) to compare workers who were referred for physical therapy and those workers who were not referred in terms of age, gender, occupational group, diagnosis, region of residence, cost of care, and absence from work.

Method

Study Population

In 1988, the Quebec Worker's Compensation Board (QWCB) compensated 54,401 workers for an absence from work due to lower back (lumbar spine) problems. We randomly selected a sample of 2,500 subjects. A subject was included in the study if he or she (1) had had no compensation for back injury in the 2 years (1986 and 1987) before the study, (2) was being compensated for at least one full day of absence from work at the time of the study, and (3) was between the ages of 15 and 65 years. Based on these criteria, 353 subjects were excluded, yielding a final sample of 2,147 compensated workers.

Data Collection

Each subject was followed for 2 years from the date of entry into the study. Data were obtained from the QWCB computerized files, which included variables such as gender, age, occupation, sector of work, region (place of residence of the worker), daily compensation rate, recurrence, absence from work, dates of commencement of and discharge from physical therapy, cost of physical therapy per day, and total cost of treatment over 2 years of follow-up. Validation of occupation and industrial sector, that is, confirming the records were correct, was done on a random sample of 50 files.

Medical files were reviewed, and information was abstracted from them regarding diagnosis and prescribed treatments. The diagnoses written by the treating physicians in their medical reports to the QWCB within 7 days following the first day of absence from work in 1988 were classified as specific or nonspecific.8 Nonspecific diagnoses included low back pain, low back strain, low back sprain, and lumbago. Specific diagnoses included sciatica and specific anatomical abnormalities or pathology such as disk disease, spondylolysis, and so on. In addition, those subjects who received physical therapy from a licensed physical therapist had an initial report completed by the therapist, which was submitted to the QWCB. Information in that report described the proposed treatment plan (which included up to seven different therapeutic interventions per subject). In Quebec (Canada), physical therapy is an insured service with the QWCB, and the entire cost of this service is defrayed by the QWCB. This is not the case in the general health care system, in which coverage is not extended to private physical therapy clinics.

Statistical Analysis

Variables formed from the information supplied by the QWCB files included recurrence (yes or no) and cumulative absence from work within the 2-year span of follow-up. Chronicity was defined as absence from work for a period of 6 months or longer following back injury.

Continuous variables were compared by using two-tailed t tests. Categorical data were assessed by the chi-square test. When comparisons were made between those subjects who had received physical therapy and those subjects who had not received physical therapy, we limited the comparisons to those subjects with only one episode of back pain. This was done so that absence was consecutive and directly related to that episode in contrast to subjects who had numerous episodes of back pain over the follow-up period and accumulated absence from work as a result of these multiple episodes.

To determine whether receiving physical therapy earlier had an effect on absence from work, we used a multivariate model. The multiple logistic regression model was selected, in which the dependent variable is dichotomized. We defined the dependent variable as follows: Absence from work for a period of less than 60 days represented early return to work, whereas absence from work of more than 60 days represented late return to work. The independent variable was dichotomized as to whether the individual received physical therapy early (within the first 30 days following the injury) or did not receive physical therapy early (ie, was never referred for physical therapy or was referred after the first 30 days following the injury). The covariates included in this analysis were selected on the basis of their statistical significance in the univariate analysis. These covariates were age (in 10-year groupings), gender (female versus male), and diagnosis (specific versus nonspecific). The cutoff of 60 days was chosen because we were attempting to find out whether receiving physical therapy early helped return people to work earlier. We decided that if after approximately 1 month of physical therapy a subject returned to work, then the treatment would be considered beneficial.

To avoid biasing the results, subjects who were absent for less than 1 month were excluded from this analysis. Had they been included in the analysis, these subjects would not have had the opportunity to be referred to physical therapy because they represented cases that were extremely mild and resolved rather quickly on their own. In addition, we excluded 35 subjects who received surgical treatment following back injury because these subjects were absent from work for long periods of time related to their treatment. All assumptions for the multivariate model (multiple logistic regression) were verified and conformed to the model's specifications. Interactions were used at the outset, but all of these interactions were nonsignificant and also did not improve the fit of the model.

