As the physical therapy profession

Transcription

1 JASON RODEGHERO, PT, PhD 1,2,5 YING-CHIH WANG, OT, PhD 3,4 TIMOTHY FLYNN, PT, PhD 1,5 JOSHUA A. CLELAND, PT, PhD 6 ROBERT S. WAINNER, PT, PhD 5,7,8 JULIE M. WHITMAN, PT, DSc 5,8 The Impact of Physical Therapy Residency or Fellowship Education on Clinical Outcomes for Patients With Musculoskeletal Conditions TTSTUDY DESIGN: A retrospective cohort design was conducted using data from an electronic survey and an existing commercial outcomes database. TTOBJECTIVE: To compare the clinical outcomes of patients with musculoskeletal conditions treated by physical therapists who had completed residency or fellowship programs versus those who had not. TTBACKGROUND: There is an increasing focus on specialization through postprofessional education in physical therapy residency and fellowship programs. Scant evidence exists that evaluates the influence of postprofessional clinical education on actual patient outcomes. TTMETHODS: Physical therapists using a national outcomes database were surveyed to determine their level of postprofessional education. Survey responders were categorized into 1 of 3 groups that included no residency or fellowship training, residency trained, or fellowship trained. Outcomes for patients with musculoskeletal conditions treated by 363 therapists from June 2012 to June 2013 were extracted from the database. These data were analyzed to identify any differences in functional status change and efficiency achieved between the 3 groups. Potentially confounding variables were controlled for statistically. TTRESULTS: The fellowship-trained group of physical therapists achieved functional status changes and efficiency that were greater than those of the other groups. No difference in functional status change was observed between the residency group and the therapists without residency or fellowship training. The group without residency or fellowship training was more efficient than the residencytrained group. Fellowship-trained therapists were more likely to achieve greater treatment effect sizes than therapists without residency or fellowship training. Residency-trained therapists were less likely to achieve greater treatment effect sizes than the therapists without residency or fellowship training. TTCONCLUSION: These data demonstrate that fellowship training may contribute to statistically greater patient outcomes. Residency training did not appear to contribute to improved patient functional status change or efficiency. It is unknown whether the statistical differences observed would be clinically meaningful for patients. J Orthop Sports Phys Ther 2015;45(2): Epub 10 Jan doi: /jospt TTKEY WORDS: residency, fellowship, outcomes As the physical therapy profession has advanced to a doctoral-level profession, there has been greater emphasis on autonomous practice and on physical therapists achieving recognition as the musculoskeletal health provider of choice. 12,30 This may contribute to a desire or sense of obligation among clinicians to pursue postprofessional education. 17 For physical therapists, this education may be in the form of continuing education or the systematic mentored training of a residency or fellowship program. Clinical experience and postprofessional education are dimensions of expertise in physical therapy. 7,12,16 Knowledge and clinical reasoning have been identified as critical elements of clinical expertise. 32 Knowledge can be obtained through many formats, whereas some argue that clinical reasoning is a dimension of expertise that may require some level 1 OSF Saint James John W. Albrecht Medical Center, Pontiac, IL. 2 Rocky Mountain University of Health Professions, Provo, UT. 3 Focus On Therapeutic Outcomes, Inc, Knoxville, TN. 4 University of Wisconsin-Milwaukee, Milwaukee, WI. 5 Evidence In Motion, Louisville, KY. 6 Franklin Pierce University, Manchester, NH. 7 Texas State University, San Marcos, TX. 8 South College, Knoxville, TN. Dr Rodeghero is the program director of the Evidence In Motion (EIM) Orthopaedic Residency Program. He did not receive any direct or indirect funds from EIM or Focus On Therapeutic Outcomes, Inc (FOTO) for this research. Drs Flynn and Wainner are owners of EIM, which offers postprofessional residency and fellowship programs. Evidence In Motion uses services provided by FOTO to collect and track patient outcome data for residents and fellows. Neither of these authors received any direct or indirect funds from FOTO or EIM for this research. Dr Whitman is the program director of the EIM Fellowship Program. She did not receive any direct or indirect funds from FOTO or EIM for this research. Dr Wang is a paid consultant for FOTO as well as a member of the FOTO Research Advisory Board. The other authors do not hold any ownership stake in the commercial database used in this project. They do practice in clinics and teach in postprofessional education programs that use the database for patient outcome management. No benefits were solicited or received from the company for the completion of this research. This study was approved by the Institutional Review Board at Rocky Mountain University of Health Professions, Provo, UT. Address correspondence to Dr Jason Rodeghero, Rehabilitation Services Manager, OSF Saint James John W. Albrecht Medical Center, 2500 West Reynolds Street, Pontiac, IL Jason.R.Rodeghero@osfhealthcare.org t Copyright 2015 Journal of Orthopaedic & Sports Physical Therapy 86 february 2015 volume 45 number 2 journal of orthopaedic & sports physical therapy

