• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of ptjournalLink to Publisher's site
Phys Ther. Sep 2008; 88(9): 989–1004.
PMCID: PMC2527215

Predictors of Physical Therapy Clinic Performance in the Treatment of Patients With Low Back Pain Syndromes

Abstract

Background and Purpose: Little is known about organizational and service delivery factors related to quality of care in physical therapy. This study sought to identify characteristics related to differences in practice outcomes and service utilization.

Subjects: The sample comprised 114 outpatient clinics and 1,058 therapists who treated 16,281 patients with low back pain syndromes during the period 2000–2001. Clinics participated with the Focus on Therapeutic Outcomes, Inc (FOTO) database.

Methods: Hierarchical linear models were used to risk adjust treatment outcomes and number of visits per treatment episode. Aggregated residual scores from these models were used to classify each clinic into 1 of 3 categories in each of 3 types of performance groups: (1) effectiveness, (2) utilization, and (3) overall performance (ie, composite measure of effectiveness and utilization). Relationships between clinic classification and the following independent variables were examined by multinomial logistic regression: years of therapist experience, number of physical therapists, ratio of physical therapists to physical therapist assistants, proportion of patients with low back pain syndromes, number of new patients per physical therapist per month, utilization of physical therapist assistants, and setting.

Results: Clinics that were lower utilizers of physical therapist assistants were 6.6 times more likely to be classified into the high effectiveness group compared with the low effectiveness group, 6.7 times more likely to be classified in the low utilization group compared with the high utilization group, and 12.4 times more likely to be classified in the best performance group compared with the worst performance group. Serving a higher proportion of patients with low back pain syndromes was associated with an increased likelihood of being classified in the lowest or middle group. Years of physical therapist experience was inversely associated with being classified in the middle utilization group compared with the highest utilization group.

Discussion and Conclusion: These findings suggest that, in the treatment of patients with low back pain syndromes, clinics that are low utilizers of physical therapist assistants are more likely to provide superior care (ie, better patient outcomes and lower service use).

Low back pain syndromes (LBPS) affect more than 65 million Americans, with cumulative costs making LBPS the sixth most expensive condition treated in the United States. Direct and indirect costs associated with LBPS are estimated at more than $12 billion per year.1 For approximately 16 million people (8%), back pain is persistent or chronic, leading to increased costs of health care and lost wages from work and disability payments.1 Physical therapy is a common treatment choice for patients with LBPS—a quarter of all referrals for outpatient physical therapy and one half of all outpatient physical therapy visits are related to patients with LBPS.2,3

Measuring and improving quality of rehabilitation services for patients with LBPS should be a priority for health services researchers, given the prevalence and economic impact of LBPS.2,4 Quality of rehabilitation services has been defined by the Institute of Medicine as “the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge.”5 However, research on health service delivery for patients with LBPS in outpatient rehabilitation is in its infancy, with only a few studies describing service delivery factors related to patient outcomes.

Two studies6,7 suggest that staffing patterns are associated with outcomes in outpatient physical therapy. In a study comparing therapists classified as expert and average on the basis of patient self-reported health-related quality-of-life (HRQL) outcomes, Resnik and Jensen6 found that different patterns of staffing, service delivery, and delegation of care to support personnel were observed. In a follow-up study, Resnik et al7 reported that patients who spent more than half of their treatment episode of care with a physical therapist assistant reported worse functional outcomes and utilized more visits compared with patients with less physical therapist assistant involvement.

Although there is a paucity of research focusing on physical therapy staffing, appropriate staffing levels have been shown to be associated with quality measures and adverse patient outcomes in inpatient and long-term care. Adequate nursing staffing levels, for example, are associated with decreased likelihood of adverse events such as pneumonia and pressure ulcer development in inpatient hospital settings.8 Although physical therapy treatment is rendered under the direction and supervision of a primary physical therapist, portions of the treatment plan are sometimes delivered by support personnel such as physical therapist assistants. Based on prior research, we expect that in clinics that have higher ratios of physical therapists to physical therapist assistants, more of the direct care would be delivered by physical therapists and that this would positively affect patient outcomes and result in lower utilization of services.

It is generally assumed that practitioners must possess many years of clinical experience to achieve the best results with patients and that years of experience are associated with better clinical outcomes.911 However, 2 recent studies that evaluated the relationship between therapist years of experience and patient outcomes have challenged this assumption.6,12 Neither years of experience6,12 nor the presence of an experienced therapist on staff12 was associated with improved patient outcomes in outpatient rehabilitation. However, neither of these studies evaluated the impact of therapist experience on the combination of outcomes and utilization of services.

Prior research suggests that expert therapists utilize collegial knowledge more often than average therapists.13 Given this finding, we considered whether the size of the professional staff within a clinic would be related to clinic outcomes and utilization. Therapists working in solo practice or small settings would have fewer opportunities to engage with colleagues for consultation on challenging cases. To our knowledge, the relationship between size of the professional staff and clinic outcomes and utilization had not been previously examined.

