Associations Between Treatment Processes, Patient Characteristics, and Outcomes in Outpatient Physical Therapy Practice

Article Outline

• Abstract
• Methods
  • Design
  • Subjects
  • Clinics and Therapists
  • Patient Characteristics and Treatment Process Data Collection
  • Outcomes Data Collection
  • Electronic Database Preparation
  • Data Analysis
• Results
  • Subjects
  • Patient Demographic and Health Data
  • Treatment Process
  • Associations Between Patient Characteristics, Process of Care, and Outcomes
• Discussion
  • Patient Demographic and General Health
  • Process-Related and Intervention Variables
  • Study Limitations
• Conclusions
• Acknowledgments
• Appendix 1
• References
• Copyright

Abstract 

Deutscher D, Horn SD, Dickstein R, Hart DL, Smout RJ, Gutvirtz M, Ariel I. Associations between treatment processes, patient characteristics, and outcomes in outpatient physical therapy practice.

Objective

To identify how treatment processes are related to functional outcomes for patients seeking treatment for musculoskeletal impairments while controlling for demographic and health characteristics at intake.

Design

Prospective, observational cohort study. Treatment processes were not altered. Data were collected continuously from June 2005 to January 2008. Descriptive statistics were applied to compare patient characteristics, interventions, and outcomes between impairment categories. Ordinary least-squares multiple regressions were used to examine associations between patient characteristics at intake, treatment processes, and functional outcomes.

Setting

Fifty-four community-based outpatient physical therapy clinics of Maccabi Healthcare Services, a public health plan in Israel.

Participants

A consecutive sample of 22,019 adult patients (mean age 51.2y, standard deviation=15.7, range 18–96, 58% women) seeking treatment due to lumbar spine, knee, cervical spine, or shoulder impairments with functional measurements at intake and discharge.

Interventions

Not applicable.

Main Outcome Measure

Functional status at discharge.

Results

Explanatory power ranged from 30% to 39%. Better outcomes were associated with patient compliance with self-exercise and therapy attendance, application of therapeutic exercise and manual therapy, and completion of 3 or more functional surveys during the episode of care. Worse outcomes were associated with women, electrotherapy for pain management, and therapeutic ultrasound for shoulder impairments. Mixed results were found for group exercise programs.

Conclusions

The study of associations between treatment processes, patient characteristics, and outcomes helps to describe practice and can be used to suggest ways to improve outcomes in outpatient physical therapy practice.

Key Words: Database, Evidence-based practice, Musculoskeletal system, Rehabilitation, Therapy, physical (specialty)

List of Abbreviations: APTA, American Physical Therapy Association, BMI, body mass index, CAT, computerized adaptive testing, EHR, electronic health record, FS, functional status, ICD-9, International Classification of Diseases, 9th Revision, IRT, item response theory, PT, physical therapy

PHYSICAL THERAPISTS are continuously challenged to provide evidence of contribution to patients' health. Desired outcomes of PT include enhanced physical and functional abilities and restored, maintained, or promoted optimal physical function, wellness, fitness, and optimal quality of life related to movement and health.¹ The World Confederation for PT promotes physical functioning as a primary goal of the profession,² emphasizing the importance of activities and participation components as defined in the International Classification of Functioning, Disability and Health.³ Therefore, physical functioning, defined as a patient's ability to perform activities,⁴, ⁵ is an essential construct that should guide clinical reasoning when establishing treatment goals and assessing outcomes.

Improvement in PT services and outcomes depends on the therapist's ability to identify treatment processes that are related to best outcomes, controlling for large variability in patient characteristics. To date, although research in PT is increasing, reports on associations between treatment processes, patient characteristics, and outcomes are limited. These associations can be investigated through experimental or observational studies, with the purpose of enhancing the ability to choose the right treatment for the right patient on the basis of the evidence.

In observational studies, also called practice-based evidence studies,⁶ evidence is based on daily medical practice without establishing a specific experimental environment.⁷ Practice-based evidence studies are more available today as a result of computerization of medical services and formation of large clinical databases.⁸, ⁹, ¹⁰ Within these databases, data on daily practice provided to all patients are captured, enabling the study of relationships between treatments and outcomes. Patient characteristics are accounted for by multivariate statistics, in which many variables can be considered simultaneously and covariates can be identified and controlled to evaluate intervention effects. The practice-based evidence design (initially called clinical practice improvement study design)¹¹, ¹² allows nonstandardized treatment protocols to be included, so data reflect real-world clinical environments. Practice-based evidence studies have few selection criteria for participation, thus enhancing generalizability of findings (external validity).¹¹ For example, patients with comorbidities are commonly excluded from randomized, controlled trials but are included in practice-based evidences, thus extending generalizability of findings to the targeted population.¹³ Practice-based evidence studies require comprehensive, complex databases including detailed patient demographic and health descriptions, characterizations of rehabilitation treatments, and recordings of valid outcomes measures.¹³ If the data set is large enough and if factors influencing distributions of both interventions and outcomes of interest are measured and controlled, treatment effects can be identified from observational data.⁷ A practice-based evidence design is not valid for identification of cause and effect but can identify associations between treatments and outcome variables, which then can be used to develop future before/after studies or randomized, controlled trials.

Of the 114,953 episodes of care provided to 101,310 patients who received PT services during 2007 in outpatient clinics of a medical service in Israel (Maccabi Healthcare Services), 93% were initiated as a result of musculoskeletal disorders, which are known to be associated with high treatment costs.⁸, ⁹, ¹⁴, ¹⁵, ¹⁶, ¹⁷ The 4 most frequently affected body parts were the lumbar spine (21%), knee (14%), cervical spine (15%), and shoulder (13%), which is similar to the incidence in outpatient clinics in the United States.¹⁵ Therefore, we included in our study patients receiving PT treatments in Maccabi for these 4 impairments groups. The current study, which is based on a practice-based evidence design of a large clinical data set, seeks to identify those treatment processes that were related to functional outcomes, controlling for patient characteristics.

Back to Article Outline

Methods 

Design 

We conducted a prospective, observational cohort study. Treatment processes were not altered. Data were collected prospectively from June 2005 to January 2008 from all patients referred to PT who met the inclusion criteria described below.

Subjects 

The study was conducted in the PT service of Maccabi, a public health maintenance organization responsible for providing health care for approximately 1.8 million people in Israel. Patients were Israeli born and new immigrants, including people whose primary language was Hebrew, Russian, Arabic, English, or Spanish. The study population included all adult patients (at least 18 years of age) who were treated with PT. Some had more than 1 episode of care. We analyzed each episode separately; therefore, we refer in the text to episodes of care as patients. This study was approved by the Institutional Review Board for the Protection of Human Subjects of Maccabi Healthcare Services—HMO, Israel.

Patients were selected for analyses if they had a musculoskeletal disorder related to 1 of the 4 impairment categories described above and had FS measurements at intake (first visit) and discharge (last visit). The study population came from a larger group of patients that had similar characteristics, including patients who either completed the treatment episode without completing a functional survey at discharge (incomplete protocol), or dropped out of treatment during the episode of care. Given the possibility for patient selection bias for patients with complete data, a comparison of patient demographic and health characteristics was made between the 2 groups of patients: those with complete and those with incomplete data for characteristics known to be related to functional outcomes, eg, age, days from onset of symptoms (acuity), and FS at intake.¹⁵, ¹⁷, ¹⁸, ¹⁹

Clinics and Therapists 

Outpatient rehabilitation clinics were defined operationally as clinics where patients with neuromusculoskeletal impairments not requiring hospitalization were treated.¹⁶ Patients were managed by 386 physical therapists (aged 36±10y, range 22–63y, 72% women) employed in 54 clinics (2–20 therapists per clinic) located in 50 cities throughout Israel. All therapists held a bachelor's degree in PT, and 10% had also earned a master's degree. The average number of years of clinical experience ± standard deviation was 10±9 years. Data were collected during routine practice where electronic outcomes were integrated into an EHR system.⁸

Patient Characteristics and Treatment Process Data Collection 

Patient characteristics and treatment process data were collected by the Maccabi EHR system, as described elsewhere.⁸ Briefly, the EHR was developed to standardize data collection related to all medical encounters for all Maccabi members, including medical data regarding treatments, patient responses, and clinical assessment. Detailed documentation of medical encounters was accomplished by using standardized data screens with each patient related to a single central medical therapy file that is accessible from any Maccabi facility throughout the country. The system provides online patient characteristics and treatment process data in real time for all clinicians. Data entered into the central data file for each therapy visit include patient personal identification number, age, sex, visit dates and time, treatment duration, clinician and clinic identification, episode identification, type of care (eg, orthopedic, neurologic), type of payer, referring doctor information, diagnosis based on ICD-9 coding,²⁰ physical therapist's classification, treatments used, and information on falls for elderly patients. Data including reason for discharge, goal achievement, attendance and exercise compliance, and capacity to return to previous activities are entered by therapists at discharge from therapy. Data generated from the EHR system are regularly imported to the Maccabi PT database. A thorough data validation process was conducted each month during 2004 and first half of 2005 when the EHR system was implemented, and it was validated by a random electronic file selection process to compare imported and original data. Whenever inconsistencies or inaccuracies were found, EHR programming or import procedures were modified until data were validated. Data entered by clinicians to the EHR were checked for inaccuracies during monthly educational sessions conducted between each clinician and their clinic manager.