2 of mentored clinical practice. 10,12,16 Frustration with poor clinical outcomes may motivate physical therapists to seek continuing education. 16 The traditional path to attaining advanced knowledge in physical therapy has been through continuing education (CE) courses that vary widely in content, time, and delivery methods. Most states have mandatory CE requirements for physical therapists, but there is no existing national standard. 4 However, few researchers have investigated the effect of education, CE with a monitoring and mentorship program, and formal postprofessional residency or fellowship training on actual clinical outcomes. Two studies have investigated the impact of a CE course on patient outcomes in physical therapy. 6,10 Brennan et al 6 studied the effect of a 2-day CE course on patient outcomes in 34 physical therapists from 13 outpatient clinics. This course consisted of education and laboratory skills practice of manual therapy and exercise techniques. After completion of the 2-day course, a subset of physical therapists (n = 11) from 4 of the clinics participated in a clinical improvement project for 6 months. Clinical outcomes data were analyzed over a 25-month period. There was no difference in patient outcomes achieved between attendees and nonattendees of the 2-day course, except for the physical therapists who participated in the clinical improvement project, who achieved better outcomes compared to those who did not participate in CE. The authors argued that the passive model of CE is potentially insufficient to influence practice patterns to a level that actually benefits patients, unless there is a standardized, formal process for reinforcement of the material into clinical practice. 6 Cleland et al 10 conducted a subsequent trial in which physical therapists participated in a 2-day (8-hour) CE course focused on examination and treatment of patients with cervicothoracic disorders. After the course, physical therapists were randomized into either a group that received ongoing education (two 1.5-hour follow-up presentations and 1 cotreatment session with a mentor) or a group that did not receive any further education or mentoring. The results supported those of Brennan et al, 6 in that physical therapists who only attended the CE program did not achieve superior outcomes compared to those who did not attend the course. However, the physical therapists who attended the course in conjunction with the ongoing education and mentorship session did achieve significantly greater improvements in disability outcomes for patients with neck pain when compared to the CE group without ongoing education or mentorship. 10 These studies suggest that traditional, postprofessional CE may not improve clinical practice compared to educational programs that include some form of clinician oversight and formalized reinforcement. Additional research has explored the association between clinical outcomes and physical therapist related variables, such as postprofessional certifications and years of clinical experience. 20,36,37 In these studies, data were extracted from a national database (Focus On Therapeutic Outcomes, Inc [FOTO]) to access patient clinical outcome data, as well as selected physical therapist related information. In 2000, Hart and Dobrzykowski 20 reported on the association of board certification in orthopaedics (OCS) with patient outcomes. Outcome data from 14 physical therapists (7 OCS, 7 not OCS) and 258 patients suggested that OCS-credentialed physical therapists required fewer visits and less overall cost compared to non OCS-certified therapists, though the sample of physical therapists and patients was small. Next, Resnik and Hart 37 examined the difference between expert (having the best patient outcomes) and average (having moderate-level patient outcomes) physical therapists to determine the effect of years of experience and/or postprofessional training or credentials on clinical outcomes. The physical therapists classified as experts were more likely to be board certified in orthopaedics (OCS), graduates of programs approved by the American Academy of Orthopaedic Manual Physical Therapists (AAOMPT), or certified manual therapists (MTC). Clinical experience in years was not identified as an indicator of physical therapists who achieved superior (expert) patient