Research on medical care indicates clinics that specialize in treatment of a specific population or that have a larger volume of a specific population are assumed to have better outcomes of care compared with smaller, less specialized clinics. Volume of patients has been shown to be related to better outcomes of care for patients with angiography,14 cardiac surgery,15,16 aneurisms,17 and cancer.18 If a similar pattern were found in outpatient rehabilitation, clinics that serve a higher volume of patients with LBPS would have better treatment outcomes. However, no previous research has examined the relationship between volume of patients with specific types of musculoskeletal problems and outcomes or service utilization in outpatient rehabilitation clinics.

We also questioned whether practice setting might be associated with outcomes and utilization of care. As the health care industry becomes increasingly for-profit, concerns have surfaced that quality of care has suffered.19,20 Financial incentives to provide fewer services have aroused suspicions that some providers may not be providing the highest care to patients.19,20 Despite these concerns, there is little research on the effect of physical therapist practice settings on patient outcomes and service utilization.

Given the need to understand organizational and service delivery factors associated with outcomes and utilization of physical therapy care of patients with LBPS, the purpose of this study was to identify organizational and service delivery characteristics related to practice outcomes and utilization.

Method

Subjects

We conducted a secondary analysis of previously collected data from the Focus On Therapeutic Outcomes, Inc (FOTO) database. The FOTO database is the largest outpatient rehabilitation outcomes database available for researchers in the United States. FOTO is an international medical rehabilitation data management company that was established in 1992 to develop a standardized measurement set for outpatient orthopedic rehabilitation outcomes21,22 FOTO provides a standardized set of data collection instruments and contains demographic, intake and discharge HRQL (instrument described below), and administrative data from outpatient rehabilitation services, as well as data on characteristics of health care providers and organizations. Providers pay an annual and per-patient fee to participate in the FOTO data collection system.

FOTO provided the first author with the data set of patients from 2000–2001. Our analytical sample was selected from a larger data set of clinics (n=487) treating patients with a variety of syndromes that participated in the FOTO database in 2000–2001. Clinics were included in the sample if they met the following criteria: (1) had completed clinician and facility registration forms (n=398), (2) had at least one physical therapist on staff (n=379), (3) had entered intake data for at least 40 patients with LBPS (n=258), and (4) had follow-up data on at least two thirds of its patient population (n=283). Three hundred seventy-three clinics were excluded because they did not meet one or more of these criteria. Thus, the final analytic sample consisted of 114 outpatient clinics with 1,058 therapists. All patients in the analytic sample were being treated for a low back problem; all other patients were excluded from the sample. The final study sample consisted of 16,281 patients with LBPS.

Measures

This study used a large number of variables in 2 different types of analyses (described in the “Data Analysis” section): (1) hierarchical linear models and (2) multinomial logistic regression models. Table 1 provides a synopsis of the variables included in each type of analysis. Each of the variables is described below.

Table 1.
Type and List of Variables Used in Each Analysisa

Dependent variables.

In this study, we classified clinics based upon measures of both patient outcomes and service utilization. The patient outcome measure was the FOTO overall health status measure (OHS), an HRQL measure derived from the Medical Outcomes Study 36-Item Short-Forum Health Survey (SF-36) that assesses both mental and physical dimensions of health. Health constructs of the SF-36 are the primary generic HRQL measures with which other self-report measures are commonly compared.23 Generic instruments are designed for broad use in a variety of patient populations. The SF-36 has been studied in populations of patients with common orthopedic diagnoses, and measures of reliability and validity have been published.3,2426 Internal consistency of items in the OHS constructs with 2 or more items has been reported (α=.57–.91).27,28 Internal consistency reliability statistics of the items of the OHS constructs27,28 were comparable to internal consistency reliability statistics calculated from the same items embedded in the SF-36 questionnaire29 and the 12-Item Short-Form Health Survey (SF-12) questionnaire.30 Test-retest reliability of data obtained with the OHS was good (intraclass correlation coefficient [2,1]=.92).28 Responsiveness in the treatment of patients with low back pain (effect size=0.83) and the validity of the OHS measure to discriminate expert from average therapists have been reported.6

The OHS scores were calculated by averaging scores from the 8 embedded HRQL constructs: general health (1 item from the SF-12),30 physical functioning (10 items from the physical functioning scale [PF-10] of the SF-36),29 role–physical (2 items from the SF-12),30 bodily pain (2 items from the SF-36),29 vitality (1 item from the SF-12),30 mental health (2 items from the SF-12),30 role–emotional (2 items from the SF-12),30 and social functioning (1 item from the SF-12).30 The OHS physical functioning construct also included 3 new questions pertinent to clients with upper-extremity impairments.31 Scoring of item responses followed published algorithms.29 Raw ordinal scores were transformed to interval scores varying from 0 to 100 for each question, with higher values indicating better health.29 Transformed item scores were grouped by construct and averaged to obtain the score for each of the 8 OHS functional scales. The service utilization measure was the number of visits per treatment episode as recorded by the therapist at the time of discharge.