Outcomes Data Collection 

Patient-centered FS outcomes data were collected via CAT⁴, ⁵, ²¹, ²², ²³, ²⁴, ²⁵, ²⁶, ²⁷, ²⁸, ²⁹, ³⁰, ³¹ based on IRT.³², ³³ The use of CATs to collect outcomes data in routine clinical work is a relatively new concept,⁴, ⁵, ²³, ²⁴, ²⁹, ³⁰, ³¹ but small-²⁵, ²⁶, ²⁷, ²⁸ and large-scale⁸, ¹⁵ applications have been described. Primary benefits of CAT are efficient data collection (eg, reduced respondent burden) with little loss of measurement precision⁴, ⁵, ¹⁵, ²⁴, ²⁵, ²⁶, ²⁷, ²⁸, ²⁹, ³¹, ³⁴, ³⁵, ³⁶, ³⁷ and ability to integrate data collection with an EHR in an automated manner. The FS outcomes data were integrated with the EHR data into a comprehensive database.⁸ Manual verifications were conducted to ensure patient EHR data and outcomes data were merged accurately.

We used 4 body part–specific CATs³⁸ to estimate patient-specific FS measures: cervical spine,¹⁵ shoulder,⁴ lumbar spine,²⁹ and knee.⁵, ³⁹ FS was defined as the patient's perception of his or her ability to perform functional tasks described in the FS items⁴, ⁵, ¹⁵, ²⁹ and represents the activity dimension of the International Classification of Functioning, Disability and Health.³ Briefly, CATs were developed by calibrating items into scales using the Andrich rating scale IRT model.⁴⁰ Results from previous studies⁴, ⁵, ¹⁵, ²⁹ suggested data fit the IRT model, items represented unidimensional scales with adequate local independence, and thus supported the FS measures were interval scaled. Measures of FS generated by IRT mathematics and applied with CAT methods³⁸ were transformed to a range from 0 to 100 (low [0] to high [100] functioning) on a linear metric.⁴, ⁵, ¹⁵, ²⁹ Although each CAT is transformed to a 0 to 100 FS metric, the 4 body part–specific FS scales were not equated,⁴¹ so all regressions were performed separately for the body part–specific scales. Each CAT was developed to assess FS for patients with a specific type of impairment. Therefore, the item bank for each CAT is different (different items and item calibrations). However, all CATs use a specific starting item, the same item selection routine, estimate FS with the same technique with similar 0 to 100 linear metrics, and apply the same stopping rules. Specifics of each CAT have been published.⁴, ⁵, ¹⁵, ²⁹ Studies describing sensitivity to change,³⁹, ⁴², ⁴³ responsiveness,³⁹, ⁴², ⁴³ and clinical interpretation⁴⁴ of FS measures estimated using the CATs are starting to be published, all of which support the notion that CAT FS measures have sound psychometric properties and can be interpreted in clinically meaningful ways.

We used Patient Inquiry version 5.0 software^45,a to collect functional outcomes data. The software was customized by translating the original English surveys into Hebrew, Russian, and Arabic. Spanish surveys were available from the original Patient Inquiry software. Because outcomes data were collected using CATs, each clinic had at least 1 computer survey stand, including a touch screen connected to the Maccabi network for survey administration and data capture.

Electronic Database Preparation 

We integrated 2 data sets (the functional outcomes database and existing EHR database) to form a PT database.⁸ The PT database was integrated with a third database stored on the Maccabi central computer that contains information on comorbidities based on ICD-9 codes,²⁰ or by attribution to disease-specific registries,⁴⁶, ⁴⁷, ⁴⁸ chronic use of medication (based on prescribed pharmaceutical purchasing database), and appointment scheduling.

Data Analysis 

Descriptive statistics were used to examine frequencies of categorical patient and treatment variables, and average and amount of variation with standard deviation for continuous measures. We compared patients with complete or incomplete outcomes collection protocol, as well as patient characteristics, therapy interventions, and outcomes between patients with lumbar spine, cervical spine, shoulder, and knee impairments. Chi-square tests were used for analysis of categorical data, and t tests or analyses of variance were used for continuous data. The α was set at .05.

We used ordinary least-squares multiple regression to examine associations between variables describing demographic and health characteristics at intake and treatment process variables with each patient's functional outcome at discharge. Use of CAT estimated measures of functional outcomes are not common in regression models of health data, but we thought that the interval nature of the FS measures based on a rating scale IRT model was appropriate for regression techniques that assume linearity of continuous data.⁴⁹ Only variables with frequencies of 2% or more of the sample were allowed to enter regression models. Stepwise R² selection procedure for ordinary least-squares regressions allowed independent variables to enter and leave each model. We created the most parsimonious model for each outcome by allowing only significant variables to remain in the model.⁵⁰ Variables were entered if the significance level of its T value was less than .05 (entry value) and removed if significance was greater than 0.1 (removal value). The importance of each predictor was determined by its T value. Variables entering regression models were checked for multicollinearity; no correlations were greater than 0.6. All analyses were performed with SPSS statistical software, version 15.^b

Data on age, intake FS, number of comorbidities, number of disease specific registries, number of visits per treatment episode, duration in days of treatment episode, and number of procedural interventions throughout the episode of care were allowed to enter the regression models as continuous measures. For categorical data, we created a dummy variable for each category to be used as independent variables in the regression models. Each category was represented as a binary predictor coded 1 if present and 0 if absent. All categories that were allowed to enter as dummy variables into the regression models are described in Table 1, Table 2, Table 3, Table 4. For example, we used dummy variables to represent the attribution of each specific comorbidity, each disease specific registry, and each specific chronic medication use.

Table 1. Patient Characteristics

Variables⁎	Total (N=22,019)	Lumbar Spine (n=7216)	Knee (n=4394)	Cervical Spine (n=5157)	Shoulder (n=5252)	P
Patient Demographic Characteristics
Age†	51±16	50±16	53±16	48±15	56±15	<.001
Age groups (%)						<.001
18–45	35.6	41.4	30.7	45.0	22.5
45–65	42.8	38.2	43.3	41.5	49.8
65–75	15.2	14.2	18.1	10.3	18.9
Over 75	6.4	6.2	7.9	3.2	8.8
Sex (%)						.003
Male	42.0	42.8	41.1	40.2	43.4
Female	58.0	57.2	58.9	59.8	56.6
Payer (%)						<.001
Maccabi	78.0	79.5	90.7	55.5	87.3
MVA	19.2	17.8	6.1	41.4	10.2
Work	2.7	2.7	3.0	3.0	2.3
Others	0.1	0.0	0.2	0.1	0.2
Daily activity (%)						<.001
Office	29.3	30.1	25.6	33.9	26.9
Physical	11.8	12.9	10.7	11.4	11.6
Combined	58.9	57.0	63.7	54.7	61.5
Language (%)						<.001
Hebrew	67.8	66.5	68.7	70.0	66.7
Russian	26.6	28.2	25.4	24.8	27.1
English	3.2	2.8	3.9	2.4	4.1
Arabic	1.4	1.7	0.8	1.9	1.0
Spanish	1.0	0.8	1.2	0.9	1.1
Health and Functional Status Data at Intake
Intake FS	—	48.3±11.8	46.2±13.8	45.3±12.8	51.8±13.4	<.001
Acuity (%)						<.001
Acute	18.2	18.3	13.2	25.4	15.4
Sub acute	34.8	32.4	34.4	35.0	38.1
Chronic	47.0	49.3	52.4	39.6	46.5
Medication use at intake (%)						<.001
Yes	43.2	45.3	36.9	49.8	39.0
No	56.8	54.7	63.1	50.2	61.0
No. of related surgeries (%)						<.001
None	92.5	94.4	83.6	96.3	93.7
1	5.1	3.5	12.3	2.2	3.9
2	1.4	1.2	2.6	0.8	1.3
3	0.6	0.5	1.0	0.2	0.6
4 or more	0.4	0.4	0.5	0.5	0.5
Physical activity (%)						<.001
None	51.3	54.4	50.9	50.6	48.4
1–2/wk	18.6	18.7	16.9	19.9	18.4
3 or more	30.1	27.0	32.2	29.5	33.2

NOTE. Values are mean ± SD unless otherwise noted.

Table 2. Patient Comorbidities or Attribution to Disease Registries

Comorbidity/Medical Registry	Total (N=22,019)	Lumbar Spine (n=7216)	Knee (n=4394)	Cervical Spine (n=5157)	Shoulder (n=5252)	P
Comorbidities⁎	0.84±1.14	0.81±1.14	0.89±1.14	0.69±1.04	0.98±1.23	<.001
Comorbidities groups (%)						<.001
None	53.6	55.6	50.5	58.8	48.3
1	24.0	23.3	24.6	23.5	25.0
2	12.5	11.3	14.3	10.8	14.2
3 or more	9.9	9.8	10.6	6.9	12.5
Registry count⁎	0.66±0.86	0.63±0.85	0.71±0.86	0.53±0.77	0.78±0.91	<.001
Registry count groups (%)						<.001
None	54.8	56.6	52.0	60.9	49.0
1	28.6	27.9	29.2	27.5	30.1
2	12.7	11.6	15.2	9.1	15.5
3 or more	3.9	3.9	3.6	2.5	5.4
Asthma or other lung disease (%)	4.0	3.8	4.1	4.1	4.2	.737
Cardiovascular registry (%)	10.3	10	10.4	8.2	12.8	<.001
CVA/TIA (%)	NA†	NA†	NA†	NA†	2.0	—
Depressive disorder (%)	2.2	2.2	2.2	2.2	2.2	.984
Diabetes registry (%)	10.5	9.9	11.1	7.4	14.1	<.001
Fertility registry (%)	4.1	4.6	2.6	5.9	3.0	<.001
Heart disease (%)	3.9	4.2	3.6	2.8	5.0	<.001
Hyper cholesterol (%)	17.2	16.3	18.8	13.9	20.5	<.001
Hyper tension registry (%)	35.9	34.4	40.5	28.0	41.9	<.001
Hypothyroidism (%)	4.1	4.0	4.6	3.7	4.4	.119
Oncology registry (%)	4.9	4.4	5.8	3.8	5.9	<.001
Osteoporosis (%)	4.2	4.3	4.4	3.0	5.0	<.001
Overweight or obesity (%)	3.8	3.7	4.6	2.9	4.1	<.001
Tobacco use disorder (%)	8.2	8.6	6.5	9.8	7.7	<.001

NOTE. Values are mean ± SD unless otherwise noted.

