Physical therapy utilization by patients with acute low back pain.

Key Words: Back pain, Health care services, Outcomes, Physical therapy, Utilization.

Low back pain (LBP)is one of the most common problems confronting outpatient physical therapists. More than one quarter of outpatient physical therapy referrals and almost one half of outpatient physical therapy visits are for treatment of LBP of varying acuity.[1-5] Most physicians agree that almost all patients with LBP should receive a trial of conservative therapy, such as physical therapy before invasive interventions are tried.[6] This study examined the utilization of physical therapy services by a group of patients with acute LBP in North Carolina.

The rising costs of health care in the United States have led public and private payers to examine the utilization patterns and outcomes of health care services.[1,2] Health care services research is needed in physical therapy to help providers of services manage costs and satisfy patients as well as to formulate effective clinical management strategies.[1]

Several researchers have reported on the utilization of physical therapy services for LBP. Ehrmann-Feldman et al described physical therapy for workers compensated for LBP in Canada, and they compared the characteristics of patients who were treated by physical therapists and the characteristics of patients who were not treated by physical therapists. Physical therapy recipients were older than patients who did not receive physical therapy (mean age of 36.4 years versus 34.1 years), and women were more likely than men to be treated by physical therapists. Ehrmann-Feldman et al[7] also reported that 36% of patients with a specific diagnosis were treated by physical therapists, whereas 17% of patients with a nonspecific diagnosis were treated by physical therapists.

Jette et al[2] used secondary analysis from the American Physical Therapy Association's Survey of Physical Therapy Outpatient Practice to examine a sample of outpatients with LBP who were treated by physical therapists.[2] The typical physical therapy recipient was a 45-year-old Caucasian Patients with Workers Compensation insurance had a 36% increase in the cost of their episodes of LBP, as compared with patients with other payment sources. The care provided to Caucasian patients for one episode of LBP cost 21% less than the care provided to non-Caucasian patients. Regional differences were found in the utilization and costs of physical therapy services, with the Northeast showing a 13% lower charge per visit but a 42% longer duration of treatment than other regions. The Midwest had a lower charge per visit but a higher frequency of visits. Fifty percent of the sample had symptoms of LBP for 4 weeks or less.[2]

Other research[1,8-11] has shown that physicians refer from 15% to 30% of patients with LBP of varying acuity to physical therapists The average number of visits per episode of LBP ranges from 6 to 25.[2,3,7,12] Private insurance is the most common source of payment, followed by Workers Compensation [2,13]

Our study, in our view, has important advantages over the studies by Jette et al[2] and Ehrmann-Feldman et al.[7] Jette et al[2] described only patients with LBP who were treated by physical therapists. They did not study patients who were not treated by physical therapists. The study by Ehrmann-Feldman et al[7] was a secondary data analysis confined to patients with LBP in Canada who received Workers Compensation. Our study is the first to examine the characteristics of patients who were treated by physical therapists within a cohort of patients with LBP who were seen by a variety of practitioners.

Our study analyzed data collected in a prospective observational cohort study of patients with acute LBP.[14] The purposes of our study were (1) to describe the demographic and clinical characteristics of patients with acute LBP, (2) to examine the demographic and clinical characteristics of those patients who were being treated by physical therapists, and (3) to analyze their service utilization. This information has important implications in the current health care market. To effectively allocate resources, it is important to know where people with LBP receive treatment and who provides care for what types of patients.

The following hypotheses were formulated from the previous research[14] on this data set, a review of the literature, and clinical experience:

1. Higher baseline Roland-Morris Questionnaire[15] scores, indicating greater disability, will increase the likelihood of being treated by physical therapists.

2. The type of initial provider will influence whether patients will be treated by physical therapists. Patients who are seen initially by chiropractors will be least likely to be treated by physical therapists, followed by patients who first see primary care providers. Patients who are seen initially by orthopedic surgeons will be most likely to be treated by physical therapists.

3. The location of pain will influence referral for physical therapy, and patients with LBP but without pain below the knee in one or both legs will be referred less often than will patients with LBP and with pain below the knee in one or both legs.

4. The location of pain will influence utilization of physical therapy services; patients with LBP but without pain below the knee in one or both legs will have fewer physical therapy visits than will patients with LBP and with pain below the knee in one or both legs.

Method

Our study examined the utilization of physical therapy services by a group of patients with acute LBP in North Carolina Sample selection for this study has been described previously, by Carey et al[14] and is summarized in the following text. The enrollment period for new patients was from June 1992 to March 1993. A total of 1,633 patients were enrolled Data from 1,580 patients who completed all the interviews are included in this analysis. Subjects were recruited from 208 health care providers who were randomly selected, using medical and chiropractic licensure files, to represent six strata. The 208 providers were 39 urban primary care physicians; 48 rural primary care physicians; 32 urban chiropractors; 32 rural chiropractors; 29 orthopedic surgeons; and 28 physicians, nurse practitioners, or physician's assistants in a group-model health maintenance organization (HMO). These providers are referred to as initial providers because the patients first sought care from them.