Primary independent variables.

We used the following organizational characteristics in our analyses: size of the professional staff, ratio of full-time equivalent (FTE) physical therapists to physical therapist assistants, average years of experience of physical therapy staff in the clinic, volume of new patients, proportion of patients with LBPS, and type of practice setting. Size of the professional physical therapy staff was a continuous variable, defined as the number of FTE physical therapists employed at each clinic. The staffing ratio of full-time physical therapists to physical therapist assistants was calculated by dividing the number of FTE physical therapist assistants by the number of FTE physical therapists. Type of setting was classified as outpatient hospital clinic or “other.”

Average years of experience of physical therapists was estimated by calculating the mean number of years of clinical experience of FTE physical therapists. Volume of new patients per month was defined as the number of new patients per month divided by the number of FTE physical therapists. The proportion of patients with LBPS was calculated as the ratio of number of patients with LBPS divided by the total number of all types of patients seen at the clinic.

We used 2 measures of service delivery characteristics in our analyses: overall volume of new patients at the clinic per therapist and a measure of utilization of the physical therapist assistant. Overall volume of new patients at the clinic per therapist was calculated by dividing the number of new patients seen at the facility each month by the number of FTE physical therapists on staff. The measure of utilization of the physical therapist assistant was a variable we created by combining data from the patient discharge survey where the clinician was asked to estimate the percentage of time that each type of provider (including physical therapist, physical therapist assistant, and physical therapy aide) spent in that specific patient's episode of care. At the time of discharge, providers had the option of checking 1 of 5 possible responses for proportion of the treatment episode that the patient had spent with the physical therapist assistant: 0%, 1% to 25%, 26% to 50%, 51% to 75%, or 76% to 100%. We used the data to calculate the percentage of patients within each clinic who therapists reported had spent more than 50% of their treatment episode with a physical therapist assistant, which we considered to be “higher physical therapist assistant utilization.” We used this cut-point because prior research7 showed that patients who spent more than 50% of their treatment episode with a physical therapist assistant had poorer outcomes and increased service utilization (more visits).

Because we were interested in examining the impact of patterns of service delivery within clinics on clinic aggregate outcomes, we sought to create an indicator variable for whether or not the clinic had an overall pattern of high physical therapist assistant utilization. Thirty-six percent of the clinics in our sample had no patients with higher physical therapist assistant utilization, creating a highly skewed variable distribution. Therefore, we first removed clinics with no instances of higher physical therapist assistant utilization and calculated the percentage of patients within each of the remaining clinics that had higher physical therapist assistant utilization. We then created a dichotomous variable to indicate lower or higher physical therapist assistant utilization. We classified clinics as lower physical therapist assistant utilizers if they had no patients in the clinic with high physical therapist assistant utilization or if their percentage of patients with higher physical therapist assistant utilization was below the mean of those with any higher physical therapist assistant utilizers. We classified clinics as higher physical therapist assistant utilizers if the percentage of patients with higher physical therapist assistant utilization within the clinic was above the mean.

Patient-level variables.

We used the following variables to control for potential confounding due to patient characteristics: sex, age, severity of condition at intake, number of days since onset of condition, primary diagnosis, surgeries, reimbursement type, exercise history, and employment status. Severity of the condition was assessed by the intake score of the OHS scale. The variable called “onset of condition” represents the number of days from the onset of the condition until the beginning of intervention. We classified onset of condition as: acute (0–21 days), subacute (22–90 days), or chronic (91 days or more). Primary diagnosis was classified using the diagnostic categories for LBPS as defined by Hart et al32 and Freburger et al33: herniated disk, spinal stenosis, spondylosis, sprain or strain, pain syndrome, deformity, or “other.” Number of surgeries for the low back was categorized as: none, 1, 2, 3, and 4 or more. Reimbursement type is the primary source of the payment for the patient's physical therapy, classified as: indemnity, litigation, Medicaid, Medicare, patient, health maintenance organization or preferred provider organization, workers’ compensation, or “other.” Exercise history is a measurement of the patient's self-reported exercise prior to the episode of physical therapy, classified as: at least 3 times a week, 1 to 2 times a week, or seldom or never. Employment status at the time of intake for physical therapy was classified as: full-time, modified work, employed but not working, previously employed and receiving disability, unemployed, retired, or student.

Procedure

Data collection procedures have been described elsewhere.27,31 When a clinic starts collecting FOTO data, staff provide information describing the clinic, and clinic staff are trained in the data collection process. Patients complete self-report health status surveys prior to their initial evaluation and following discharge from their rehabilitation episode. Patient demographic data are collected at intake. Clinicians enter number of visits and treatment dates at discharge. Data from patients and staff are entered on paper surveys, which are submitted to FOTO where data are checked manually for completeness. Complete data are entered into a computer data bank where computer programs check to ensure data are complete and within appropriate ranges for each variable. Data identified as incomplete or inappropriate are returned to the clinic for correction.