Abbreviations: CVA, cerebral vascular accident; NA, not applicable; TIA, transient ischemic attack.

Table 3. Chronic Use of Medication

Condition Treated/Medication Group	Total (N=22,019)	Lumbar Spine (n=7216)	Knee (n=4394)	Cervical Spine (n=5157)	Shoulder (n=5252)	P
Anticonvulsant	4.3	4.3	3.5	4.5	4.7	.021
Antidepressant	12.7	12.5	11.9	13.7	12.7	.044
Antineoplastic	NA⁎	NA⁎	NA⁎	NA⁎	2.0	—
Antipsychotic	2.1	2.2	NA⁎	2.4	NA⁎	.195
Antithrombotic	21.0	19.8	23.5	15.7	25.8	<.001
Asthma	8.3	8.1	9.0	7.9	8.5	.186
Cardiovascular	32.2	30.7	36.6	24.7	38.1	<.001
Corticosteroids	3.6	3.7	3.3	3.1	4.0	.044
Diabetes	9.1	8.4	9.7	6.4	12.5	<.001
Hormonereplacementtherapy	6.6	6.5	6.9	6.2	7.1	.258
Hyperlipidemic	32.1	29.3	35.3	25.5	39.7	<.001
Hypothyroidism	7.2	6.8	7.9	6.4	7.8	.004
Migraine	2.7	2.8	2.3	3.6	2.2	<.001
Osteoporosis	7.9	7.8	8.9	5.5	9.4	<.001
Prostate	6.0	5.6	6.6	4.9	7.3	<.001
Sedatives	10.6	10.4	10.8	10.5	10.9	.811

NOTE. Values are percentage of N.

Abbreviation: NA, not applicable.

Table 4. Treatment-Related Variables

Variables⁎	Total (N=22,019)	Lumbar Spine (n=7216)	Knee (n=4394)	Cervical Spine (n=5157)	Shoulder (n=5252)	P
No. of visits	7.6±4.4	7.1±3.6	7.7±4.8	7.5±3.7	8.5±5.5	<.001
Duration (d)	52.8±38.0	50.8±37.1	49.6±35.3	55.4±39.0	55.5±40.1	<.001
Referring doctor specialty (%)						<.001
General practitioner	17.0	17.4	9.8	24.3	15.2
Orthopedic	75.0	75.1	84.4	64.5	77.5
Other	8.0	7.5	5.8	11.2	7.3
Waiting days from referral†	30.7±34.9	31.5±32.3	30.4±34.1	29.9±35.0	30.5±38.7	.105
Waiting days from referral categories (%)						<.001
0 to 7	20.4	18.4	23.2	20.4	21.0
7.1 to 14	12.9	12.7	12.1	13.9	12.9
14.1 to 30	24.4	24.6	22.3	25.7	24.6
Over 30	42.3	44.3	42.4	40.0	41.5
Mean appointment time (min)†	27.3±8.9	28.2±9.1	25.4±8.1	28.8±9.9	26.0±7.6	<.001
Mean appointment time categories (%)						<.001
0 to 19.9	13.0	11.3	17.0	11.0	14.0
20 to 24.9	27.8	24.9	31.3	25.1	31.6
25 to 29.9	24.4	24.4	24.4	23.6	25.2
30 or more	34.8	39.4	27.3	40.3	29.2
Attendance compliance (%)						<.001
Good	87.5	87.1	87.8	86.2	88.9
Moderate	6.5	6.6	5.6	8.2	5.7
Not good	0.7	0.8	0.9	0.7	0.6
Irrelevant‡	5.3	5.5	5.7	4.9	4.8
Home exercise compliance (%)						<.001
Good	66.7	64.4	69.5	64.4	69.8
Moderate	15.4	16.5	13.0	17.1	14.1
Not good	5.6	6.4	5.3	5.8	4.7
Irrelevant§	12.3	12.7	12.2	12.7	11.4
No. of surveys during episode of care (%)						.001
2	78.2	78.8	80.2	77.3	76.8
3 or more	21.8	21.2	19.8	22.7	23.2

NOTE. Values are mean ± SD unless otherwise noted.

Back to Article Outline

Results 

Subjects 

Intake surveys were administered for 57,008 patients. There were 17,485 patients (31%) who completed treatment but who did not complete a functional survey at discharge (incomplete protocol), and 17,504 patients (31%) who dropped out of treatment during the episode of care. That left 22,019 patients with intake and discharge measures, a 39% completion rate.⁸ Descriptive data of these patients are provided in Table 1, Table 2, Table 3, Table 4, Table 5. The definitions of variables and categories described in tables are in appendix 1. Comparison of patient's age, acuity, and FS at intake between patients with complete versus incomplete outcomes collection protocols suggested no difference in age (51.2±15.7y and 51.2±16.2y, respectively, P=.97), slight differences in sex frequency (58.0% and 59.5%, respectively, P=.02), and a statistically significant but clinically identical difference in FS at intake (48.0±13.1 and 48.3±14.1, respectively, P=.02). However, the percentage of patients with a chronic condition at intake (conditions existing for >90 days as perceived by the patient) was higher in the incomplete protocol group than in the complete protocol group (51% and 47%, respectively, P<.001), as was the average number of comorbidities (.88±1.2 and .84±1.1, respectively, P<.001).

Table 5. Physical Therapy Procedural Interventions and Outcome Measure

NOTE. Values are mean ± SD of average number of times each code was used during the episode of care.

Abbreviation: NA, not applicable.

Patient Demographic and Health Data 

Patient demographic and general health characteristics at intake are described in table 1. Patients with lumbar and cervical spine impairments were younger than patients treated for knee and shoulder impairments. Nearly half of all patients were referred to treatment for chronic conditions existing at least 3 months, and 51% were not involved in any weekly physical activity before onset of the condition being treated. Most patients answered the surveys in Hebrew, and 27% of patients answered in Russian. For 19% of patients, the payer was motor vehicle insurance due to a motor vehicle accident. The rate of motor vehicle accident payer among cervical spine impairments was more than double that for the full sample.

Patient comorbid conditions or attribution to disease registries are described in table 2. The highest number of comorbidities was usually found in patients with shoulder impairments, followed by knee impairments. Patients with cervical spine impairments had the largest frequency of tobacco use or women registered in the fertility registry, and patients with knee impairments had the highest rate of overweight or obesity. No significant differences between impairment groups were found for frequencies of hypothyroidism, asthma or other lung diseases, and depressive disorders.

Data on chronic use of medication are described in table 3. Patients treated for shoulder impairments usually used more medications than patients in the other groups. Yet patients with spinal impairments (mainly cervical spine) used more antipsychotic and antimigraine medication than patients with shoulder and knee problems. No substantial differences among impairment groups were found for frequencies of chronic use of sedatives, antidepressants, asthma medications, corticosteroid drugs, and hormone replacement therapy.

Treatment Process 

Variables related to treatment process are described in table 4. Number of visits per treatment episode, attendance compliance, and home exercise compliance were higher in patients with knee and shoulder compared with patients with lumber and cervical spine impairments. In all categories, most referrals were from orthopedic surgeons, with the highest rate for patients with knee impairments. The longest appointment length of time was scheduled for patients with spinal impairments. The lowest rate of patients completing at least 1 survey during the treatment episode in addition to the surveys completed at intake and discharge was in the knee impairment group. There were no differences between impairment groups in waiting days from referral to first visit.

PT procedural interventions, intervention groups as defined in the guide to PT practice,¹ and FS outcome measures are described in table 5. The data in the table represent the average number of times each intervention was used during the overall episode of care. Exercises performed in the clinic were recorded more than twice as much for knee and shoulder than for lumbar and cervical spine impairments. Self-exercise education also was recorded more frequently in patients with knee and shoulder impairments than in spinal impairments, as was the case for the aggregated therapeutic exercise code. Group exercise was most frequent in patients with cervical spine impairments, and joint mobilization was least frequent for patients with knee impairments. Shortwave therapy was applied most frequently for knee impairments, and therapeutic ultrasound was applied most frequently for shoulder impairments.