outcomes. Finally, in 2004, Resnik and Hart 36 examined the differences in outcomes for patients with low back pain treated by physical therapists with or without postprofessional certifications. Physical therapists were identified based on the credentials associated with expertise (OCS, FAAOMPT, or MTC). Results from analysis of FOTO data from 930 total physical therapists (26 OCS, 5 FAAOMPT, 7 MTC, 2 FAAOMPT/OCS) suggested that physical therapists with MTC credentials achieved better outcomes on all patient-centered outcome variables, such as the Medical Outcomes Study 36-Item Short-Form Health Survey. While this study 36 demonstrated value in advanced training, specifically manual therapy certification, it was inconclusive for others due to a small sample of physical therapists with various types of education/certifications. Furthermore, this study 36 only examined the outcomes for patients with low back pain. Additionally, in all 3 studies, the researchers did not differentiate whether those with advanced credentials had actually matriculated through a formal training process. Residency programs have been shown to offer certain perceived benefits to graduates. In a survey of orthopaedic residency graduates, there were high ratings from graduates on their ability to logically reason, perform examination skills, treat effectively and efficiently, and diagnose properly. 39 As with CE courses, perceptions of graduates have been studied, but there is a lack of evidence addressing the impact of such programs on patient outcomes. 17,39 There is the potential that a standardized program with an emphasis placed on mentorship may impact patient outcomes in a more positive manner. 10 Formal clinical residency and fellowship programs are potential journal of orthopaedic & sports physical therapy volume 45 number 2 february

3 mechanisms to provide this educational experience. There is an increasing focus on matriculation through formal residency and fellowship training in physical therapy, with scant available evidence to demonstrate how this additional education may actually impact patient outcomes. 2 The specific aim of this study was to investigate the impact of completing an accredited residency or fellowship program on clinical outcomes in patients with musculoskeletal conditions. A secondary aim was to provide initial insight about the value of advanced postprofessional education for the physical therapy profession. METHODS A retrospective cohort design was conducted using data from a survey and an existing commercial database. Physical therapist users of the FOTO database were surveyed to determine what level of education (entry-level and all postprofessional educational programs) each physical therapist had completed. Clinical outcome data for these therapists were then extracted from the database and analyzed to identify any differences in the outcomes achieved and clinical efficiency. The protocol for this study was approved by the Institutional Review Board at Rocky Mountain University of Health Professions in Provo, UT. Outcome Database Focus On Therapeutic Outcomes is a national patient outcome assessment system. The FOTO database contains outcome data from clinics in outpatient rehabilitation settings, nursing homes (skilled nursing), and adult daycare settings. The patient data-collection process has been previously reported. 19,21,22 Clinics that use FOTO administer surveys to each patient using Patient Inquiry computer software developed by FOTO (Focus On Therapeutic Outcomes, Inc, Knoxville, TN). The functional status (FS) outcome measure, developed and used by FOTO, TABLE 1 Minimal Clinically Important Difference for Each Body Region Based on Functional Status Score at Intake Body Region Intake FS MCID* Lumbar > > > Hip > > > Knee > > > Foot/ankle > > > Shoulder > > > Elbow > > > Cervical > > > Abbreviations: FS, functional status; MCID, minimal clinically important difference. *MCID estimates are different for each respective measure and vary based on FS intake score. The MCID values have been reported in previous studies. is body-region specific, with demonstrated validity and reliability for common musculoskeletal conditions. 18,21-27,29 Patients enter demographic information into the system and complete a baseline FS measure prior to their evaluation and receiving intervention. The FS measure requires patients to answer questions about their level of difficulty with various functional activities, as related to the body region requiring treatment. For example, someone receiving treatment for a lumbar disorder could be asked about the level of difficulty with bending or stooping activities. The FOTO system utilizes a computer adaptive testing process, so the type and amount of questions vary based on the patient response. The final FS score represents an estimate of the patient s functional level on a scale from 0 to 100, with higher measures representing higher functioning. At subsequent visits, the therapist or administrative staff issues follow-up surveys, including the discharge FS survey that includes patient satisfaction questions. That episode of care is then completed after the therapist or administrative staff enters 88 february 2015 volume 45 number 2 journal of orthopaedic & sports physical therapy