There were two inclusion criteria for the providers: (1) The, had to see ambulatory patients 50% of the time, and (2) they had to be the first medical entry point for patients with acute LBP. Consecutive patients with acute LBP who were seen by the providers were asked to participate in the study. The inclusion criteria for the subjects were (1) LBP for less than 10 weeks duration, (2) no previous care received for the current episode of LBP, (3) no history of back surgery, (4) no history of cancer, (5) no pregnancy at the time of the initial visit, (6) accessible via telephone, and (7) fluent in English. Fifty percent of the patients with back pain who were seen by the providers were eligible for the study. Only 8% of the patients who were eligible for the study declined to participate.

Procedure

Methods for this study have been described previously by Carey et al[14] and are briefly summarized in the following paragraphs. The subjects' consent for participation was obtained by the providers at the time of the initial office visit, and the rights of the subjects were protected. Subjects were paid $20 for participating and were told that the study's purpose was to determine the duration of LBP and the types of treatments used. Each subject's history was recorded and a physical examination was conducted during the initial visit. The providers were surveyed after the data were collected, and no attempt was made to influence the providers' clinical decisions. All follow-up survey instruments were collected by the staff members of the University of North Carolina Survey Research Unit (UNCSRU).

The subjects were contacted by telephone after the initial office visit for a telephone interview by a member of the UNCSRU. The subjects were asked for information on demographics, medical history, health care services utilization, functional status, employment, and satisfaction with treatment for back pain. The subjects were contacted again by telephone by a member of the UNCSRU 2, 4, 8, 12, and 24 weeks later, or until the subjects reported that they had completely recovered from their LBP if before 12 weeks. All subjects were interviewed at 24 weeks even if they had recovered earlier.

The follow-up interviews focused on health care services utilization, functional status, and satisfaction. The Roland-Morris adaptation of the Sickness Impact Profile (a validated 23-item scale that assesses loss of function due to back problems) was used to measure the subjects' functional status at each interview.[15] Stratford et al[16] compared several different disability questionnaires and found the Roland-Morris Questionnaire to be the best instrument for detecting change over time in patients with LBP. The initial provider's care was further evaluated in chart abstractions obtained by the study personnel, and these evaluations correlated well with the information reported by the patients.

Data Analysis

The number of subjects who were being treated by physical therapists and the number of subjects who were not being treated by physical therapists were computed for one episode of LBP. The demographic and clinical characteristics of subjects who were being treated by physical therapists were compared with those of subjects who were not being treated by physical therapists. The subjects' demographic characteristics were age, gender, race, marital status, education, payment source, receipt of Workers' Compensation, and income Prior episodes of LBP, prior treatment for LBP by physical therapists, duration of LBP prior to the initial visit, self-reports of LBP and of pain below the knee in one or both legs, neurological abnormalities, pain scores, and Roland-Morris Questionnaire scores were the reported clinical characteristics. Subjects were asked whether they had experienced LBP within the past 24 hours. If they answered "yes," then they were asked whether they currently had LBP and pain below the knee in one or both legs. If they responded "no," then they were asked whether, when they were experiencing LBP, the pain traveled below the knee. Neurological abnormalities were defined as asymmetrical ankle reflex and muscle weakness in ankle dorsiflexion. A higher Roland-Morris Questionnaire score represents greater self-rated disability secondary to LBP.[15] Patterns of referral for physical therapy were analyzed according to the type of provider the subjects saw on their initial provider visit. Service utilization for the subjects who were being treated by physical therapists was defined by number of physical therapy visits and by the procedures and modalities used.

Standard statistical packages (SAS(*) and SPSS([dagger]) were used. A two-sample t test and an analysis of variance were used to analyze the data obtained for the continuous variables. Pearson's chi-square statistic was used to analyze the categorical data. The level of statistical significance was set at P[is less than or equal to].05 for all tests.

Multiple logistic regression was used to establish the probability of receiving treatment by a physical therapist, using covariates found to be significant in the univariate analysis and other relevant covariates.[17] The bivariate dependent variable was whether physical therapy was received by the subject. The independent variables were the clinical and demographic characteristics of the subjects. The variables selected were age, gender, education, receipt of Workers' Compensation, income, prior physical therapy for LBP, duration of LBP prior to index visit, mean baseline Roland-Morris Questionnaire score, LBP and pain below the knee in one or both legs, and provider type.

Results

Hypotheses

The first three hypotheses were supported by the regression findings. A higher baseline Roland-Morris Questionnaire score, indicating greater disability, increased the likelihood of receiving treatment by physical therapists. The subjects who were being treated by physical therapists had a baseline Roland-Morris Questionnaire score that was five points higher than that of subjects who were not being treated by physical therapists. Subjects who saw orthopedic surgeons initially were most likely to be treated by physical therapists. Subjects who saw chiropractors initially were least likely to be treated by physical therapists. Location of pain influenced referral to physical therapists; subjects with self-reported LBP and pain below the knee in one or both legs were more likely to be treated by physical therapists. The location of pain did not influence the number of visits to a physical therapist; therefore, the fourth hypothesis was rejected.