Data Analysis

Censoring.

Preliminary analyses were conducted to estimate potential bias introduced by omitting clinics from the sample due to incomplete data. We compared differences in characteristics for the clinics included and excluded from the analytic sample using analyses of variance for continuous variables and Pearson chi-square tests for categorical variables.

Modeling of outcomes and utilization.

We used statistical risk adjustment, also known as case-mix adjustment, techniques to control for effects of confounding variables seen in patient populations.3436 Using SAS version 9.1,*,36 two separate 3-level hierarchical linear models (HLMs)7 were constructed: one for OHS score at discharge and one for number of visits in the treatment episode. Multilevel models were appropriate for this analysis, given the hierarchically structured data, where patients are nested within therapists and therapists are nested within practice or clinic and where the primary goal is to compare clusters. Due to the complexity of these models, a full description is presented in our companion article in the issue.37 In summary, these models controlled for potential confounders at the patient level (see description of risk-adjustment variables below) and adjusted for bias due to missing data through the use of inverse probability weighting.38 Because selection of the clinic or reason why a patient receives treatment in one specific clinic and not another might be related to expected outcomes, we added 2 sets of variables at the clinic level in an attempt to control for selection bias. These variables were the volume of new patients per month and variables representing the proportion of patients referred by physician type (see descriptions of variables in the “Measures” section).

Classification of clinics into performance groups.

To determine the mean patient OHS outcomes and mean number of visits per clinic, we aggregated residual scores within each clinic after fitting the 3-level models to form 2 separate clinic-specific residual scores, one for OHS scores and one for visits. Residual scores are the difference between actual scores and the predicted scores after risk-adjusted modeling. As such, they represent the differences in outcome, as predicted by variables included in the model and the actual outcomes that patients obtained in treatment. Thus, a patient who did better than predicted would have a higher residual score, and a patient who did worse than predicted would have a lower residual score. Thus, clinics whose patients did better than predicted would have higher mean OHS residual scores than clinics whose patients had OHS outcomes as expected or worse. The rationale for using residual scores to estimate provider performance is that residuals represent the amount of variance unexplained by the models that could potentially be explained by factors other than patient characteristics, including type of treatment given or some other aspect of service delivery such as care continuity. Other researchers39,40 have used differences as well as ratios between observed and expected rates of change to assess performance of providers. The use of residuals to estimate provider performance has been used in previous studies that classified expert and average physical therapists.6,41

We used the clinic-specific aggregated residual score after modeling OHS scores to classify clinics into 3 effectiveness groups, which were determined based on the distribution of the residual scores. We considered the upper 76th–100th percentile of residual scores to denote the high effectiveness group, the 26th–75th percentile to be the middle group, and the 1st–25th percentile to be the low effectiveness group. We used the clinic-specific aggregated residual scores after modeling of visits to classify clinics into 3 utilization groups based on the distribution of the residual scores. We considered clinics with residual scores in the 1st–25th percentile to be the low utilizers, clinics in the 26th–75th percentile to be the middle utilizers, and those in the 76th–100th percentile to be the high utilizers.

Finally, we classified clinics into overall performance groups based on the combination of the effectiveness and visit classification categories described above. We found little precedent in the literature to guide our classification of clinics using a combination of visits and outcomes. However, a recent pay-for-performance demonstration project funded by the Centers for Medicare and Medicaid Services classified clinics on the basis of both outcomes and visits into 9 performance groups.42 We expanded on the approach used in that demonstration project to develop a classification scheme that we believed had face validity by classifying clinics into the best performance group if they were in the upper quartile of effectiveness and had a middle or low quartile of utilization. Clinics were classified into the middle performance group if they were in the middle quartiles of effectiveness and were in the best (ie, lowest) utilization group or if they were in the high effectiveness group but were in the low utilization group. Clinics were classified into the worst performance group if they were in the low or middle effectiveness group and the high utilization group (Figure).

Figure.
Classification of clinics into performance groups.

Relationship between clinic groups and clinic characteristics.

We examined the relationships between organizational and service delivery characteristics (independent variables) and the dependent variables of: (1) clinic effectiveness group, (2) utilization group, and (3) clinic overall performance group. For ease of interpretation, we performed multinomial logistic regressions within each of the groups in each case using the group we considered the “lowest quality” as the reference group. Thus, the first model compared the high and middle effectiveness groups with the group with lowest effectiveness, the second model compared the lowest and middle utilization groups with the group with the highest (or most) utilization, and the third model compared the best and middle overall performance groups with the worst overall performance group. Each of the multinomial regression models included the following variables: size of the professional staff, ratio of FTE physical therapists to physical therapist assistants, average years of experience of physical therapy staff in the clinic, volume of new patients, proportion of patients with LBPS, type of practice setting, and utilization of the physical therapist assistant.