4 Physical therapists without residency or fellowship training, n = 387 Physical therapists with patients in FOTO from June 2012 to June 2013, n = 306 Patients treated by physical therapists without residency or fellowship training included in analyses, n = additional information, including last date of service and number of treatment sessions. This database was selected because it allowed access to a large group of physical therapists and patients from various geographical regions to examine the influence of education on patient outcomes. Physical Therapists Physical therapists using FOTO were surveyed via to identify level of education and personal demographics. The FOTO user survey was created with the assistance of content experts on postprofessional education, including a current director of an orthopaedic residency program accredited by the American Board of Physical Therapy Residency and Fellowship Education (ABPTRFE), a current director of an ABPTRFE-accredited manual physical therapy fellowship program, and a former director of an Physical therapists registered using FOTO linked to addresses, n = 6809 Physical therapists who returned completed surveys, n = 479 Residency-trained physical therapists, n = 73 Physical therapists with patients in FOTO from June 2012 to June 2013, n = 45 Patients treated by residency-trained physical therapists included in analyses, n = 4042 Fellowship-trained physical therapists, n = 19 Physical therapists with patients in FOTO from June 2012 to June 2013, n = 12 Patients treated by fellowship-trained physical therapists included in analyses, n = 1083 FIGURE 1. Identification of study sample. Abbreviation: FOTO, Focus On Therapeutic Outcomes, Inc. ABPTRFE-accredited manual physical therapy fellowship program. Three of the experts also hold academic appointments in entry-level programs, postprofessional degree programs, residency programs, and fellowship programs. The survey was uploaded to a web-based surveying company for electronic distribution and collection. Completion of the survey implied consent of the physical therapists to be included in this study. The survey remained open for 6 weeks, with reminders sent at 2 weeks and 4 weeks. A total of s were sent to a potential sample of 6809 registered physical therapist users of FOTO. The discrepancy in the number of s and total therapists was due to some practices only providing 1 address for all of their providers. Therefore, it is possible that the survey was not made available to all physical therapists using FOTO. It is unknown how many total therapists received the survey due to this discrepancy. Data from the completed surveys were downloaded to a PASW Statistics 18 (SPSS Inc, Chicago, IL) data set. Patients A deidentified patient data set was provided from FOTO. Specific patient information and outcome data from June 2012 to June 2013 for each physical therapist were extracted. All patients 18 years or older with musculoskeletal impairments (neck, shoulder, elbow/wrist/ hand, lumbar, hip, knee, and foot/ankle) and complete data (intake scores, discharge scores, and number of treatment sessions) were included in the data set. Personal patient information was removed, and each patient was assigned a patient identification number. This data set was merged with the therapist survey data in PASW. Variables Physical therapists were classified into 1 of 3 clinical groups based on their postprofessional education: (1) residency program completion, (2) fellowship program completion, or (3) no residency or fellowship training. There was uncertainty about clarity of dual program completion, as residency and fellowship have not always been clearly defined. 36 Any therapist who had reported completing both types of programs (n = 4) was classified into the fellowship-trained group, as this reflects a higher level of postprofessional clinical education. Patient-specific independent variables in this study were consistent with other studies, including those that have utilized FOTO and others with theoretical relevance. 3,9,36,37 To compare the effectiveness between clinical groups, both patient outcomes and clinical efficiency were dependent variables. Patient outcomes were assessed using the FS change score, which was calculated by subtracting the intake FS score from the FS score at discharge. Clinical efficiency was calculated by dividing the FS change score by the number of treatment sessions. 6,9 journal of orthopaedic & sports physical therapy volume 45 number 2 february