Associations Between Patient Characteristics, Process of Care, and Outcomes 

Multiple regression analyses for the 4 impairment categories are presented in Table 6A, Table 6B. The table has 2 sections. The first section (6A) includes significant associations between patient demographic and health characteristics at intake and FS at discharge (outcome). The second section (6B) includes significant associations between treatment process variables and functional outcome at discharge, controlling for patient characteristics. The explanatory power of the 4 regression models (1 model per patient group), described by R² values, ranged from 30% to 39%. The R² value did not change when allowing 4 aggregated procedural intervention variables (Electro^APTA, Man_ther^APTA, Pamm^APTA, Therap_ex^APTA)¹ to enter models, compared with allowing specific process variables. Therefore, we selected specific process variables for the 4 regression models because they are more informative than aggregated intervention variables from a clinical perspective. B coefficients indicated the amount of expected change in discharge FS given a 1-unit change in the value of the respective predictor variable, given that all other variables in the model are held constant. T values indicated the importance of each independent variable on predicting discharge FS (dependent variable).

Table 6A. Multiple Regression Analyses by Impairment Categories: Patient Characteristics

Dependent Variable	Lumbar Spine Impairment			Knee Impairment			Cervical Spine Impairment			Shoulder Impairment
Discharge FS	(n=7216; R2=0.35)			(n=4394; R2=0.36)			(n=5157; R2=0.39)			(n=5252; R2=0.30)
Predictors	B⁎	T†	P	B⁎	T†	P	B⁎	T†	P	B⁎	T†	P
Constant	34.0	35.4	<.001	36.0	25.1	<.001	33.4	27.5	<.001	40.6	31.9	<.001
Demographic
Age	−0.1	−5.3	<.001	−0.1	−4.6	<.001	−0.1	−7.4	<.001	−0.1	−8.0	<.001
Daily activity=office	0.6	2.0	.041
Female				−1.5	−3.7	<.001	−1.9	−5.1	<.001	−2.2	−6.1	<.001
Language=Hebrew										0.9	2.5	.013
Language=Russian	1.8	5.7	<.001	2.3	4.7	<.001
Payer=Maccabi	3.2	8.8	<.001	3.1	3.8	<.001	3.7	8.6	<.001	4.0	7.2	<.001
Payer=Work				−3.4	−2.5	.013
Health
Acuity=acute	3.0	7.4	<.001	2.0	3.0	.002	1.5	3.1	.002	2.2	4.2	<.001
Acuity=chronic	−2.8	−9.2	<.001	−3.4	−8.1	<.001	−3.5	−8.3	<.001	−2.2	−5.9	<.001
Intake functional status	0.5	45.5	<.001	0.5	33.9	<.001	0.6	40.8	<.001	0.5	33.4	<.001
Medication use at intake				−1.3	−3.0	.003	−1.3	−3.5	<.001
No. of related surgeries=1 or more	−3.8	−6.4	<.001
Physical activity=1–2/week										1.1	2.5	.012
Physical activity=none	−1.4	−5.2	<.001	−1.6	−4.0	<.001	−1.1	−3.1	.002
Comorbidities
No. of comorbidities	−0.5	−3.5	.001
No. of medical condition registries				−0.7	−2.7	.006
Cardiovascular registry										−1.1	−2.0	.042
Tobacco use disorder							−1.4	−2.4	.015	−1.4	−2.1	.035
Chronic medication use
Antidepressant	−1.5	−3.7	<.001	−1.5	−2.4	.016	−1.2	−2.4	.018	−1.8	−3.5	<.001
Asthma	−1.5	−3.0	.003
Migraine							−3.0	−3.2	.001
Osteoporosis	−1.3	−2.4	.015	−1.5	−2.1	.035

Table 6B. Multiple Regression Analyses by Impairment Categories: Treatment Process

Dependent Variable	Lumbar Spine Impairment			Knee Impairment			Cervical Spine Impairment			Shoulder Impairment
Discharge FS	(N=7216; R2=0.35)			(N=4394; R2=0.36)			(N=5157; R2=0.39)			(N=5252; R2=0.30)
Predictors	B⁎	T†	P	B⁎	T†	P	B⁎	T†	P	B⁎	T†	P
Constant	34.0	35.4	<.001	36.0	25.1	<.001	33.4	27.5	<.001	40.6	31.9	<.001
Process related
Attendance compliance=good	2.1	4.7	<.001	1.7	2.5	.013	2.9	5.0	<.001	2.5	4.1	<.001
Referring doctor's specialty=general practitioner	1.3	3.3	.001
Referring doctor's specialty=orthopedic							−1.1	−3.1	.002
Duration	0.0	−2.0	.049
Home exercise compliance=good	3.0	8.9	<.001	3.4	6.6	<.001	3.1	7.2	<.001	4.2	9.6	<.001
Home exercise compliance=not good	−2.4	−4.1	<.001	−2.1	−2.3	.021	−2.6	−3.2	.001	−2.0	−2.4	.017
Mean appointment time=30min or more				1.3	2.7	.008
No. of surveys=3 or more	1.0	2.6	.008				1.8	3.9	<.001	1.2	2.8	.005
No. of visits	−0.2	−4.0	<.001				−0.6	−8.3	<.001
Waiting days from referral=0–7				1.8	3.5	<.001
Waiting days from referral=7.1–14	−1.3	−3.2	.002
Waiting days from referral=14.1 to 30	−0.7	−2.0	.041
Interventions
Assupron_ex		NA‡			NA‡			NA‡		0.4	2.7	.006
Cold_pack		NA‡		−0.5	−2.4	.017		NA‡			NA‡
Consult	−0.8	−2.3	.019
Electro_pain	−0.2	−3.7	<.001	−0.2	−3.2	.001				−0.2	−3.3	.001
Group_ex				−0.2	−2.7	.006	0.2	2.6	.008	−0.2	−3.3	.001
Joint_mob				0.1	2.0	.045	0.2	3.4	.001
Neural_mob					NA‡					−0.5	−2.5	.014
Passupro_mob										0.2	2.3	.024
Proprio_ex		NA‡		0.4	2.2	.025		NA‡			NA‡
Reassess	−0.3	−2.7	.007
Shortwave				−0.3	−4.3	<.001				−0.3	−4.5	<.001
Stabil_ex	0.6	2.9	.004		NA‡
Stretch_ex										0.4	2.4	.014
Ultrasound										−0.2	−3.5	<.001

For all models, patient variables explained most of the variation in outcomes (ie, in discharge functional score), with intake FS score being positively associated with outcomes and the strongest predictor. For clarity, we note that variables associated with poorer or better outcomes refer to poorer or better than their reference group—for example, acute compared with not acute, or women compared with not women. Older age, chronic conditions, and greater use of antidepressant medications were associated with poorer outcomes, whereas acute conditions and regular Maccabi payer (ie, regular health insurance as opposed to workers' compensation or car insurance coverage) were associated with better outcomes. In all models except for the lumbar spine impairment group, women were associated with poorer outcomes, and in all models except for patients with shoulder impairments, no physical activity before onset of the condition was associated with poorer outcomes.

Across all models, good home exercise compliance (reference is moderate or poor compliance) was the strongest process variable associated with better outcomes and was 1 of 3 second most predictive variables overall, along with symptom acuity and age. In addition, poor home exercise compliance was associated with poorer outcomes, and good visit attendance compliance was associated with better outcomes. In all models except for knee impairments, completion of at least 1 functional survey during the episode of care in addition to intake and discharge surveys (labeled 3 surveys or more with reference being 2 surveys only at intake and discharge) was associated with better outcomes. In models for spinal impairments (lumbar and cervical spine), a high number of visits per episode of care was associated with poorer outcomes. Patients with lumbar or cervical spine impairments who were referred by general practitioners were associated with better outcomes compared with referral by an orthopedic surgeon. For lumbar spine and knee impairments, longer waiting days from referral to admission were associated with poorer outcomes.

Greater use of electrotherapeutic modalities for pain management was associated with poorer outcomes in all models except for a cervical spine impairments model. In models for patients with knee and cervical spine impairments, manual joint mobilization was associated with better outcomes. Additionally, in all models, higher frequencies of personal active exercise interventions were associated with better outcomes. The only active intervention that had mixed associations with outcomes was group exercise, with negative associations for knee and shoulder impairments and positive associations for cervical spine impairments.

Back to Article Outline

Discussion 

Our purpose was to identify associations between FS outcomes during PT, patient characteristics, and treatment processes for patients with lumbar spine, knee, cervical spine, or shoulder musculoskeletal impairments. Many associations were identified. For brevity, we discuss only some of these associations, which characterize our findings and emphasize benefits of practice-based evidence studies to outpatient therapy clinical practice.

Patient Demographic and General Health 

Demographic variables and health characteristics are considered outcome confounders.⁷, ¹¹, ⁵¹, ⁵², ⁵³, ⁵⁴, ⁵⁵ Intake FS, age, acuity level, payer, the existence of various comorbidities, chronic use of antidepressant medication, and exercise history were associated with functional outcomes across all 4 regression models. The 4 strongest predictors of outcomes were intake FS, age, acuity levels, and payer. These have been reported by others,¹⁵, ³⁹, ⁴², ⁵⁴ a finding that emphasizes the importance of controlling for these factors when studying associations between treatment processes and outcomes in PT.

The presence of comorbid conditions was associated with poorer outcomes across all 4 regression models (see Table 6A, Table 6B). This supports use of comorbidity indices known to be predictive of functional outcomes.³⁹, ⁴², ⁴³, ⁵⁶ However, although the highest rate of comorbidities usually found in patients with shoulder impairments, this group had the weakest associations between comorbidities and outcomes. This result has been reported previously,⁵⁷ suggesting that comorbidities may not have important negative effects on shoulder functional outcomes.