Results

Preliminary Analyses of Biased Censoring

Table 2 shows the descriptive characteristics of the clinics included in the sample compared with those excluded from the sample. Consistent with the sampling criteria, clinics included in the sample had a higher proportion of patients with LBPS and more complete data than clinics omitted from the sample. The OHS intake scores were significantly lower (worse) for patients in the clinics omitted from analysis compared with patients of the sample group. Clinics included in the sample had more physical therapists and physical therapist assistants on staff and saw more new patients per month compared with clinics excluded from the sample. There was a higher percentage of men in clinics retained in the sample compared with clinics omitted from the sample. A greater percentage of patients in the clinics omitted from the sample had acute symptoms compared with clinics selected for the analytic sample, and a lower percentage of patients in the clinics omitted from the sample had chronic symptoms compared with patients in clinics selected for the analytic sample. Finally, a greater proportion of clinics included in the analytic sample were hospital based compared with clinics excluded from the sample.

Table 2.
Characteristics of Clinics Included and Excluded From Analytic Samplea

Characteristics of Analytic Sample

Sample characteristics with descriptors of all patient-level variables used in the analysis are shown in Table 3. Almost half of the sample had chronic low back pain. Most (82.0%) of patients had not had prior surgery for their condition. The top 3 diagnoses were pain (34.8%), sprain or strain (25.5%), and herniated disk (19.3%). On average, patients improved more than 16 OHS points during therapy.

Table 3.
Characteristics of Patients by Overall Performance Groupa

Characteristics of clinics by overall performance group are shown in Table 4. Unadjusted mean OHS discharge scores were 73.9 (SD=17.9) for the best clinic performance group, 67.9 (SD=19.4) for the middle clinic performance group, and 65.1 (SD=20.6) for the worst clinic performance group. On average, patients in the best clinic performance group improved 19.2 OHS points during therapy, while patients in the worst clinic performance group improved an average of 16.4 OHS points. Patients in the best clinic performance group utilized, on average, 7.7 (SD=4.1) visits per treatment episode compared with 7.9 (SD=4.1) in the middle clinic performance group and 9.3 (SD=4.9) in the worst clinic performance group. One hundred nine clinics were classified into performance groups. Five clinics were dropped in the modeling process because they were missing data on number of new patients per month.

Table 4.
Characteristics of Clinics by Overall Performance Groupa

Results of Hierarchical Modeling

The results of the HLM for OHS are shown in Table 5, and the results of the HLM for visits are shown in Table 6. The primary purpose of these models in the current study was for analytic control of confounding (ie, risk adjustment) so as to obtain the residual scores from these models for use in classifying the clinics into effectiveness group, visits group, and overall performance group as described in the “Method” section. Thus, the results of these analyses will not be discussed in detail in this report.

Table 5.
Three-Level Linear Regression Model Results for Residual Discharge FOTO Overall Health Status (OHS) Measure (N=16,281 Patients Treated by 1,058 Physical Therapists in 114 Clinics)a
Table 6.
Three-Level Linear Regression Model Results for Visits per Treatment Episode (N=16,281 Patients Treated by 1,058 Physical Therapists in 114 Clinics)a

Explaining Clinic Variation in Outcomes, Utilization, and Overall Performance Groups

The results of the multinomial logistic regression of effectiveness groups (Tab. 7) showed that clinics that were lower utilizers of physical therapist assistants were 6.6 times more likely to be classified into the high effectiveness group compared with the low effectiveness group (P=.04), 6.7 times more likely to be classified into the low utilization group compared with the high utilization group (P=.03), and 12.4 times more likely to be classified into the best performance group compared with the worst performance group (P=.03).

Table 7.
Results of Multinomial Logistic Regressions: Probability of Being in Each Clinic Group by Organizational and Staffing Characteristicsa

Clinics that had a higher proportion of patients with LBPS were more likely to be classified into the lowest utilization group (P=.001) or the middle utilization group (P=.009). The model results can be interpreted as follows: each 1-point increase in the proportion of patients with LBPS treated in the clinic would lead to a 1.2 times increased likelihood of being classified into the lower utilization group compared with the highest utilization group and to a 1.1 times greater likelihood of being classified into the middle utilization group compared with the highest utilization group. Staff years of experience was inversely associated with being classified into the middle utilization group compared with the highest utilization group (P=.05). No other differences were noted between any groups in ratio of physical therapists to support staff, number of patients per therapist per month, number of full-time physical therapists on staff, or practice setting.

Discussion

We found that quality of physical therapy clinics, as estimated by grouped, aggregated, risk-adjusted measures of outcomes, service utilization, and a composite measure of overall performance, was related to organizational and service delivery factors. Our strongest finding was that clinics that had lower utilization of physical therapist assistants were much more likely to be in the “best” category of each type of group (ie, highest effectiveness, lowest utilization, and overall performance). Together, these findings suggest that use of physical therapist assistants in place of physical therapists is associated with less effective care delivery and higher utilization of services in the treatment of patients with LBPS. Findings in the current study support previous findings that high physical therapist assistant utilization was associated with lower functional outcomes.6

Our results also suggested that clinics with therapists who had, on average, more years of experience were less likely to be classified into the middle utilization group compared with the highest utilization group. We found no association between years of experience and likelihood of being in the lowest utilization group compared with the highest utilization group. Thus, this result should be interpreted with caution. One possible explanation for this finding is that more recent graduates have entered the profession in times of more limited resources and reimbursement and, thus, might utilize visits more conservatively than “older” graduates. However, this hypothesis would need to be tested in future studies.