5 The minimal clinically important difference (MCID) values of the FS measures for each body region have previously been reported 24-27,29,40-44 and are listed in TABLE 1. Different MCID cut points have been used to assess predictive clinical models. 38 We created additional variables, using established MCID values and patient FS change to dichotomize patients who achieved various treatmenteffect cut points (MCID 1-5). Patients whose FS change met or exceeded the desired MCID cut point were classified as a responder at that level (coded as 1), and patients who did not achieve the levels of FS change were classified as a nonresponder (coded as 0). For example, a patient with a lumbar condition presenting with an FS intake score of 53 would have an MCID of 3 (TABLE 1). The variables created (MCID 1-5) for this patient would be based on their initial MCID of 3 and would include 3 (MCID 1), 6 (MCID 2), 9 (MCID 3), 12 (MCID 4), and 15 (MCID 5). If the patient s FS change was 10, the patient was classified as a responder for the MCID 1 to MCID 3 cut points but a nonresponder for the MCID 4 and MCID 5 cut points. These data were used to assess the odds of achieving a treatment effect of various magnitudes between groups. Data Analysis Selection Bias We used chi-square statistics with standardized deviates for categorical data and analyses of variance for continuous variables to assess differences among the 3 clinical groups for physical therapist characteristics (age, years of experience, entry-level degree, postprofessional training, and practice setting) and patient characteristics (eg, age, sex, and number of comorbidities). Effectiveness Comparisons For each dependent variable (FS change score and clinical efficiency), 3 general linear models or 1-way analyses of covariance were employed. Each model was designed to determine if there was a difference in the dependent variables, while controlling for important independent variables, TABLE 2 with the main factor being the clinical group (fellowship trained, residency trained, or neither). In the first model, the researchers compared the differences between clinical groups, while controlling for the intake FS score (covariate). In the second model, the researchers controlled for patient variables provided by FOTO, including type of impairment (lumbar, shoulder, elbow/wrist/hand, hip, knee, foot/ankle, cervical, other), age, symptom acuity, sex, surgical history, payer, number of functional comorbidities, exercise history, use of medication at intake, and fear-avoidance beliefs, and intake FS Characteristics of Physical Therapists Variables No Residency or Fellowship (n = 306) Residency (n = 45) Fellowship (n = 12) Sex (female), % Age, n y y y y Experience, n* 0-5 y y y Entry-level degree, n* Certificate Bachelors Masters Doctorate Earned tdpt, n* Earned advanced masters, n Earned academic doctorate, n* Manual therapy certified, n* OCS, n* SCS, n* Practice setting, % Private practice Hospital practice Abbreviations: OCS, board-certified orthopaedic clinical specialist; SCS, board-certified sports clinical specialist; tdpt, transitional Doctor of Physical Therapy. *Significant difference observed on Pearson chi-square (P<.05). Significantly fewer observations than expected (standardized residual less than critical value of 1.96). More observations than expected (standardized residual greater than critical value of 1.96). score (covariate). Patient age and intake FS score were entered as continuous variables, whereas the rest of the independent variables were treated as categorical variables. Symptom acuity, which was operationally defined as the number of calendar days from the date of onset of the condition being treated in therapy to the date of initial therapy evaluation, was categorized as acute (less than 22 days), subacute (between 22 and 90 days), and chronic (greater than 90 days). Surgical history related to the impairment being treated was classified into 5 categories (none, 1, 2, 3, 4 or more). Payer was categorized into 16 payer sources (eg, health 90 february 2015 volume 45 number 2 journal of orthopaedic & sports physical therapy

6 TABLE 3 Variable maintenance organization, preferred provider organization). Number of functional comorbid conditions was assessed using a list of 30 conditions common to patients entering an outpatient rehabilitation clinic (eg, arthritis, asthma, diabetes, heart attack, autoimmune deficiency syndrome, sleep disturbance, overweight, and cancer). Exercise history prior to receiving therapy was categorized as exercising 3 times a week or more, exercising 1 to 2 times a week, or exercising seldom or never. Use of medication at intake was a dichotomous response (yes, no). Last, patients were classified into high (elevated) or low (not-elevated) fear-avoidance beliefs, based on screening items developed by Hart et al. 28 In the third model, the researchers controlled for both the patient variables, Characteristics of Patients No Residency or Fellowship (n = Patients) Residency (n = 4042 