Additionally, overweight, with highest frequency in patients with knee impairments, was not associated with functional outcomes in any model. Evidence exists supporting a BMI effect on incidence of knee osteoarthritis, particularly in women, although several studies failed to detect an effect of BMI on knee osteoarthritis progression.⁵⁸ Future analyses should perhaps identify smaller and more homogeneous subsets of patients with knee impairments that might show associations that are masked when all knee impairments are analyzed together. For example, if women with knee impairments were analyzed separately from men, or patients with knee osteoarthritis were analyzed separately from patients with knee internal derangements, associations between BMI and functional outcomes might emerge. Furthermore, we noted that rates of overweight and obesity in our study were much lower than rates reported in a recent survey in Israel.⁵⁹ This suggests possible underreporting of BMI measures in Maccabi records, which might have influenced our results.

Chronic use of medication had a stronger association with poorer outcomes than comorbidities. This stronger association could result from the fact that the medication data being extracted from a database of purchased prescriptions are not affected by the comprehensiveness of doctors' registration in the EHR. For example, for the total sample, only 2% of patients were labeled as experiencing a depressive disorder, but 13% purchased antidepression medication, which was the most predictive chronic medication use of poorer outcomes in all groups. This finding is supported by previous publications⁶⁰, ⁶¹ and emphasizes the need to control for use of medications as well as to identify patients with depression when studying associations between treatments and outcomes in PT. From a clinical perspective, it is important to find ways to improve outcomes of patients with depression when seeking PT treatment. Previous studies have suggested integrating cognitive behavioral treatments to improve functional and mental status, including depression in patients with low back pain.⁶² Additional studies are required to study the contribution of these approaches.

Lack of physical activity (defined as at least once a week for a duration of 20min or more) before the onset of the condition being treated was associated with poorer outcomes in lumbar spine, knee, and cervical spine impairments. Moderate physical activity (once to twice a week) was associated with better outcomes in shoulder impairments. Exercise history has been previously reported to be associated with functional outcomes,³⁹, ⁴², ⁴³, ⁵⁴, ⁶³ which suggests that health promotion measures targeted to improve participation in routine physical activity could benefit patients attending PT.

Sex was associated with outcomes in 3 of the 4 regression models. Women were associated with poorer outcomes for patients with knee, cervical spine and shoulder impairments. English-speaking men and women in the United States responded to FS items in similar ways,⁵, ²⁹, ⁶⁴ suggesting that sex plays a role in how a person responds to PT. Previous evidence indicates that women are more likely to respond positively to multidisciplinary⁶⁵ and cognitive-behavioral treatments for chronic pain.⁶⁶, ⁶⁷ A recent study on sex differences for work-disabled people with chronic musculoskeletal pain found sex differences in the influence of sociodemographic and psychosocial factors on rehabilitation outcomes of functional health status.⁶⁸ Future studies are needed to identify ways to adapt treatment procedures to sex. Subset analyses by sex could help discover whether different treatment plans should be designed for women and men.

Process-Related and Intervention Variables 

Compliance with self-exercise programs was one of the strongest predictors overall and the strongest predictor among process variables (see table 6B). Better outcomes were achieved when patients were more compliant with their exercise program. This result has important implications for clinicians. Ability to improve patient compliance is probably more of an educational skill than a clinical skill. One could perceive this as a marketing skill. Physical therapists need to know how to educate and persuade patients that what they are “selling” actually works. These persuasive skills are especially important because self-exercise programs are time-consuming and demand self-discipline and persistence. It is only when the patient is convinced that it is worth the time and effort that good compliance with exercise programs can be achieved. Another patient compliance issue is compliance with attending therapy. Again, better compliance was associated with better outcomes. These results suggest clinicians should strengthen their educating skills. Although similar recommendations were published recently,⁶⁹ the best ways to achieve this goal are yet to be determined.

Interestingly, patients with spinal impairments had a lower rate of good home exercise compliance than patients with peripheral impairments. The same trend was observed for good attendance compliance. These differences may result from therapists' clinical skills, which seem to be more persuasive when educating their patients to do exercises targeted to peripheral joint structures. Also, exercises for peripheral joints may be easier to understand and perform, and therefore they are related to higher compliance than exercises addressing spinal structures. Support for this claim is seen in the frequencies of therapeutic exercise use by impairment categories (see table 5, Therap_ex^APTA); therapeutic exercises were used more frequently per treatment episodes for peripheral compared with spinal impairments. We conjecture that physical therapists consider exercises for peripheral joints to be more effective and easier to perform than exercises for spinal joints, and therefore they are prescribed more often. Physical therapists may need more education regarding evidence-based knowledge on effectiveness of exercises for spinal impairments. An enhanced postgraduate educational program on methods that are supported scientifically can help therapists to better adjust exercises to their patients, be more persuasive, and improve patient compliance. Our data strongly suggest that if this happens, outcomes could be improved.

For patients to comply with exercise programs, they first must be prescribed by their therapists. Active exercise variables that entered the regression models were associated with better outcomes. An exception was group exercise, which was associated with poorer outcomes for knee and shoulder impairments. For lumbar spine impairments, group exercise was not associated with outcomes, a result reported previously.⁷⁰ Only patients with cervical spine impairments, who had the highest frequency for use of group exercises, seemed to benefit from them. This result calls for further investigation of the specific contents of group exercise programs and their appropriateness for different musculoskeletal conditions.

Joint mobilization was positively related to outcomes in cervical spine impairments and had borderline significance for knee impairments. Thus, in spite of little support in the literature for passive interventions,⁷¹, ⁷² the use of manual therapy (a passive intervention) is important. Further analyses are needed to identify more specific groups of patients that do or do not benefit from manual therapy techniques, as reported previously for spinal manipulation techniques.⁷³, ⁷⁴, ⁷⁵ However, electrotherapy, including shortwave therapy and therapeutic ultrasound, were always associated with poorer outcomes. Similar results have been reported in the past.⁷¹, ⁷⁶, ⁷⁷, ⁷⁸, ⁷⁹, ⁸⁰ Nonetheless, there is some evidence to support that with knee osteoarthritis, transcutaneous electrical nerve stimulation was more effective than placebo⁸¹ and that combined exercise therapy with physical agents increases the effectiveness of exercise.⁸² This suggests the need for further investigation to detect groups of patients that might benefit from these interventions, and for reevaluation of electrotherapeutic modalities and therapeutic ultrasound, which in many cases may have little benefit in promoting functional outcomes.

Interestingly, in all but knee impairments (where 3 or more surveys were least frequent), patients who took only the intake and discharge surveys had poorer outcomes than those who took 3 or more surveys. Three or more surveys indicate that a functional survey was completed at least once during the episode of care. This could have encouraged therapists to adjust treatment strategies when realizing that patients were not improving as expected. This suggests that therapists ought to initiate a wider use of functional measures during therapy in order for the process of care to be continuously based on outcomes.

For treated patients with spinal impairments, poorer outcomes were associated with more visits per treatment episode. This finding needs careful interpretation. Because of the public nature of the Maccabi system, therapists' reimbursement is not dependent on number of visits. Therefore, the main reason to prolong therapy is when a patient does not get better. Our data suggest that when outcomes improve during treatment of spinal impairments, the episode of care is usually shorter. When outcomes do not improve, the tendency of therapists might be to add visits, hoping that this will help improve outcomes, but simply doing more might not be beneficial. Further testing is needed to study whether this explains the association found between number of visits and outcomes. Because overall costs of PT service are closely related to number of visits per treatment episode, we suggest that a reasonable way to improve the cost-effectiveness ratio in PT is to focus on improving outcomes rather than on lowering costs. When outcomes improve, it is likely that cost will eventually decrease, a view that has been acknowledged previously for medical care in general⁸³ and is supported by pay-for-performance simulations in outpatient therapy.¹⁵

Study Limitations 

Patient selection bias is a possible limitation attributed to observational studies. In our data, all patients referred to therapy were assessed, but not all patients who had intake data also had discharge data, which produces the potential for patient selection bias. We chose to include patients treated in all clinics participating in the outcomes collection system, regardless of how long the clinic was using the system. Consequently, inclusion of clinics that just recently implemented the outcomes measurement system resulted in a relatively low overall completion rate (39%).⁸ With large data sets, which minimizes the possibility of a consistent selection bias,¹⁵ the only potential for selection bias related to low completion rates would be if therapists would consistently collect both intake and discharge functional surveys from specific groups of patients thought to have achieved better outcomes. If that were to occur, higher frequencies for characteristics known to be related to better outcomes would be found in the complete protocol group. Thus, we would expect patients from the complete protocol group to be younger, more acutely injured, and with lower intake scores compared with the incomplete protocol group, because all of these variables were found to be strongly associated with better FS outcomes.³⁵, ⁵³, ⁵⁴, ⁸⁴, ⁸⁵, ⁸⁶, ⁸⁷ Given our findings, we could also expect these patients to have a higher rate of men and in some cases more comorbidities. However, in our sample, both groups of patients (with or without functional measures at discharge) were similar in age, sex, and intake FS measures. Still, there was a difference between the 2 groups regarding symptom acuity, with a larger percentage of patients with chronic symptoms and a slightly higher average number of comorbidities in the incomplete protocol group compared with the complete protocol group. Although previous findings have been equivocal related to differences between complete and incomplete data related to symptom acuity,³⁹, ⁴² our current finding suggests possible patient selection bias. Therefore, we recommend that all clinics achieve high completion rates in order to minimize the potential for patient selection bias.