Demographic and Clinical Characteristics

Baseline demographic characteristics of the subjects with acute LBP are reported in Table 1. One hundred ninety-nine (12.6%) of the subjects reported that they were being treated by physical therapists. Overall, the subjects had a mean age of 41 years and were Caucasian, married, and insured. Thirty-one percent of the subjects received Workers' Compensation, and 31% had a family income less than or equal to $20,000. The subjects were divided equally with regard to gender and education beyond high school.

Table 1. Baseline Demographic Characteristics of Cohort of Subjects With Acute Low Back Pain (N=1,580)

                                     No. of            Percentage
Characteristic                       Subject           of Cohort

Age (y)
  [bar]X                                41
  SD                                    13
  Range                                 20-75

Male                                   748                  47

Caucasian                            1,314                  83

Married                              1,110                  70

Less than high-school education        783                  50

Fully insured                        1,433                  91

Workers' Compensation                  488                  31

Family income of
[is less than or equal to]
$20,000                                488                   31

Table 2 presents the subjects' baseline clinical characteristics. Eighty-two percent of the subjects had had prior episodes of LBP, and 24% of the subjects had LBP and pain below the knee in one or both legs. Sixty-seven percent of the subjects reported that they had LBP for 2 weeks or less prior to the initial visit. The mean baseline Roland-Morris Questionnaire score was 11.1 on a scale of 0 to 23, with a higher score representing more disability. The mean age of subjects who were being treated by physical therapists was 39 years, compared with 42 years for subjects who were not being treated by physical therapists (Tab. 3).

Table 2. Baseline Clinical Characteristics of Cohort of Subjects With Acute Low Back Pain (LBP) (N= 1,580)

                                     No. of     Percentage
Characteristic                      Subjects    of Cohort

Prior episodes of LBP
  0 before now                        286       18
  1-5 episodes                        801       51
  >5 episodes                         491       31

Prior physical therapy For LBP        218       14

Duration  of LBP of                 1,044       67
    [is less than or equal to]2
    weeks prior to initial visit

Subjects with LBP and with pain       371       24
    below the knee in one or
    both legs

Neurological abnormalities            293       19

Self-assessed pain score at
    baseline(a)

  [bar]X                                5.4
   SD                                   2.4

Roland-Morris Questionnaire[15]
score(b) at baseline

  [bar]X                               11.1
  SD                                    7.0

(a) Scale of 0 to 10, where 0=no pain and 10=pain as bad as it could be. (b) Scale of 0 to 23, where 0=no disability and 23=severe disability.

Table 3. Age of Cohort of Subjects (N= 1,580) Associated With Utilization of Physical Therapy for Acute Low Back Pain

                              Not
                Treated       Treated
                by a          by a
Demographic     Physical      Physical
Variable        Therapist     Therapist     p(a)

Age (y)
  [bar]X        39             42           .0056
  SD            11             13
  Range         39-75          42-75

(a) Significance level set at P[is less than or equal to].05

The subjects' demographic characteristics associated with utilization of physical therapy for acute LBP are summarized in Table 4. Eleven percent of the subjects with a high-school education or less were being treated by physical therapists, whereas 15% of the subjects who were being treated by physical therapists were educated beyond high school. Seventeen percent of the subjects with Workers' Compensation were being treated by physical therapists, whereas only 11% of subjects without Workers' Compensation were being treated by physical therapists. Workers' Compensation and education beyond high school were associated with receiving treatment by physical therapists. Gender, race, marital status, payment source, and income were not associated with receiving treatment by physical therapists.

Table 4. Demographic Characteristics of Cohort of Subjects (N=1,580) Associated With Utilization of Physical Therapy for Acute Low Back Pain

                              No. of
                              Subjects
                              Treated
                              by a        Percentage
                              Physical    of
Variable                      Therapist   Cohort

Gender
 Male (n=748)                  83         11
 Female (n=832)               116         14

Race (n=1,579)
 Caucasian (n=1,314)          163         12
 Non-Caucasian (n=265)         36         14

Marital status
 Married (n=1,110)            142         13
 Not married (n=470)           57         12

Education
 [is less than or equal to]
 High school (n=797)
 >High school (n=783)          84         11
                              115         15
Payment source (n=1,575)
 None (or Medicaid/
  Medicare without
  coinsurance) (n=142)         16         11
 Full insurance (n=1,433)     183         13

Workers' Compensation
 Yes (n=488)                   83         17
 No (n=1,092)                 116         11

Income
 [is less than or equal to]
 $20,000 (n=488)               71         15
 >$20,000 (n=1,092)           128         12