For all analyses, a probability value of less than .05 was considered to be statistically significant. Data were analyzed using the SAS for Personal Computers software package* for both univariate and multivariate models.[9]

Results

Of the 2,147 subjects in the study cohort, 389 subjects (18%) had received physical therapy after having been referred by their treating physician. For those subjects who had only one episode of back pain during the 2 years of follow-up, the average delay to begin physical therapy was 31.9 days and the average duration of the treatment period was 48.3 days, or 24.5 actual treatment days. The average duration of physical therapy did not differ with respect to gender or age group. The duration of physical therapy did differ, however, between those subjects who ended up being off work for more than 6 months (ie, subjects with chronic absence from work) and those who were absent from their jobs for less than 6 months (P<.0001). The frequency of the different therapeutic techniques is shown in the Figure. The treatments most frequently chosen by the treating physical therapists were therapeutic exercise, heat, therapeutic ultrasound, back education, vertebral manipulation and massage, and transcutaneous electrical nerve stimulation (TENS), which together accounted for 87% of all treatment techniques selected.

The mean age of the subjects receiving physical therapy was 36.4 years compared with 34.1 years for those subjects who did not receive physical therapy. There were differences between these two subgroups with respect to cumulative absence from work, average daily compensation salary, and number of recurrent episodes over the 2 years of follow-up, with the physical therapy recipients having a greater cumulative absence from work, a higher mean daily compensation salary, and more recurrences over the 2-year period. Additionally, in the group that received physical therapy, there were a greater proportion of workers with specific diagnoses (P<.001) and a greater proportion of female workers (P <.006) (Tab. 1). With regard to chronicity, there was a difference between the two groups, with more subjects with chronic absence from work in the group that received physical therapy than in the group that did not receive physical therapy. The per capita physical therapy cost was $619.30 (Canadian), or approximately $523 (US), for those subjects who received physical therapy.

Table 1.
Physical Therapy Utilization by Age, Gender, Occupation, Region, and
Diagnosis

                                            Percentage of Subjects
                            Total Number    Who Received Physical
                            of Subjects     Therapy(a)

                            2,147           18(389)
Age group (y)
   15-24                      425           12(52)
   25-34                      679           18(119)
   35-44                      377           22(98)
   45-54                      272           1952)
   55-64                      119           25(30)
Gender
   Male                     1,648           17(278)
   Female                     499           22(111)
Occupational group
   Blue collar              1,030           16(166)
   Drivers                    139           21(32)
   Nurses                     169           21(36)
   White collar               535           17(92)
Region
   Montreal                 1,049           20(209)
   Quebec City                192           12(23)
   Other urban centers        214           10(22)
   Rural areas                692           20(135)
Diagnosis
   Specific                   165           36(59)
   Nonspecific              1,683           7(279)

a Number of subjects in parentheses.

Receiving physical therapy was associated with increasing absence from work; only 2% of the entire cohort who were absent for 1 month or less received physical therapy, whereas between 61% and 78% of those absent more than 1 month were referred for physical therapy. Delays in commencing physical therapy tended to increase among those subjects with longer absences from work (Tab. 2). Those subjects with chronic absence from work (ie, were absent from work for 6 months more) had a greater delay in commencing physical therapy compared with subjects without chronic absence from work (P <.O1).

[TABULAR DATA OMITTED]

Multivariate Analysis

We performed a multivariate logistic regression to determine whether absence from work for greater than 60 days was associated with age, diagnosis, gender, and utilization of physical therapy services within the first month after back injury. The results indicated that, adjusted for age, receipt of physical therapy services within 30 days following back injury had a strong protective effect on absence from work (ie, it predicted return to work within 60 days following injury [odds ratio=0.13, 95% confidence interval=0.06,0.3]). In addition, female gender (odds ratio=1.8, 95% confidence interval=1.0,3.1) and specific diagnosis (odds ratio=2.2, 95% confidence interval=1.1,4.5) were predictors for absence from work of greater than 60 days' duration (Tab. 3).