Possible missing or underreporting of patient characteristics could limit our ability to control for baseline patient characteristics. Therefore, we should encourage addition of other patient descriptors such as educational and socioeconomic levels, known to be related to outcomes in general health care⁸⁸, ⁸⁹ and in musculoskeletal rehabilitation.⁹⁰, ⁹¹ Furthermore, we need improvements in registrations of patients characteristics thought to be underreported—for example, BMI values. We conducted additional cross-checks for possible underreporting of comorbidities and identified possible underreporting of tobacco use disorders. One of the ways that the Maccabi health care system works to improve the comprehensiveness of patient specific comorbidities data is the formation of condition-specific registries that are based on several inclusion criteria from several different data sources.⁴⁶, ⁴⁷, ⁴⁸ This not only can improve the ability to control for these characteristics, but also can help identify health promotion and prevention activities directed to decrease risk factors that inhibit the achievement of high physical function in PT.

Another possible limitation results from unstudied reliability and validity of treatment process coding among therapists. Reliability verification of data registration coding has been acknowledged as an important step of practice-based evidence studies.⁶, ¹¹, ¹², ¹³ We assume that improved reliability and validity of the treatment coding process will enable us to discover additional relationships between treatment process and outcome, and will raise the explanatory power (R²) by treatment processes in regression models. Therefore, the Maccabi PT management team has initiated an educational program for all physical therapists to improve representation of treatment processes within the electronic database.

Finally, the clinical interpretability of the regression models described in our study is limited as a result of the heterogeneity of patients in the 4 subsets we analyzed. For example, patients with lumbar spine or cervical spine symptoms are known to include many different subgroups of patients that can be reliably and validly classified.⁷³, ⁹², ⁹³, ⁹⁴, ⁹⁵, ⁹⁶, ⁹⁷, ⁹⁸, ⁹⁹, ¹⁰⁰, ¹⁰¹, ¹⁰², ¹⁰³, ¹⁰⁴, ¹⁰⁵, ¹⁰⁶, ¹⁰⁷ When analyzing all subgroups together, some important clinical associations might be masked. To overcome this limitation, therapists must learn to classify their patients in a reliable and valid way and record these classifications accordingly. Only then will we be able to identify clinically valid subgroups of patients that, when analyzed separately, could reveal additional and more specific relationships between process of care and outcomes, helping therapists to improve the outcomes of care.

Back to Article Outline

Conclusions 

This study offers an introductory examination of a large integrated database to investigate associations between patient characteristics, treatments, and outcomes in outpatient PT practice. We found that better functional outcomes were associated with the following: high patient compliance with self-exercises; use of active exercises; number of completed functional assessments during therapy; high participation in routine physical activity; and use of manual therapy. Further studies are needed to identify ways to improve outcomes for women. The wide use of electrotherapeutic modalities in all impairment categories, and therapeutic ultrasound for shoulder impairments, was not supported and therefore should be reevaluated. Finally, we conclude that practice-based evidence studies work well with integrated outcomes and EHR databases to help clinicians identify ways to improve care.

Suppliers

Back to Article Outline

Acknowledgments 

We thank Rafi Rolnik, BSc, Department of Medical Information System, Avi Hosit and his team, Information Technology Department, and Shlomo Oz, David Amital, MSc, Eyal Kedem, MBA, Orly Shpigel, BA Econ, and Hillel Alapi, BBSc, System Development Department, who all played key roles in the data collection process from the various computerized registries of Maccabi Healthcare Services. We thank the PT research and development coordinators and PT district directors of Maccabi who managed the process of implementing a continuous data collection strategy. A special acknowledgment is dedicated to the physical therapists and their clinic managers of Maccabi who participated in the ongoing data collection process.