We found that clinics that treated a higher proportion of patients with LBPS were more likely to be classified into the lower and middle utilization groups compared with the highest utilization group, suggesting that greater volume of patients with LBPS contributed to more care efficiency. However, volume of patients with LBPS was not associated with either effectiveness of care or overall performance group.

Limitations

Our analysis has a number of limitations. The study's cross-sectional design limits our ability to make causal inferences regarding the relationship of staffing utilization, experience of staff, visits utilization, and outcomes. Due to the observational design, we can observe associations among variables and generate hypotheses, but we cannot establish causal relationships.

We recognize that selection of clinics into our sample that had at least an average amount of uncensored data is a limitation of our design in that providers with more complete data may be more alike than providers who have less complete data and may be of higher quality or have other characteristics that are endogenous to the study outcome measure (discharge OHS score). We were able to provide control of selection bias of patients within our sample clinics due to missing data by using statistical techniques, but we could not account for differences between clinics selected for study and those not selected. We evaluated and found differences between the group of clinics included in the sample (ie, those that had complete follow-up data on at least 67% of their patients) and the group of clinics omitted from the sample. Specifically, clinics included in the sample were more likely to be hospital based; have larger professional staffs; and see a greater proportion of patients with LBPS, patients with chronic conditions, and patients who have greater functional disability compared with clinics excluded from the sample. One explanation, which awaits future analysis, for some of these differences is that hospital-based clinics are more rigorous in their data collection efforts because they have a larger professional staff and are motivated to collect performance data to meet accreditation requirements.

There are limits to the generalizability of our conclusions to the full population of clinics participating in the FOTO database. Furthermore, we do not know whether our sample of clinics is representative of physical therapist practices in the United States. We are not aware of any available national data quantifying the number of physical therapy clinics in the United States or their characteristics. Our sample included clinics that participated in the FOTO database. No effort was made to contact clinics that do not participate in this database. Because FOTO is an independent data management company with practices choosing for a variety of reasons to participate, there are threats to external validity. Thus, our findings cannot be generalized to clinics that do not participate with FOTO.

There also are limitations in the strategy we used to classify clinics into groups based on outcomes and utilization. We believe that a performance classification based on a combination of outcomes and utilization makes sense. Because the ultimate concern of public agents and payers is the providers’ impact on patient outcomes, many quality measures that have been developed are outcomes based. Several health care quality indicators are based on measurement of the patient's performance or self-reported performance of functional tasks using HRQL instruments. For example, several of the nursing home quality indicators are based on performance measurement data available in the Minimum Data Set assessments.43 As another example, the Health Outcomes Survey conducted by the Centers for Medicare and Medicaid Services uses the SF-36, a patient self-assessment tool, to assess health outcomes of Medicare beneficiaries in managed care settings and to measure performance of health plans.44 In some instances, due to either measurement problems or ambiguity about what an outcome of a service is, many systems still focus on process measures in their assessment of provider quality. Because there is little ambiguity about the goal of physical therapy for people with low back pain, we believe that measuring provider performance via patient outcomes and utilization is a reasonable approach. We recognize that, in the future, other investigators might choose to categorize clinics using different methods.

Our statistical techniques, although more sophisticated than previous models, could be further improved by controlling for other potential confounders, which were not available in the FOTO database in 2000–2001. There is preliminary evidence that the addition of an index of comorbidities related to functional outcomes, for example, may control additional variance of change in functional status outcomes.4245 Unfortunately, functional comorbidity data were not available in our data set. In addition, there are several variables related to the classification of patients with LBPS into more homogeneous groups that, if available, might improve the risk-adjustment model.4650 Finally, the addition of other variables related to psychosocial constructs, such as fear avoidance,51 may be helpful in future risk-adjustment models.

Conclusions

We found that outpatient rehabilitation performance in the care of patients with LBPS, as estimated by risk-adjusted measures of outcome and utilization, was related to organizational and service utilization factors. Clinics that were lower utilizers of physical therapist assistants were characterized as more effective and lower utilizers of visits and, as such, also were more likely to be classified as higher performing. Clinics with higher proportions of patients with LBPS were characterized as lower utilizers of services, but they did not have better outcomes or better overall performance. These findings add to the body of knowledge on organizational and service delivery factors related to treatment of patients with LBPS.

Notes

Dr Resnik, Dr Liu, and Dr Hart provided concept/idea/research design. All authors provided writing. Dr Resnik, Dr Liu, and Dr Mor provided data analysis. Dr Resnik provided project management and fund procurement. Dr Hart provided data collection, subjects, and consultation (including review of manuscript before submission). The authors acknowledge the invaluable assistance of Sharon-Lise T Normand, PhD, who served as statistical consultant for this study.