Table 3.
Probability of Absence From Work Greater Than 60 Days in Relation to
Receiving Physical Therapy Within 30 Days of Injury(a)

                              Adjusted   95% Confidence
                              Odds       Interval for
Independent Variable          Ratio      Odds Ratio

Age (10-y intervals)          1.1        (0.88,1.38)
Diagnosis (specific versus    2.2         (1.1,4.5)
  nonspecific)
Gender  female versus male)   1.8         (1.0,3.1)
Received physical therapy     0.13       (0.06,0.3)
  within 30 days postinjury

a From logistic regression model, including all variables in the table.
Restricted to those subjects with only one episode of back pain who were
absent from work for at least 30 days and to those who did not receive
surgical treatment Within the 2-year follow-tip of the study. Odds ratio less
than, 1 is indicative of a protective effect; nonutilization of physical
therapy within 1 month postinjury odds ratio is 7.7,

Discussion

Physical therapy services were utilized by 18.1% of all workers compensated for back injuries in this representative cohort of compensated workers in Quebec. Swedlow et al[1] reported that 11.6% of all compensated workers in their study of the California workers' compensation system received physical therapy. They also reported that 60.5% of workers with musculoskeletal injuries were referred for physical therapy, although the proportion of those injuries that were back problems is unknown. Data from the Dutch[5] and Scottish[6] studies illustrate that 33% and 39%, respectively, of all referrals by general practitioners were for back ailments. Jette et al[7] reported that 25% of discharges from physical therapy outpatient departments were for low back pain. These studies reflect the general population and not only compensated workers. Additionally, the denominators differ considerably because in our study we investigated only individuals referred for back injuries, whereas in the aforementioned studies these figures represent a proportion of all referrals for all types of conditions.

The patients who were referred for physical therapy by their treating physician tended to be slightly older, to have a specific diagnosis, to be female, to be compensated at a slightly higher salary rate, and to be absent from work for more than 2 months. These results agree in part with those of Akpala and colleagues in Scotland,[6] who described that older age groups were more often referred for physical therapy , as were female patients.

Within the group that received physical therapy, the duration of physical therapy was higher than in the study by Jette et al[7] (48 days versus 36 days). This finding may be a reflection of the composition of the two cohorts. In our study, only compensated subjects were included (all costs were covered by the QWCB), whereas the sample in the study of Jette et al comprised subjects who paid for the services themselves in addition to those who were insured privately or had received workers' compensation. Jette et al reported that in their study, episodes of care charged to workers' compensation cost 36% more than those charged to other insurance carriers, which may occur as a result of longer duration of physical therapy. Another plausible explanation is that the subjects in our study, were more functionally disabled and required a longer duration of physical therapy to achieve their goals.

The mean age of the subjects in the study by Jette et al7 was slightly older (43.3 years) than in our study (36.4 years), again indicative of the differences between the two samples. Because our cohort was exclusively limited to a working population, the subjects were all under the age of 65 years and this may account for the younger age of our sample. Both studies support the fact that increasing delays in starting physical therapy are associated with longer episodes of physical therapy care. Within the group that received physical therapy, the duration or length of time spent in physical therapy was related to being a female and to being absent from work for a longer time.

The question that remains to be resolved is whether physical therapy occurs because the worker has been absent from the job for increasing periods of time or whether receiving physical therapy increases the absence time for the worker. In order to address this question, we compared absence from work for those subjects who received physical therapy (ie, absence from work until the time they began physical therapy) and absence from work for those subjects who were not referred for physical therapy. To avoid biases from multiple episodes of back pain and intermittent absences, we restricted this analysis to those subjects with only one episode of back pain (258 subjects in the group that received physical therapy, and 1,594 subjects in the group that did not receive physical therapy).