Back to Article Outline

Appendix 1 

Appendix 1. VARIABLE DESCRIPTIONS

Back to Article Outline

References 

1. Guide to physical therapist practice. 2nd ed. American Physical Therapy Association. Phys Ther 2001;81:9-746.
   • View In Article
2. World Confederation for Physical Therapy. Declarations of principle and position statements. 1999. http://wcpt.org/node/29602Accessed July 1, 2009
   • View In Article
3. International Classification of Functioning, Disability and Health (ICF). Geneva: World Health Organization; 2001;
   • View In Article
4. Hart DL, Cook KF, Mioduski JE, Teal CR, Crane PK. Simulated computerized adaptive test for patients with shoulder impairments was efficient and produced valid measures of function. J Clin Epidemiol. 2006;59(3):290–298
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (152 KB)
   • CrossRef
5. Hart DL, Mioduski JE, Stratford PW. Simulated computerized adaptive tests for measuring functional status were efficient with good discriminant validity in patients with hip, knee, or foot/ankle impairments. J Clin Epidemiol. 2005;58(6):629–638
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (227 KB)
   • CrossRef
6. Horn SD, Gassaway J. Practice-based evidence study design for comparative effectiveness research. Med Care. 2007;45(10 Suppl 2):S50–S57
   • View In Article
   • CrossRef
7. Horn SD, DeJong G, Ryser DK, Veazie PJ, Teraoka J. Another look at observational studies in rehabilitation research: going beyond the holy grail of the randomized controlled trial. Arch Phys Med Rehabil. 2005;86(12 Suppl 2):S8–S15
   • View In Article
   • MEDLINE
8. Deutscher D, Hart DL, Dickstein R, Horn SD, Gutvirtz M. Implementing an integrated electronic outcomes and electronic health record process to create a foundation for clinical practice improvement. Phys Ther. 2008;88:270–285
   • View In Article
   • CrossRef
9. Swinkels IC, Ende CH, de Bakker D, et al. Clinical databases in physical therapy. Physiother Theory Pract. 2007;23:153–167
   • View In Article
   • MEDLINE
   • CrossRef
10. Swinkels IC, Hart DL, Deutscher D, et al. Comparing patient characteristics and treatment processes in patients receiving physical therapy in the United States, Israel and the Netherlands (Cross sectional analyses of data from three clinical databases). BMC Health Serv Res. 2008;8:163
    • View In Article
    • CrossRef
11. Horn SD. Clinical practice improvement methodology: implementation and evaluation. New York: Faulkner & Gray; 1997;
    • View In Article
12. Horn SD, Hopkins DSP. Clinical practice improvement: a new technology for developing cost-effective quality care. New York: Faulkner & Gray; 1994;
    • View In Article
13. Gassaway J, Horn SD, DeJong G, Smout RJ, Clark C, James R. Applying the clinical practice improvement approach to stroke rehabilitation: methods used and baseline results. Arch Phys Med Rehabil. 2005;86(12 Suppl 2):S16–S33
    • View In Article
    • MEDLINE
14. Deutscher D. Quality assurance in physical therapy [thesis]. Tel-Aviv: Tel-Aviv Univ; 2002;
    • View In Article
15. Hart DL, Connolly JB. Pay-for-performance for physical therapy and occupational therapy: Medicare Part B Services (Final report). Grant 18-P-93066/9-01 Health & Human Services/Centers for Medicare & Medicaid Services; 2006;http://www.cms.hhs.gov/TherapyServices/downloads/P4PFinalReport06-01-06.pdfAccessed July 1, 2009
    • View In Article
16. Hart DL, Wright BD. Development of an index of physical functional health status in rehabilitation. Arch Phys Med Rehabil. 2002;83:655–665
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (112 KB)
    • CrossRef
17. Khoudri I, Ali Zeggwagh A, Abidi K, Madani N, Abouqal R. Measurement properties of the short form 36 and health-related quality of life after intensive care in Morocco. Acta Anaesthesiol Scand. 2007;51:189–197
    • View In Article
    • MEDLINE
    • CrossRef
18. Badke MB, Boissonnault WG. Changes in disability following physical therapy intervention for patients with low back pain: dependence on symptom duration. Arch Phys Med Rehabil. 2006;87:749–756
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (872 KB)
    • CrossRef
19. Boissonnault WG, Badke MB. Influence of acuity on physical therapy outcomes for patients with cervical disorders. Arch Phys Med Rehabil. 2008;89:81–86
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (170 KB)
    • CrossRef
20. In:  Hart AC,  Schmidt KM,  Aaron WS editor. ICD-9-CM code book. Reston: St Anthony's; 1999;
    • View In Article
21. Ader D. Developing the Patient-Reported Outcomes Measurement Information System (PROMIS). Med Care. 2007;45(5 Suppl 1):S1–S2
    • View In Article
    • MEDLINE
    • CrossRef
22. Cella D, Yount S, Rothrock N, et al. The Patient-Reported Outcomes Measurement Information System (PROMIS): progress of an NIH Roadmap cooperative group during its first two years. Med Care. 2007;45(5 Suppl 1):S3–S11
    • View In Article
    • MEDLINE
    • CrossRef
23. Fliege H, Becker J, Walter OB, Bjorner JB, Klapp BF, Rose M. Development of a computer-adaptive test for depression (D-CAT). Qual Life Res. 2005;14:2277–2291
    • View In Article
    • MEDLINE
    • CrossRef
24. Haley SM, Coster WJ, Andres PL, Kosinski M, Ni P. Score comparability of short forms and computerized adaptive testing: simulation study with the activity measure for post-acute care. Arch Phys Med Rehabil. 2004;85:661–666
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (136 KB)
    • CrossRef
25. Haley SM, Fragala-Pinkham M, Ni P. Sensitivity of a computer adaptive assessment for measuring functional mobility changes in children enrolled in a community fitness programme. Clin Rehabil. 2006;20:616–622
    • View In Article
    • MEDLINE
    • CrossRef
26. Haley SM, Ni P, Fragala-Pinkham MA, Skrinar AM, Corzo D. A computer adaptive testing approach for assessing physical functioning in children and adolescents. Dev Med Child Neurol. 2005;47:113–120
    • View In Article
    • MEDLINE
    • CrossRef
27. Haley SM, Raczek AE, Coster WJ, Dumas HM, Fragala-Pinkham MA. Assessing mobility in children using a computer adaptive testing version of the pediatric evaluation of disability inventory. Arch Phys Med Rehabil. 2005;86:932–939
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (153 KB)
    • CrossRef
28. Haley SM, Siebens H, Coster WJ, et al. Computerized adaptive testing for follow-up after discharge from inpatient rehabilitation: I (Activity outcomes). Arch Phys Med Rehabil. 2006;87:1033–1042
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (233 KB)
    • CrossRef
29. Hart DL, Mioduski JE, Werneke MW, Stratford PW. Simulated computerized adaptive test for patients with lumbar spine impairments was efficient and produced valid measures of function. J Clin Epidemiol. 2006;59(9):947–956
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (305 KB)
    • CrossRef
30. Ware JE. Conceptualization and measurement of health-related quality of life: comments on an evolving field. Arch Phys Med Rehabil. 2003;84(4 Suppl 2):S43–S51
    • View In Article
    • Abstract
    • Full-Text PDF (209 KB)
    • CrossRef
31. Ware JE, Kosinski M, Bjorner JB, et al. Applications of computerized adaptive testing (CAT) to the assessment of headache impact. Qual Life Res. 2003;12:935–952
    • View In Article
    • MEDLINE
    • CrossRef
32. Hambleton RK, Swaminathan H. Item response theory: principles and applications. Boston: Kluwar; 1985;
    • View In Article
33. Hambleton RK, Swaminathan H, Rogers HJ. Fundamentals of item response theory. Newbury Park: Sage Publications; 1991;
    • View In Article
34. Jette AM, Haley SM. Contemporary measurement techniques for rehabilitation outcomes assessment. J Rehabil Med. 2005;37:339–345
    • View In Article
    • MEDLINE
    • CrossRef
35. Jette AM, Haley SM, Tao W, et al. Prospective evaluation of the AM-PAC-CAT in outpatient rehabilitation settings. Phys Ther. 2007;87:385–398
    • View In Article
    • MEDLINE
    • CrossRef
36. In:  Sands WA,  Waters BK,  McBride JR editor. Computerized adaptive testing: from inquiry to operation. Washington: American Psychological Association; 1997;
    • View In Article
37. Wainer H. Introduction and history. In:  Wainer H editors. Computerized adaptive testing: a primer. 2nd ed.. Mahwah: Lawrence Erlbaum; 2000;p. 1–21
    • View In Article
38. Mills CN. Computer-based testing: building the foundation for future assessments. Mahwah: Lawrence Erlbaum; 2002;
    • View In Article
39. Hart DL, Wang YC, Stratford PW, Mioduski JE. Computerized adaptive test for patients with knee impairments produced valid and responsive measures of function. J Clin Epidemiol. 2008;61(11):1113–1124
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (506 KB)
    • CrossRef
40. Andrich D. A rating formulation for ordered response categories. Psychometrika. 1978;43:561–573
    • View In Article
    • CrossRef
41. In:  Lord FM editors. Applications of item response theory to practical testing problems. Hillsdale: Lawrence Erlbaum; 1980;
    • View In Article
42. Hart DL, Wang YC, Stratford PW, Mioduski JE. A computerized adaptive test for patients with hip impairments produced valid and responsive measures of function. Arch Phys Med Rehabil. 2008;89:2129–2139
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (323 KB)
    • CrossRef
43. Hart DL, Wang YC, Stratford PW, Mioduski JE. Computerized adaptive test for patients with foot or ankle impairments produced valid and responsive measures of function. Qual Life Res. 2008;17:1081–1091
    • View In Article
    • CrossRef
44. Wang YC, Hart DL, Stratford PW, Mioduski JE. Clinical interpretation of computerized adaptive test–generated outcome measures in patients with knee impairments. Arch Phys Med Rehabil. 2009;90:1340–1348
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (720 KB)
    • CrossRef
45. Patient Inquiry. Version 5. Knoxville: Focus on Therapeutic Outcomes. 2008. http://www.fotoinc.com/demoinstructions.htmAccessed July 1, 2009
    • View In Article
46. Chodick G, Heymann AD, Shalev V, Kookia E. The epidemiology of diabetes in a large Israeli HMO. Eur J Epidemiol. 2003;18:1143–1146
    • View In Article
    • MEDLINE
    • CrossRef
47. Heymann AD, Chodick G, Halkin H, Kokia E, Shalev V. [Description of a diabetes disease register extracted from a central database]. Harefuah. 2007;146:15–1779
    • View In Article
    • MEDLINE
48. Shapiro J, Cohen AD, David M, et al. The association between psoriasis, diabetes mellitus, and atherosclerosis in Israel: a case-control study. J Am Acad Dermatol. 2007;56:629–634
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (220 KB)
    • CrossRef
49. Wright BD, Linacre JM. Observations are always ordinal; measurements, however, must be interval. Arch Phys Med Rehabil. 1989;70:857–860
    • View In Article
    • MEDLINE
50. Horn SD, DeJong G, Smout RJ, Gassaway J, James R, Conroy B. Stroke rehabilitation patients, practice, and outcomes: is earlier and more aggressive therapy better?. Arch Phys Med Rehabil. 2005;86(12 Suppl 2):S101–S114
    • View In Article
    • MEDLINE
51. Iezzoni LI. Using risk-adjusted outcomes to assess clinical practice: an overview of issues pertaining to risk adjustment. Ann Thorac Surg. 1994;58:1822–1826
    • View In Article
    • MEDLINE
    • CrossRef
52. Iezzoni LI, Shwartz M, Ash AS, Mackiernan Y, Hotchkin EK. Risk adjustment methods can affect perceptions of outcomes. Am J Med Qual. 1994;9:43–48
    • View In Article
    • MEDLINE
    • CrossRef
53. 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
    • View In Article
    • MEDLINE
54. Resnik L, Hart DL. Using clinical outcomes to identify expert physical therapists. Phys Ther. 2003;83:990–1002
    • View In Article
    • MEDLINE
55. Resnik L, Liu D, Mor V, Hart DL. Predictors of physical therapy clinic performance in the treatment of patients with low back pain syndromes. Phys Ther. 2008;88:989–1004