Patients) Fellowship (n = 1083 Patients) Mean age, y* Sex (female), % Comorbidities, n* Fear-avoidance beliefs (FOTO score, 0-100)* Surgical history (without surgery), % Duration of symptoms (>91 d), % FS intake score (0-100)* Mean treatment sessions per episode, n* Body region of primary symptoms, n (%) Lumbar 4979 (24) 1086 (27) 314 (29) Shoulder 4207 (20) 816 (20) 191 (18) Knee 4040 (20) 673 (17) 124 (11) Cervical 2491 (12) 484 (12) 163 (15) Other 5001 (24) 983 (24) 291 (27) Abbreviations: FOTO, Focus On Therapeutic Outcomes, Inc; FS, functional status. *Significant difference with analysis of variance (P<.05). Age: no residency or fellowship greater than residency greater than fellowship (P<.001). Number of comorbidities: no residency or fellowship greater than residency (P<.001) and no residency or fellowship greater than fellowship (P =.02). Fear-avoidance beliefs: residency greater than no residency or fellowship (P =.03). FS intake: fellowship greater than residency (P =.03) and fellowship and residency greater than no residency or fellowship (P<.001). Treatment sessions: fellowship less than residency and no residency or fellowship (P<.001). Significant difference with Pearson chi-square analysis (P<.05). Significantly fewer observations than expected (standardized residual less than critical value of 1.96). FOTO fear-avoidance beliefs score ranges from 0 (lowest fear) to 100 (greatest fear) and is calculated based on the Fear-Avoidance Beliefs Questionnaire-physical activities subscale. More observations than expected (standardized residual greater than critical value of 1.96). FS score ranges from 0 to 100, with a lower score indicating greater functional limitations. as described above (in the second model), and physical therapist variables obtained from the survey, including years of experience, earned transitional Doctor of Physical Therapy degree, board certification in orthopaedics, board certification in sports, and certification as a manual therapist. Selection of physical therapist related variables was based on significant differences observed between the clinical groups. Years of physical therapy experience was operationally defined as years actively practicing, and physical therapists were categorized as 0 to 5, 6 to 9, and 10 or more years. Board certification, manual therapy certification, and completion of a postprofessional degree were dichotomous variables categorized as either yes or no for each. A series of 5 binary logistic regression analyses was used to compute the odds ratios for achieving the desired MCID thresholds (1 to 5 times the MCID) for patients within each clinical group. Patients were coded as a responder (0 or 1) at various levels, based on the MCID level of change they achieved. The 5 responder groups were entered as the dependent variable in subsequent modeling. The independent variable consisted of the clinical group, with the group of physical therapists without residency or fellowship training treated as the reference group. Finally, a chi-square test for association was conducted to assess the associations between clinical groups and MCID levels achieved. The level of statistical significance was set at P.05, with all analyses conducted using PASW. RESULTS A total of 479 therapists participated in the survey, representing a 21% rate of response to the total s sent out and 7% of the total number of reported physical therapists registered with FOTO. Physical therapists were included for further analysis if the survey was complete and the therapist had patient outcomes data in FOTO for the period being studied, resulting in a final sample of 363 physical therapists. The process of sample identification is outlined in FIGURE 1. Physical Therapist Characteristics Characteristics of physical therapists in each clinical group are presented in TABLE 2. The fellowship group was older and had more years of clinical experience compared to the other groups. There appeared to be a greater percentage of female therapists in the residency group and the no residency or fellowship group versus the fellowship group. Patient Characteristics Details for patient characteristics are presented in TABLE 3. Patients treated for each of the clinical groups were similar in age, journal of orthopaedic & sports physical therapy volume 45 number 2 february