Variable                      P(a)

Gender
 Male (n=748)                 .089
 Female (n=832)

Race (n=1,579)
 Caucasian (n=1,314)          .60
 Non-Caucasian (n=265)

Marital status
 Married (n=1,110)            .72
 Not married (n=470)

Education
 [is less than or equal to]
 High school (n=797)
 >High school (n=783)         .013

Payment source (n=1,575)
 None (or Medicaid/
  Medicare without
  coinsurance) (n=142)        .61
 Full insurance (n=1,433)

Workers' Compensation
 Yes (n=488)                  [is less than
 No (n=1,092)                 or equal to] .001

Income
 [is less than or equal to]
 $20,000 (n=488)              .12
 >$20,000 (n=1,092)

(a) Significance level set at P [is less than or equal to] .05.

Table 5 summarizes the subjects' clinical characteristics associated with utilization of physical therapy for acute LBP. Twenty-three percent of the subjects who had prior physical therapy for LBP were currently being treated by physical therapists, compared with 11% of the subjects who had not had any prior treatment. Seventeen percent of the subjects who had LBP for more than 2 weeks prior to the initial provider visit were being treated by physical therapists, compared with 11% of the subjects who had LBP for 2 weeks or less. Twenty-one percent of the subjects with self-reported LBP and pain below the knee in one or both legs were being treated by physical therapists, compared with 10% of the subjects who did not have LBP and pain below the knee in one or both legs. Sixteen percent of the subjects with neurological abnormalities were being treated by physical therapists. The mean baseline Roland-Morris Questionnaire score for the subjects who were receiving physical therapy was 15.6 (SD=5.7), 5 points higher than that of the subjects who were not receiving physical therapy ([bar]X= 10.5, SD=6.9). Factors associated with being treated by physical therapists were prior physical therapy for LBP, a duration of LBP of greater than 2 weeks prior to the initial provider visit, self-reported LBP and pain below the knee in one or both legs, neurological abnormalities, and a higher mean baseline Roland-Morris Questionnaire score. Prior episodes of LBP and the mean pun score at baseline showed no difference.

Table 5. Clinical Characteristics of Cohort of Subjects (N=1,580) Associated With Utilization of Physical Therapy for Acute Low Back Pain (LBP)

                              No. of
                              Subjects
                              Treated
                              by a        Percentage
                              Physical    of
Characteristic                Therapist   Cohort

Prior episodes of LBP
  (n=1,578)
 0 before now (n=286)          45         16
 1-5 (n=801)                   97         12
 >5 (n=491)                    56         11

Prior physical therapy for
  LBP (n=1,576)
 Yes (n=218)                   50         23
 No (n=1,385)                 149         11

Duration of LBP prior to
  initial visit
  (n=1,563)
 >2 wk (n=519)                 86         17
 [is less than or equal to]   113         11
  2 wk (n=1,044)

LBP and pains below the
  knee in one or both
  legs (n=1,575)
 Yes (n=371)                   79         21
 No (n=1,204)                 120         10

Neurological
  abnormalities
  (n=1,530)
 Yes (n=293)                   46         16
 No (n=1,237)                 140         11

Self-assessed pain score(b)
  at baseline (treated
  by a physical
  therapist/not treated
  by a physical
  therapist)
[bar]X                         5.6/5.3
SD                             2.3/2.4
Range                         1-20/0-10

Roland-Morris
 Questionnaire(15) score(c)
  at baseline (treated
  by a physical
  therapist/not treated
  by a physical
  therapist)

[bar]X                         15.6/10.5
SD                              5.7/6.9
Range                         1-23/0-23

Characteristic                P(a)

Prior episodes of LBP
  (n=1,578)
 0 before now (n=286)         .19
 1-5 (n=801)
 >5 (n=491)

Prior physical therapy for
  LBP (n=1,576)
 Yes (n=218)
 No (n=1,385)                 [is less than or equal to] .001

Duration of LBP prior to
  initial visit
  (n=1,563)
 >2 wk (n=519)                [is less than or equal to] .001
 [is less than or equal to]
  2 wk (n=1,044)

LBP and pains below the
  knee in one or both
  legs (n=1,575)
 Yes (n=371)
 No (n=1,204)                 [is less than or equal to] .001

Neurological
  abnormalities
  (n=1,530)
 Yes (n=293)
 No (n=1,237)                 .039

Self-assessed pain score(b)
  at baseline (treated
  by a physical
  therapist/not treated
  by a physical               .10
  therapist)
[bar]X
SD
Range

Roland-Morris
 Questionnaire(15) score(c)
  at baseline (treated
  by a physical               [is less than or equal to] .001
  therapist/not treated
  by a physical
  therapist)

[bar]X
SD
Range

(a) Significance level set at P [is less than or equal to] .05.

(b) Scale of 0 to 10, where 0=no pain and 10=pain as bad as it could be.