    • View In Article
    • CrossRef
56. Groll DL, To T, Bombardier C, Wright JG. The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol. 2005;58:595–602
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (196 KB)
    • CrossRef
57. Boissonnault WG, Badke MB, Wooden MJ, Ekedahl S, Fly K. Patient outcome following rehabilitation for rotator cuff repair surgery: the impact of selected medical comorbidities. J Orthop Sports Phys Ther. 2007;37:312–319
    • View In Article
58. Reijman M, Pols HA, Bergink AP, et al. Body mass index associated with onset and progression of osteoarthritis of the knee but not of the hip: the Rotterdam Study. Ann Rheum Dis. 2007;66:158–162
    • View In Article
    • MEDLINE
    • CrossRef
59. Keinan-Boker L, Noyman N, Chinich A, Green MS, Nitzan-Kaluski D. Overweight and obesity prevalence in Israel: findings of the first national health and nutrition survey (MABAT). Isr Med Assoc J. 2005;7:219–223
    • View In Article
    • MEDLINE
60. Kim JM, Stewart R, Glozier N, et al. Physical health, depression and cognitive function as correlates of disability in an older Korean population. Int J Geriatr Psychiatry. 2005;20:160–167
    • View In Article
    • MEDLINE
    • CrossRef
61. Russo A, Cesari M, Onder G, et al. Depression and physical function: results from the aging and longevity study in the Sirente geographic area (ilSIRENTE Study). J Geriatr Psychiatry Neurol. 2007;20:131–137
    • View In Article
    • CrossRef
62. Guzman J, Esmail R, Karjalainen K, Malmivaara A, Irvin E, Bombardier C. WITHDRAWN: multidisciplinary bio-psycho-social rehabilitation for chronic low-back pain. Cochrane Database Syst Rev. 2006;18:CD000963
    • View In Article
63. Hart DL, Dobrzykowski EA. Impact of exercise history on health status outcomes in patients with musculoskeletal impairments. Orthop Phys Ther Clin North Am. 2000;9:1–16
    • View In Article
64. Crane PK, Hart DL, Gibbons LE, Cook KF. A 37-item shoulder functional status item pool had negligible differential item functioning. J Clin Epidemiol. 2006;59:478–484
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (134 KB)
    • CrossRef
65. Jensen IB, Bodin L. Multimodal cognitive-behavioural treatment for workers with chronic spinal pain: a matched cohort study with an 18-month follow-up. Pain. 1998;76:35–44
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (148 KB)
    • CrossRef
66. Jensen IB, Bergstrom G, Ljungquist T, Bodin L, Nygren AL. A randomized controlled component analysis of a behavioral medicine rehabilitation program for chronic spinal pain: are the effects dependent on gender?. Pain. 2001;91:65–78
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (175 KB)
    • CrossRef
67. Jensen IB, Nygren A, Lundin A. Cognitive-behavioural treatment for workers with chronic spinal pain: a matched and controlled cohort study in Sweden. Occup Environ Med. 1994;51:145–151
    • View In Article
    • MEDLINE
    • CrossRef
68. Lillefjell M. Gender differences in psychosocial influence and rehabilitation outcomes for work-disabled individuals with chronic musculoskeletal pain. J Occup Rehabil. 2006;16:659–674
    • View In Article
    • MEDLINE
    • CrossRef
69. Moore A, Jull G. Educator skills for musculoskeletal therapy practice: do we use these skills effectively and how and when are they used?. Man Ther. 2008;13:91–92
    • View In Article
    • Full Text
    • Full-Text PDF (94 KB)
    • CrossRef
70. Carr JL, Klaber Moffett JA, Howarth E, et al. A randomized trial comparing a group exercise programme for back pain patients with individual physiotherapy in a severely deprived area. Disabil Rehabil. 2005;27:929–937
    • View In Article
    • MEDLINE
    • CrossRef
71. Hurley MV, Bearne LM. Non-exercise physical therapies for musculoskeletal conditions. Best Pract Res Clin Rheumatol. 2008;22:419–433
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (356 KB)
    • CrossRef
72. Jette AM, Delitto A. Physical therapy treatment choices for musculoskeletal impairments. Phys Ther. 1997;77:145–154
    • View In Article
    • MEDLINE
73. Childs JD, Fritz JM, Flynn TW, et al. A clinical prediction rule to identify patients with low back pain most likely to benefit from spinal manipulation: a validation study. Ann Intern Med. 2004;141:920–928
    • View In Article
74. Fritz JM, Brennan GP, Leaman H. Does the evidence for spinal manipulation translate into better outcomes in routine clinical care for patients with occupational low back pain? (A case-control study). Spine J. 2006;6:289–295
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (159 KB)
    • CrossRef
75. Fritz JM, Whitman JM, Flynn TW, Wainner RS, Childs JD. Factors related to the inability of individuals with low back pain to improve with a spinal manipulation. Phys Ther. 2004;84:173–190
    • View In Article
    • MEDLINE
76. Gross AR, Aker PD, Goldsmith CH, Peloso P. Physical medicine modalities for mechanical neck disorders. Cochrane Database Syst Rev. 2000;CD000961
    • View In Article
77. Kroeling P, Gross A, Houghton PE. Electrotherapy for neck disorders. Cochrane Database Syst Rev. 2005;CD004251
    • View In Article
78. Kroeling P, Gross AR, Goldsmith CH. A Cochrane review of electrotherapy for mechanical neck disorders. Spine. 2005;30:E641–E648
    • View In Article
    • CrossRef
79. McCarthy CJ, Callaghan MJ, Oldham JA. Pulsed electromagnetic energy treatment offers no clinical benefit in reducing the pain of knee osteoarthritis: a systematic review. BMC Musculoskelet Disord. 2006;7:51
    • View In Article
    • MEDLINE
    • CrossRef
80. Michener LA, Walsworth MK, Burnet EN. Effectiveness of rehabilitation for patients with subacromial impingement syndrome: a systematic review. J Hand Ther. 2004;17:152–164
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (158 KB)
    • CrossRef
81. Osiri M, Welch V, Brosseau L, et al. Transcutaneous electrical nerve stimulation for knee osteoarthritis. Cochrane Database Syst Rev. 2000;CD002823
    • View In Article
82. Cetin N, Aytar A, Atalay A, Akman MN. Comparing hot pack, short-wave diathermy, ultrasound, and TENS on isokinetic strength, pain, and functional status of women with osteoarthritic knees: a single-blind, randomized, controlled trial. Am J Phys Med Rehabil. 2008;87:443–451
    • View In Article
    • CrossRef
83. Porter M, Teisberg E. Redefining health care (Creating value-based competition on results). Boston (MA): Harvard Business School Press; 2006;
    • View In Article
84. Hart DL. The power of outcomes: FOTO Industrial Outcomes Tool—initial assessment. Work. 2001;16:39–51
    • View In Article
85. Hart DL, Dobrzykowski EA. Influence of orthopaedic clinical specialist certification on clinical outcomes. J Orthop Sports Phys Ther. 2000;30:183–193
    • View In Article
    • MEDLINE
86. Jette DU, Jette AM. Physical therapy and health outcomes in patients with knee impairments. Phys Ther. 1996;76:1178–1187
    • View In Article
    • MEDLINE
87. Jette DU, Jette AM. Physical therapy and health outcomes in patients with spinal impairments. Phys Ther. 1996;76:930–941discussion 942-5
    • View In Article
    • MEDLINE
88. Huisman M, Kunst A, Deeg D, Grigoletto F, Nusselder W, Mackenbach J. Educational inequalities in the prevalence and incidence of disability in Italy and the Netherlands were observed. J Clin Epidemiol. 2005;58:1058–1065
    • View In Article
    • MEDLINE
89. Lubetkin EI, Jia H, Franks P, Gold MR. Relationship among sociodemographic factors, clinical conditions, and health-related quality of life: examining the EQ-5D in the US general population. Qual Life Res. 2005;14:2187–2196
    • View In Article
    • MEDLINE
    • CrossRef
90. Atlas SJ, Tosteson TD, Hanscom B, et al. What is different about workers' compensation patients? (Socioeconomic predictors of baseline disability status among patients with lumbar radiculopathy). Spine. 2007;32:2019–2026
    • View In Article
    • CrossRef
91. Davis ET, Lingard EA, Schemitsch EH, Waddell JP. Effects of socioeconomic status on patients' outcome after total knee arthroplasty. Int J Qual Health Care. 2008;20:40–46
    • View In Article
    • CrossRef
92. Childs JD, Fritz JM, Piva SR, Erhard RE. Clinical decision making in the identification of patients likely to benefit from spinal manipulation: a traditional versus an evidence-based approach. J Orthop Sports Phys Ther. 2003;33:259–272
    • View In Article
    • MEDLINE
93. Childs JD, Fritz JM, Piva SR, Whitman JM. Proposal of a classification system for patients with neck pain. J Orthop Sports Phys Ther. 2004;34:686–696discussion 697-700
    • View In Article
    • MEDLINE
94. Delitto A, Cibulka MT, Erhard RE, Bowling RW, Tenhula JA. Evidence for use of an extension-mobilization category in acute low back syndrome: a prescriptive validation pilot study. Phys Ther. 1993;73:216–222discussion 223-8
    • View In Article
    • MEDLINE
95. 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;75:470–485discussion 485-9
    • View In Article
    • MEDLINE
96. Donelson R. Evidence-based low back pain classification (Improving care at its foundation). Europa medicophysica. 2004;40:37–44
    • View In Article
    • MEDLINE
97. Donelson R. Rapidly reversible low back pain: an evidence-based pathway to widespread recoveries and savings. Hanover: Self Care First; 2007;
    • View In Article
98. Flynn T, Fritz J, Whitman J, et al. A clinical prediction rule for classifying patients with low back pain who demonstrate short-term improvement with spinal manipulation. Spine. 2002;27:2835–2843
    • View In Article
    • CrossRef
99. Fritz JM, Brennan GP. Preliminary examination of a proposed treatment-based classification system for patients receiving physical therapy interventions for neck pain. Phys Ther. 2007;87:513–524
    • View In Article
    • MEDLINE
    • CrossRef
100. Fritz JM, Cleland JA, Childs JD. Subgrouping patients with low back pain: evolution of a classification approach to physical therapy. J Orthop Sports Phys Ther. 2007;37:290–302
     • View In Article
101. Fritz JM, Lindsay W, Matheson JW, et al. Is there a subgroup of patients with low back pain likely to benefit from mechanical traction? (Results of a randomized clinical trial and subgrouping analysis). Spine. 2007;32:E793–E800
     • View In Article
     • CrossRef
102. Werneke M, Hart DL. Centralization phenomenon as a prognostic factor for chronic low back pain and disability. Spine. 2001;26:758–764discussion 765
     • View In Article
     • MEDLINE
     • CrossRef
103. Werneke M, Hart DL. Discriminant validity and relative precision for classifying patients with nonspecific neck and back pain by anatomic pain patterns. Spine. 2003;28:161–166
     • View In Article
     • CrossRef
104. Werneke M, May S. The centralization phenomenon and fear-avoidance beliefs as prognostic factors for acute low back pain. J Orthop Sports Phys Ther. 2005;35:844–845author reply 845-7
     • View In Article
     • MEDLINE
     • CrossRef
105. Werneke MW, Harris DE, Lichter RL. Clinical effectiveness of behavioral signs for screening chronic low-back pain patients in a work-oriented physical rehabilitation program. Spine. 1993;18:2412–2418
     • View In Article
     • MEDLINE
     • CrossRef
106. Werneke MW, Hart DL. Categorizing patients with occupational low back pain by use of the Quebec Task Force Classification system versus pain pattern classification procedures: discriminant and predictive validity. Phys Ther. 2004;84:243–254
     • View In Article
     • MEDLINE
107. Werneke MW, Hart DL, Resnik L, Stratford PW, Reyes A. Centralization: prevalence and effect on treatment outcomes using a standardized operational definition and measurement method. J Orthop Sports Phys Ther. 2008;38:116–125
     • View In Article

• a Focus On Therapeutic Outcomes Inc, PO Box 11444, Knoxville, TN 37919.
• b SPSS Inc, 233 S Wacker Dr, Chicago, IL 60606.