The data indicated that the average time off work until beginning physical therapy for the group that received physical therapy was 33.8 days as opposed to 14 days' absence from work for the group that did not receive physical therapy (Satterthwaite's approximation for unequal variances, t= - 5.59, df= 287.3, P<.0001). This finding implies that the subjects who received physical therapy were referred for physical therapy after lengthy absences from work, which may reflect the fact that initially physicians do not refer many of their patients for physical therapy. If the condition, however, does not improve and the patient continues to have pain or remains absent from work, the physician may be more inclined to refer the patient for physical therapy. Within the group that received physical therapy, the time from injury to start of physical therapy was higher in the subjects with chronic absence from work than in the subjects without chronic absence from work. This finding indicates that those subjects who were off work for greater than 6 months received physical therapy much later than those subjects who were off work for less than 6 months.

The results of the multivariate logistic analysis support the benefit of early physical therapy intervention on decreasing work absence. Receiving physical therapy within I month of the work injury was a strong predictor of return to work within 2 months of back injury. The hypothesis that early intervention by a physical therapist has a beneficial effect on return to work is not proven by the data; however, its plausibility remains. The challenge is to identify those patients who could benefit from early physical therapy intervention. Identifying these patients is especially important in view of the fact that within the group of subjects with chronic absence from work, longer delays before receiving physical therapy did not seem to affect absence from work, implying that the benefits of physical therapy among patients with chronic absence from work is questionable.

With respect to actual treatment modalities chosen by the treating physical therapists, supervised exercise, heat, and ultrasound represent 58% of all the treatments selected. Battie et all[10] found that physical therapists expressed preferences for education on proper body mechanics followed by exercises, ultrasound, heat, and ice in a hypothetical situation for treatment of patients with low back pain. The efficacy of these various modalities has been studied, and there has been some evidence supporting the use of exercise, ultrasound, cold, manipulation, and laser.[11]

The per-capita cost of physical therapy was 619.30 (Canadian), or approximately $523 (US), for the group that received physical therapy, which represented one third of the total medical per-capita cost for these subjects. This cost compares well with the costs described by Swedlow et all in workers compensated for back pain in California: $406 to $448 (US), depending on the type of physical therapy facility. Jette et al7 found that the mean cost of physical therapy per episode of back pain was $766.70 (US), although their sample was not selected exclusively from a working population. Mitchell et al'2 reported a yearly per-capita cost for physical therapy services in Florida varying from $600 to $791 (US). This latter study referred to the general population and not exclusively to that of workers as well as including physical therapy for all types of conditions, not only back problems.

Limitations of the Study

The main limitations of this study lie in the fact that virtually all of the measures were taken from the QWCB database or from the subjects' medical files. Quality control on such measures as diagnosis could not be assessed. In addition, severity of the back pain could not be addressed except that, in general, those subjects with specific diagnoses had more severe back pain than subjects with nonspecific diagnoses, as evidenced by the longer absence from work for those subjects with specific diagnoses. For the multivariate analysis, we did adjust for diagnosis, using it as a proxy measure for severity. Data regarding physical therapy plans were taken from the forms submitted by the treating physical therapists to the QWCB. These were the initial forms, which included the proposed treatment plan only. Thus, it is not confirmed that all the treatment modalities proposed were actually used.

Conclusion

With escalating health care costs, particularly those related to workers' compensation, it is essential that management of patients be more efficient yet remain effective. Physical therapy is an important component in the treatment of back pain syndromes. More judicious use of physical therapy, both in terms of patient profile and in terms of timing of referral, is the goal in order to achieve the most favorable results and reduce morbidity. This study clearly indicates that physicians request physical therapy services based on certain patient characteristics. Whether the patients with those characteristics are those who benefit the most from physical therapy remains to be proven. Timing of referral for physical therapy is another factor to consider. The results show that those subjects who were referred within the first month following injury tended to return to work within a relatively short period of time. If an optimal time for physical therapy intervention could be determined, both costs and morbidity may be decreased.