(c) Scale of 0 to 23, where 0=no disability and 23=severe disability.

Utilization of Services

The subjects who were being treated by physical therapists had approximately twice as many visits to their initial provider compared with the subjects who were not receiving physical therapy. Subjects whose initial provider was a chiropractor had approximately five times the number of visits to the initial provider than did the subjects who first went to primary care physicians, orthopedic surgeons, and HMO primary care providers.

Table 6 shows that orthopedic surgeons, as initial providers, had the highest percentage of subjects who went on to receive treatment by physical therapists (21%), followed by HMO primary care providers (15%), primary care physicians (14%), and chiropractors (7%). No difference was found in the mean or median number of visits to physical therapists according to type of initial provider. The median number of visits to physical therapists varied from 5 for subjects who were first seen by HMO providers to 7 for subjects who were first seen by chiropractors.

Table 6.
Percentage of Cohort of Subjects (N=1,580) Treated by a Physical
Therapist by Initial Provider Type

                         No. of
                         Subjects
                         Treated
                         by a
                         Physical    Percentage
Initial Provider         Therapist   of Cohort    P(a)

Chiropractor (n=586)     43           7           [is less than or
Primary care physician                            equal to] .001
 (n=625))                90          14
Health maintenance
 organization (n=195)    30          15
Orthopedic surgeon
 (n=174)                 36          21

(a) Significance level set at P [is less than or equal to] .05.

The Figure illustrates the frequency of procedures and modalities that subjects reported physical therapists used during treatment. The frequencies of use of the procedures and modalities are reported in descending order: exercise instruction (83%), heat treatment (74%), ultrasound (65%), electrical stimulation (50%), massage (49%), cold pack (35%), spinal manipulation or adjustment (21%), traction (16%), and other (6%).

Multivariate Analysis

Table 7 shows the results of the multivariate logistic regression analysis to determine the association of multiple factors (independent variables) on receiving physical therapy (dependent variable). The logistic regression analysis confirmed all of the bivariate findings. Post--high-school education, receipt of Workers' Compensation, previous physical therapy for LBP, self-reported LBP and pain below the knee in one or both legs, and a higher Roland-Morris Questionnaire score were positively associated with receiving physical therapy.

Discussion

Hypotheses

The first hypothesis was confirmed; subjects with greater disability at baseline, as determined by the Roland-Morris Questionnaire, were more likely to be treated by physical therapists. The mean baseline Roland-Morris Questionnaire score for the subjects who were being treated by physical therapists was five points higher than that of the subjects who were not receiving physical therapy. A four- to five-point difference is considered to be sufficient to detect clinical change.[18] Jette and Jette[19] recently published an article using data from the Focus on Therapeutic Outcomes (FOTO) database that supports this finding. The FOTO database consists of information from patients who received physical therapy for lumbar or cervical pain. The purpose of their study was to analyze the effect of physical therapy on health outcomes for patients with spinal impairments. Jette and Jette reported a decrease in baseline health status, compared with the general population, of patients with spinal impairments during their initial visit to physical therapists. Specifically, they found that the patients with lumbar impairments who had been treated by physical therapists had a decrease in physical functioning.

In future studies, it will be important to control for baseline functional status when comparisons are made of outcomes by different providers. For example, if baseline characteristics are not compared and other providers see patients who are less disabled, then they may have more favorable results. Unequal baseline characteristics could lead clinicians and payers to draw inappropriate conclusions about the effectiveness of physical therapy, and physical therapy utilization could be affected.

The total number of provider visits is an outcome measurement with cost implications In our study, the total number of provider visits was higher for the subjects who were being treated by physical therapists than for the subjects who did not receive physical therapy. This finding is consistent with the finding that the subjects who were receiving physical therapy had greater baseline disability and may have required more treatments. These findings demonstrate the importance of controlling for baseline disability when comparing treatment outcomes. The number of provider visits and therefore treatment costs are influenced by baseline disability.

The second hypothesis, that the type of initial provider will influence whether patients will be treated by physical therapists, was accepted. Subjects who first went to a chiropractor were least likely to be treated by physical therapists, followed by subjects who first went to primary care providers. Subjects who went to orthopedic surgeons first were most likely to be treated by physical therapists. These findings are consistent with those in the current literature. Jette and Davist[13] examined referrals to hospital-based and private outpatient physical therapy practices and they found that orthopedists and primary care physicians were the top two types of referring providers. Professional convergence could explain why orthopedic surgeons often refer patients to physical therapists--both professions predominantly treat patients with musculoskeletal illness. Physical therapists also have mainly marketed their services to orthopedists and other subspecialists.[7]

A limitation of our study is that patients who received physical therapy may or may not have been referred by their initial provider. We examined what actually happened instead of examining provider intent. The questionnaire was designed to analyze utilization of services by the initial provider. The subjects could have referred themselves, could have been referred by their initial provider, or could have been referred by a provider other than the initial provider.