Funding for this research was provided by the National Institute of Child Health and Human Development (grant 1RO3HD051475-01).

Poster presentations of this research were given at the Combined Sections Meeting of the American Physical Therapy Association; February 4–8, 2004; Nashville, Tennessee, and at Academy Health Annual Research Meeting; June 3–5, 2006; Seattle, Washington. A platform presentation of this research was given at the Combined Sections Meeting of the American Physical Therapy Association; February 14–18; Boston, Massachusetts.

Dr Hart is an employee of and investor in Focus On Therapeutic Outcomes, Inc, the database management company that manages the data analyzed in this study.

*SAS Institute Inc, PO Box 8000, Cary, NC 27513.

References

1. Druss BG, Marcus SC, Olfson M, Pincus HA. The most expensive medical conditions in America. Health Aff (Millwood). 2002;21:105–111. [PubMed]
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. [PubMed]
3. Di Fabio RP, Boissonnault W. Physical therapy and health-related outcomes for patients with common orthopaedic diagnoses. J Orthop Sports Phys Ther. 1998;27:219–230. [PubMed]
4. Frymoyer JW, Cats-Baril WL. An overview of the incidences and costs of low back pain. Orthop Clin North Am. 1991;22:263–271. [PubMed]
5. Crossing the Quality Chasm: The IOM Quality Initiative. Available at: http://www.iom.edu/CMS/8089.aspx. Accessed June 19, 2008.
6. Resnik L, Hart DL. Using clinical outcomes to identify expert physical therapists. Phys Ther. 2003;83:990–1002. [PubMed]
7. Resnik L, Feng Z, Hart DL. State regulation and the delivery of physical therapy services. Health Serv Res. 2006;41(4 pt 1): 1296–1316. [PMC free article] [PubMed]
8. Cho SH, Ketefian S, Barkauskas VH, Smith DG. The effects of nurse staffing on adverse events, morbidity, mortality, and medical costs. Nurs Res. 2003;52:71–79. [PubMed]
9. Jensen GM, Shepard KF, Hack LM. The novice versus the experienced clinician: insights into the work of the physical therapist. Phys Ther. 1990;70:314–323. [PubMed]
10. Jensen GM, Shepard KF, Gwyer J, Hack LM. Attribute dimensions that distinguish master and novice physical therapy clinicians in orthopedic settings. Phys Ther. 1992;72:711–722. [PubMed]
11. Jensen GM, Gwyer J, Shepard KF, Hack LM. Expert practice in physical therapy. Phys Ther. 2000;80:28–52. [PubMed]
12. Constance DP. Effect of Experience on Physical Therapist Functional Outcomes [master's thesis]. St Augustine, FL: University of St Augustine; 2000.
13. Resnik L, Jensen GM. Using clinical outcomes to explore the theory of expert practice in physical therapy. Phys Ther. 2003;83:1090–1106. [PubMed]
14. Shook TL, Sun GW, Burstein S, et al. Comparison of percutaneous transluminal coronary angioplasty outcome and hospital costs for low-volume and high-volume operators. Am J Cardiol. 1996;77:331–336. [PubMed]
15. Hannan EL, Siu AL, Kumar D, et al. The decline in coronary artery bypass graft surgery mortality in New York State: the role of surgeon volume. JAMA. 1995;273:209–213. [PubMed]
16. Shahian DM, Normand SL. The volume-outcome relationship: from Luft to Leapfrog. Ann Thorac Surg. 2003;75:1048–1058. [PubMed]
17. Berman MF, Solomon RA, Mayer SA, et al. Impact of hospital-related factors on outcome after treatment of cerebral aneurysms. Stroke. 2003;34:2200–2207. [PubMed]
18. Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol. 2000;18:2327–2340. [PubMed]
19. Council on Ethical and Judicial Affairs. Ethical issues in managed care. JAMA. 1995;273:334–335. [PubMed]
20. Feldstein P. The Politics of Health Legislation: An Economic Perspective. Chicago, IL: Health Administration Press; 1996.
21. Dobrzykowski EA, Nance T. The Focus On Therapeutic Outcomes (FOTO) Outpatient Orthopedic Rehabilitation Database: results of 1994–1996. Journal of Rehabilitation Outcomes Measurement. 1997;1(1): 56–60.
22. Swinkels ICS, van den Ende CHM, de Bakker D, et al. Clinical databases in physical therapy. Physiother Theory Pract. 2007;23:153–167. [PubMed]
23. Ware JE Jr. SF-36 health survey update. Spine. 2000;25:3130–3139. [PubMed]
24. Gatchel RJ, Mayer T, Dersh J, et al. The association of the SF-36 health status survey with 1-year socioeconomic outcomes in a chronically disabled spinal disorder population. Spine. 1999;24:2162–2170. [PubMed]
25. Bronfort G, Bouter LM. Responsiveness of general health status in chronic low back pain: a comparison of the COOP charts and the SF-36. Pain. 1999;83:201–209. [PubMed]