* SAS Institute Inc, PO Box 8000, Cary, NC 27511.

D Ehrmann-Feldman, MSc, PT, is Reseacher, Centre for Clinical Epidemiology and Community Studies. Sir Mortimer B Davis-Jewish General Hospital, 3755 Cote Ste Catherine Rd, Montreal, Quebec, Canada H3T 1E2 ([e-mail]). She is also a doctoral student, Department of Epidemiology and Biostatistics, McGill University Montreal, Quebec, Canada H3A 1A2. Address all correspondence to Ms Ehrmann-Feldman.

M Rossignol, MD, MSc, is Senior Researcher, Centre for Clinical Epidemiology and Community Studies, Sir Mortimer B Davis-Jewish General Hospital, and Assistant Professor, Department of Epidemiology and Biostatistics and Department of Occupational to Ms McGill University.

L Abenhaim, MD, ScD, is Director, Centre for Clinical Epidemiology and Community Studies, Sir Mortimer B Davis-Jewish General Hospital, and Associate Professor, Department of Epidemiology and Biostatistics, McGill University.

D Gobeille, MSc, was Research Assistant, Centre for Clinical Epidemiology, Sir Mortimer B Davis-Jewish General Hospital, at the time this study was conducted and is currently employed in private industry.

This investigation was supported by grants from the Institut de Recherche en Sante et Securite du Travail du Quebec. Ms Ehrmann-Feldman was supported by a National Health Research and Development Program fellowship from Health Canada.

An abstract of this study was presented at the 12th International Congress of the World Confederation for Physical Therapy; June 25-30, 1995; Washington, DC.

References

[1] Swedlow A, Johnson G, Smithline N, Milstein A. Increased costs and rates of use in the California workers' compensation system as a result of self-referral by physicians. N Engl J Med. 1992;327:1502-1506. [2] Haig AJ, Linton P, McIntosh M, et al. Aggressive early medical management by a specialist in physical medicine and rehabilitation: effect on lost time due to injuries in hospital employees. J Occup Med. 1990;32:241-244. [3] Stanton PE, Fox FK, Frangos KM, et al. Assessment of resident physicians' knowledge of physical therapy. Phys Ther. 1985;65:27-30. [4] Ritchey FJ, Pinkston D, Goldbaum, Heerten ME. Perceptual correlates of physician referral to physical therapists: implications for role expansion. Soc Sci Med. 1989;28:69-80. [5] Kerssens JJ, Groenewegen PP. Referrals to physiotherapy: the relation between the number of referrals, the indication for referral and the inclination to refer. Soc Sci Med. 1990;30:797-804. [6] Akpala GO, Curran AP, Simpson J. Physiotherapy in general practice: patterns of utilization. Public Health. 1988; 102:263-268. [7] Jette AM, Smith K, Haley SM, Davis KD. Physical therapy episodes of care for patients with low back pain. Phys Ther. 1994;74:101-115. [8] Deyo RA. Early diagnostic evaluation of low back pain. J Gen Intern Med. 1986;1:328-338. [9] SAS/STAT User's Guide, Release 603 Edition. Cary, NC: SAS Institute Inc; 1988. [10] Battie MC, Cherkin DC, Dunn R, et al. Managing low back pain: attitudes and treatment preferences of physical therapists. Phys Ther. 1994;74:219-226. [11] Tan JC, Roux EB, Dunand J, Vischer TL. Role of physical therapy in the management of common low back pain. Ballieres Clin Rheumatol. 1992;6: 629-655. [12] Mitchell JM, Scott E. Physician ownership of physical therapy services: effects on charges, utilization, profits, and service characteristics. JAMA. 1992;268:2055-2059.