The third and fourth hypotheses concerned the effect of diagnosis or classification on the utilization of physical therapy services. Deyo[20] argued that there is no acceptable classification scheme for diagnosing patients with LBP and went on to assert that a reliable diagnostic or classification scheme would greatly influence the indication for therapy. We examined the presence of pain in the back and below the knee as a classification category and its effect on physical therapy utilization. A limitation of our study was that the presence of pain in the back and below the knee was the only diagnostic or classification scheme.

We found that the subjects with LBP and pain below the knee in one or both legs were more likely to be treated by physical therapists than were the subjects without pain below the knee. We therefore accepted the third hypothesis. The current literature supports this finding. Ehrmann-Feldman et al[7] associated the identification of a specific diagnosis with the utilization of physical therapy. They found that out of 165 subjects with a specific diagnosis, 36% were treated by physical therapists, whereas only 17% of 1,683 subjects with a nonspecific diagnosis were treated by physical therapists. In Ehrmann-Feldman and colleagues' study[7] and in our analysis, subjects with pain below the knee were twice as likely to receive treatment by a physical therapist. Overman et al[21] found that the presence of radicular symptoms was correlated positively with a physician's decision to refer patients for physical therapy.

The fourth hypothesis was that LBP with pain below the knee would influence utilization of physical therapy services. We rejected this hypothesis because subjects with LBP but without pain below the knee in one or both legs did not have fewer physical therapy visits than did subjects with LBP and pain below the knee in one or both legs. Several studies[7,22-25] have examined diagnostic categories to decide who would benefit most from physical therapy. Chavannes et al[24] studied 475 patients with acute LBP who were seen by their own general practitioner and found that 38% of the patients had acute LBP with radiation to one or both legs and that 62% of the patients had pain only in their back. Eleven percent of the patients were referred to a specialist or a physical therapist, and the patients without radiation had a greater decrease in pain than did patients with radiation at 4 weeks.

Referral rates and outcomes are affected by diagnostic or classification categories. In future studies, therefore, researchers should examine these categories when reporting outcomes. Often, studies describe characteristics of the episode and symptoms but not a clinical diagnosis for the patients' LBP. Because multiple structures and tissues are capable of mimicking each other and the physiology of healing will be different for these structures and tissues, improved diagnostic specificity will help control for differences when outcomes are reported. The development of a reliable diagnostic or classification scheme will be necessary to evaluate fully the effect of diagnosis on physical therapy utilization and outcome.

Demographic and Clinical Characteristics

One hundred ninety-nine subjects, or 12.6% of the cohort, reported that they were being treated by physical therapists, which is a lower percentage than was reported in other studies.[1,8-11] In the cohort that we studied, 67% of the subjects had back pain for 2 weeks or less. The most likely reason for the lower percentage of subjects being treated by physical therapists is that this cohort had greater acuity compared with subjects in other studies. Some subjects referred to physical therapists may have gotten better quickly and therefore did not go to physical therapists.

In Our study, a postsecondary level of education was associated with a greater likelihood of being treated by a physical therapist. Deyo and Tsui-Wu[8] reported that more educated subjects sought care from an orthopedic surgeon, whereas less educated subjects sought care from general practitioners.[8] We did not examine the relationship of education and choice of provider, but our study showed that the highest number of subjects who were treated by physical therapists had orthopedic surgeons as their initial provider.

Ninety-one percent of the entire cohort were fully insured. There was no difference in physical therapy utilization between the subjects who were fully insured and those who were not fully insured. Seventeen percent of the cohort with Workers' Compensation were being treated by physical therapists, and studies that have examined payer source for episodes of LBP have shown private insurance to be the most utilized, followed by Workers' Compensation.[3,14] The rapid expansion of managed care insurance plans, which often limit the number of physical therapy visits per time period or by diagnosis, may drastically affect the utilization of physical therapy services. The effects of managed care on physical therapy should be studied in the future.

Neither Jette et al[2] nor Ehrmann-Feldman et al[7] reported on prior treatment for LBP. Subjects in our study were twice as likely to be treated by physical therapists for their current episode of LBP if they had been treated by physical therapists in the past. A possible reason for this finding is that subjects who had received physical therapy previously may have been satisfied that it improved their functioning or decreased their symptoms. Alternatively, initial providers may have felt that physical therapy had been beneficial for particular patients and again made referrals.