26. Gatchel RJ, Polatin PB, Mayer TG, et al. Use of the SF-36 Health Status Survey with a chronically disabled back pain syndrome: strengths and limitations. J Occup Rehabil. 1998;8:237–246.
27. Hart DL, Wright BD. Development of an index of physical functional health status in rehabilitation. Arch Phys Med Rehabil. 2002;83:655–665. [PubMed]
28. Hart DL. Test-retest reliability of an abbreviated self-report overall health status measure. J Orthop Sports Phys Ther. 2003;33:734–744. [PubMed]
29. Ware JE Jr, Snow KK, Kosinksi M, Gandek B. SF-36 Health Survey: Manual and Interpretation Guide. Boston, MA: The Health Institute, New England Medical Center; 1993.
30. Ware JE Jr. How to Score the SF-12 Physical and Mental Heatlh Summary Scales. Boston, MA: The Health Institute, New England Medical Center; 1995.
31. Hart DL. The power of outcomes: FOTO Industrial Outcomes Tool-Initial Assessment. Work. 2001;16:39–51. [PubMed]
32. Hart LG, Deyo RA, Cherkin DC. Physician office visits for low back pain: frequency, clinical evaluation, and treatment patterns from a U.S. national survey. Spine. 1995;20:11–19. [PubMed]
33. Freburger JK, Carey TS, Homes GM. Physician referrals to physical therapists for the treatment of spine disorders. Spine J. 2005;5:530–541. [PubMed]
34. Iezzoni LI, ed. Risk Adjustment for Measuring Health Care Outcomes. Ann Arbor, MI: Health Administration Press; 1994.
35. Iezzoni LI. An introduction to risk adjustment. Am J Med Qual. 1996;11:S8–S11. [PubMed]
36. Iezzoni LI. Risk adjusting rehabilitation outcomes: an overview of methodologic issues. Am J Phys Med Rehabil. 2004;83:316–326. [PubMed]
37. Resnik L, Liu D, Hart DL, Mor V. Benchmarking physical therapy clinic performance: statistical methods to enhance internal validity when using observational data. Phys Ther. 2008;88:1078–1087. [PMC free article] [PubMed]
38. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11:550–560. [PubMed]
39. Selim AJ, Berlowitz D, Fincke G, et al. Use of risk-adjusted change in health status to assess the performance of integrated service networks in the Veterans Health Administration. Int J Qual Health Care. 2006;18:43–50. [PubMed]
40. Liu CF, Sales AE, Sharp ND, et al. Case-mix adjusting performance measures in a veteran population: pharmacy- and diagnosis-based approaches. Health Serv Res. 2003;38:1319–1337. [PMC free article] [PubMed]
41. Resnik L, Hart DL. Influence of advanced orthopaedic certification on clinical outcomes of patients with low back pain. Journal of Manual & Manipulative Therapy. 2004;12:32–41.
42. Hart DL, Connolly J. Pay-for-Performance for Physical Therapy and Occupational Therapy: Medicare Part B Services. Grant #18-P-93066/9–01. Washington, DC: US Dept of Health & Human Services, Centers for Medicare & Medicaid Services; 2006.
43. Zimmerman DR. Improving nursing home quality of care through outcomes data: the MDS quality indicators. Int J Geriatr Psychiatry. 2003;18:250–257. [PubMed]
44. Cooper JK, Kohlmann T, Michael JA, et al. Health outcomes. New quality measure for Medicare. Int J Qual Health Care. 2001;13:9–16. [PubMed]
45. Hart DL, Wang Y, Stratford PW, Mioduski J. Computerized adaptive test for patients with knee impairments produced valid and sensitive measures of function. J Clin Epidemiol. In press. [PubMed]
46. Werneke M, Hart DL, Cook D. A descriptive study of the centralization phenomenon: a prospective analysis. Spine. 1999;24:676–683. [PubMed]
47. Werneke M, Hart DL. Centralization phenomenon as a prognostic factor for chronic low back pain and disability. Spine. 2001;26:758–764. [PubMed]
48. 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;75:470–489. [PubMed]
49. Fritz JM, Delitto A, Erhard RE. Comparison of classification-based physical therapy with therapy based on clinical practice guidelines for patients with acute low back pain: a randomized clinical trial. Spine. 2003;28:1363–1371; discussion 1372. [PubMed]
50. Wilson L, Hall H, McIntosh G, Melles T. Intertester reliability of a low back pain classification system. Spine. 1999;24:248–254. [PubMed]
51. Waddell G, Newton M, Henderson I, et al. A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in chronic low back pain and disability. Pain. 1993;52:157–168. [PubMed]

Articles from Physical Therapy are provided here courtesy of American Physical Therapy Association
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • MedGen
    MedGen
    Related information in MedGen
  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...