Utilization of Services

The mean number of visits to physical therapists in our study was 8.5 visits. The number of visits to physical therapists per episode of LBP is varied in the literature.[2-4,12,21] Jette et al[2] reported an average of 11 visits to physical therapists, and Akpala et al[12] reported an average of 6 visits to physical therapists. Ehrmann-Feldman et al[7] reported an average of 25 visits to physical therapists per episode of LBP, and all subjects in their study were receiving Workers' Compensation. The number of visits to physical therapists can be correlated to costs to society. If the average charge per visit is $70.00 (estimated from insurance records in our study) using the overall average of 8 physical therapy visits, then the mean total physical therapy charge for one episode of LBP in our study would be $560.00. Differences found in health care services utilization and costs may be due to acuity of LBP, geographical differences, and variations in health care systems.[7,26]

Our study and the studies by Jette et al[2] and Ehrmann-Feldman et al[7] showed that therapeutic exercise was the procedure used mostly frequently by physical therapists in the treatment of patients with LBP. This finding is supported by the Agency for Health Care Policy and Research's clinical guidelines for the treatment of adults with acute LBP; exercise and spinal manipulation (under certain conditions) are the only treatments that are recommended.[27,28] Jette and Jette[19] also reported that exercise and spinal manipulation were predictors of better outcomes for patients with spinal impairments.

Evidence for the effectiveness of procedures and modalities used by physical therapists other than exercise and manipulation will be crucial as health care reform continues.[3] The analysis of procedures and modalities in our study was based on the subjects' reports of the treatments that they had received. The subjects selected procedures and modalities from a list that may not have accurately reflected the clinical treatment regimens actually used by the physical therapists. Sullivan et al[29] reported that physical therapists use several different treatment techniques during an episode of care.

Carey et al,[14] in the original study using this cohort, reported the outcomes and costs for acute LBP based on the subjects' initial provider. All instruments focused on care received from this initial provider. Physical therapy was not a provider stratum; therefore, satisfaction and functional outcomes could not be related directly to physical therapists. In a another study,[30] Carey et al examined care-seeking for acute severe LBP and found that less than 2% of the patients initially sought care from a physical therapist. Jette et al[2] and Battie et al[3] propose that future studies should describe patient satisfaction with and outcomes of physical therapy and how they vary in relation to utilization of physical therapy services.

Overman et al[21] examined physical therapists as initial providers for patients with LBP and compared them with internists. They found greater patient satisfaction and greater improvements in functional outcome among patients managed by physical therapists than among patients managed by internists. Jette and Jettel[19] recently associated treatment by a physical therapist with outcome. They reported improvement in health status in both the physical and emotional dimensions for patients who were treated by physical therapists. In the current managed care environment, physical therapy needs to be compared with other health care disciplines to evaluate the least expensive and most effective treatment for patients with acute and prolonged episodes of LBP.

Patients have direct access to physical therapy in most states; the characteristics of patients and physical therapists who utilize this service would be important to those hoping to change practitioner and patient behaviors with regard to the treatment of LBP. For example, recent research[21,31] has been directed toward the timeliness of care received by patients with LBP. Direct access to physical therapy eliminates delays due to a requirement of physician referral and may be associated with improved outcomes. This improved access to services that promote return to functional activities and prevention of LBP would reduce the costs of disability.[21]

Characteristics of physical therapists regarding education, training, years of experience, and practice styles should be included in further research. No information was provided in this our study on the characteristics of the physical therapists. Future studies also should describe the characteristics of physicians who refer patients to physical therapists and of physicians who do not refer patients to physical therapists. This information would be useful to physical therapists for targeting educational and marketing efforts.

Conclusions

This is the first population-based study in the United States that describes the subjects who actually received treatment by a physical therapist. This study, therefore, is more generalizable than previous studies. Post-high-school education and Workers' Compensation are demographic characteristics associated with being treated by physical therapists. Our study showed that the people who were treated by physical therapists were more functionally disabled, had prior episodes of LBP, had prior physical therapy for LBP, and were more likely to have pain below the knee. In this study, physical therapists were utilized in the treatment of patients with greater severity of LBP. The findings demonstrate the importance of controlling for baseline characteristics when comparing outcomes of patients with LBP when treated by different types of providers.

References

[1] Swanson G. Use of outcome reports: justifying the need for physical therapy services. Orthopaedic Physical Therapy Clinics of North America. 1995;4:253-268.

[2] 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.

[3] 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.

[4] 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.

[5] Dixon AS. Progress and problems in back research. Rheumatol Rehabil. 1973;12:165-175.

[6] Deyo RA. Conservative therapy for low back pain: distinguishing useful from useless therapy. JAMA. 1983;250:1057-1062.

[7] Ehrmann-Feldman D, Rossignol M, Abehaim L, Gobeille D. Physician referral to physical therapy in a cohort of workers compensated for low back pain Phys Ther. 1996;76:150-157.

[8] Deyo RA, Tsui-Wu YJ. Descriptive epidemiology of low-back pain and its related medical care in the United States. Spine. 1987;12:264-268.

[9] Faas A, Van Eijk J, Chavannes A, Gubbels J. A randomized trial of exercise therapy in patients with acute low back pain: efficacy on sickness absence Spine. 1995;20:941-947.

[10] Hart G, Deyo RN, Cherkin DC. Physician office visits for low back pain. Spine. 1995;20:11-19.

[11] Frymoyer Pope MH, Clements JH, et al. Risk factors in low-back pain. J Bone Joint Surg Am. 1983;65:213-218.

[12] Akpala CO, Curran AP, Simpson J. Physiotherapy in general practice: patterns of utilization. Public Health. 1988;102:263-268.

[13] Jette AM, Davis KD. A comparison of hospital-based and private outpatient physical therapy practices. Phys Ther. 1991;71:366-375.

[14] Carey T, Garrett J, Jackman A, et al. The outcomes and costs of care for acute low hack pain among patients seen by primary care providers, chiropractors, and orthopedic surgeons. N Engl J Med. 1995;333:913-917.

[15] Roland M, Morris R. A study of the natural history of back pain, part 1: developmental of a reliable and sensitive measure of disability in low-back pain. Spine. 1989;11:951-954.

[16] Stratford PW, Binkley JM, Solomon P, et al. Assessing change over time in patients with low back pain Phys Ther. 1994;74:528-533.

[17] SPSS for Windows: Base System User's Guide, Release 6.0. Chicago, Ill: SPSS Inc; 1993.

[18] Stratford PW, Binkley Solomon P, et al. Defining the minimum level of detectable change for the Roland-Morris Questionnaire. Phys Ther. 1996;76:359-368.

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

[20] Deyo RA. Invited commentary on "Physical Therapy Episodes of Care for Patients With Low Back Pain." Phys Ther. 1994;74:111-113.

[21] Overman SS, Larson JW, Dickstein DA, Rockey PH. Physical therapy care for low back pain: monitored program of first-contact non-physician care. Phys Ther. 1988;68:199-207.

[22] Di Fabio RP, Mercy G, Holte JB. Physical therapy outcomes fro patients receiving workers' compensation following treatment for herniated lumbar disc and mechanical low back pain syndrome. J Orthop Sports Phys Ther. 1996;23:180-187.

[23] 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.

[24] Chavannes A, Gubbels J, Post D, et al. Acute low back pain: patients' perceptions of pain four weeks after diagnosis and treatment in general practice. J R Coll Gen Pract. 1986;36:271-273.

[25] Delitto A, Erhard RE, Bowling RW. A treatment-based classification approach to low back syndrome: identifying and staging patients for conservative treatment. Phys Ther. 1995;75:470-485.

[26] Waddell G. Modern management of spinal disorders. J Manipulative Physio Ther. 1995;18:509-596.

[27] Bigos S, Bowyer O, Braen G, et al. Acute Low Back Problems in Adults. Rockville, Md: Agency for Health Care Policy and Research, Public Health Service, US Department of Health and Human Services; December 1994. Clinical Practice Guideline No. 14, AHCPR Publication No. 95-0642.

[28] Rothstein JM, Delitto A, Scalzitti D. Understanding AHCPR Clinical Practice Guideline No. 14: Acute Low Back Problems in Adults. Alexandria, Va: American Physical Therapy Association; 1995.

[29] Sullivan SM, Kues JM, Mayhew TP. Treatment categories for low back pain: a methodological approach. J Orthop Sports Phys Ther. 1996;24:355-364.

[30] Carey T, Evans A, Hadler N, et al. Acute severe low back pain: a population-based study of prevalence and care-seeking. Spine. 1996;21: 339-344.

[31] Linton S, Hellsing A, Andersson D. A controlled study of the effect of an early intervention of acute musculoskeletal pain problems. Pain. 1993;54:353-359.

(*) PC SAS, Version 6.04, SAS Institute Inc, PO Box 8000, Cary, NC 27511.

([dagger]) SPSS for Windows, Release 6.0, and SPSS for UNIX: Advanced Statistics, Release 5.0, SPSS Inc, 444 N Michigan Ave, Chicago, IL 60611.

TJ Mielenz, PT, is Clinical Assistant Professor, Division of Physical Therapy, Department of Medical Allied Health Professions, School of Medicine, The University of North Carolina at Chapel Hill, CB 7135, Medical School Wing E, Room 211, Chapel Hill, NC 27599-7135 (USA) ([e-mail]). Address all correspondence to Ms Mielenz.

TS Carey, MD, is Professor of Medicine, Department of Medicine, School of Medicine, The University of North Carolina at Chapel Hill.

DA Dyrek, PT, is assistant Professor, Graduate Programs in Physical Therapy,

MGH Institute of Health Professions.

JM Garret, PhD, is Research Associate Professor, Department of Medicine, School of Medicine, The University of North Carolina at Chapel Hill.

JD Darter is Application Anayst Programmer, Cecil B Sheps Center for Health Services Reserach, Chapel Hill, NC.

This study was approved by the Institutional Review Board Subcommittee on Human Studies of Massachusetts General Hospital.

This study was supported in part by a grant from the Agency for Health Care Policy and Research to Dr Carey (HS06663) at the Cecil B Sheps Center for Health Services Research.

This article was submitted December 5, 1995, and was accepted March 19, 1997.