7 TABLE 4 sex distribution, duration of symptoms, and number of comorbidities. There was a greater percentage of patients with spinal disorders (lumbar and cervical) in the fellowship group. The residency and no residency or fellowship groups had patients with similar body-region conditions. FS Change The mean FS change for patients for the 3 clinical groups is reported in TABLE 4 and FIGURE 2. Overall, patients treated by therapists in the fellowship group showed statistically greater improvements than those treated by the therapists in the groups with residency training or without residency or fellowship training (P<.001). However, no significant difference (P = ) was observed between the residency group and the no residency or fellowship training group. Clinical Efficiency The efficiency data for the 3 clinical groups are reported in TABLE 5. Overall, therapists in the fellowship group provided more efficient care (greater improvements per number of treatment sessions) than the therapists in the group without residency or fellowship training, followed by the residency group (P =.06 to P<.001). FS Change for Each Clinical Group From ANCOVAs* Group Raw Data Model 1 Model 2 Model 3 No residency or fellowship 17.5 (17.3, 17.7) 17.4 (17.1, 17.6) 10.1 (7.9, 12.3) 8.6 (6.3, 11.0) Residency 16.5 (16.0, 17.0) 17.0 (16.6, 17.5) 9.9 (7.6, 12.2) 8.8 (6.4, 11.1) Fellowship 18.3 (17.3, 19.2) 19.4 (18.6, 20.3) 12.1 (9.7, 14.5) 11.2 (8.8, 13.7) Abbreviation: ANCOVA, analysis of covariance; FS, functional status. *Values are mean (95% confidence interval) FS change. Fellowship greater than residency (P =.003). No residency or fellowship greater than residency (P<.001). No difference between fellowship and no residency or fellowship (P =.41). Adjusted mean. Model 1 controlled for FS intake (covariate). Fellowship greater than residency (P<.001) and no residency or fellowship (P<.001). No difference between residency and no residency or fellowship (P =.52). Adjusted mean. Model 2 controlled for patient characteristics, with FS intake as a covariate. Fellowship greater than residency (P<.001) and no residency or fellowship (P<.001). No difference between residency and no residency or fellowship (P = 1.00). Adjusted mean. Model 3 controlled for patient and physical therapist characteristics, with FS intake as a covariate. Fellowship greater than residency (P<.001) and no residency or fellowship (P<.001). No difference between residency and no residency or fellowship (P = 1.00). Treatment Effect Levels Patients treated by physical therapists with fellowship training had a higher likelihood of achieving 2 to 4 times the MCID for FS change than the therapists without residency or fellowship training (TABLE 6). Patients treated by residencytrained physical therapists were less likely to achieve FS change scores that were 2 to 5 times greater than the MCID (TABLE 6). Nonetheless, these differences may have small clinical significance, as all odds ratios were small (less than 2). Last, chi-square analysis revealed a statistically significant association between clinical groups and achieving the MCID at each level of change (χ 2 = 28.4, P =.002). 10 This indicates that there was a significant association between the clinical group and the magnitude of the MCID achieved. Patients treated by physical therapists with fellowship training had a tendency to achieve treatment effect sizes of greater magnitude (3 to 4 times the MCID; standardized residual, 1.9), whereas those treated by the residency-trained group did not (standardized residual, 2.5). DISCUSSION This study included a large sample of patients with a variety of musculoskeletal conditions who received treatment from physical therapists with various levels of postprofessional education, including completion of ABPTRFE-accredited residency or fellowship training programs. The results of this preliminary study suggest that physical therapists with fellowship training may provide better patient outcomes in fewer treatment sessions compared to other physical therapists. The results of this study also appear to challenge the clinical benefit of completing a residency program. The patients treated by physical therapists with residency training did not achieve greater outcomes or demonstrate greater efficiency than the physical therapists without residency training. In some cases, the therapists without residency or fellowship training were more efficient than the therapists in the residency group (TABLE 6). When known potential confounders were controlled for, patients treated by physical therapists who completed a fellowship program achieved significantly greater improvements in FS than those treated by therapists with residency training and those without residency or fellowship training (TABLE 4). Fellowship graduates achieved the greatest amount of FS change in the fewest number of treatment sessions, resulting in a significantly higher efficiency rating than those with residency training or no training (TABLE 5). The efficiency rating may arguably be the strongest indicator of quality and has been utilized in previous research. 6,9 Last, the physical therapists in the fellowship group had a higher likelihood of achieving 2 to 4 times the MCID for FS change compared to the therapists without residency or fellowship training (TABLE 6). This may further reinforce fellowship training as a way to optimize patient outcomes, although the differences observed were very small and some of the 95% confidence intervals barely exceeded 1. The limited previous research has demonstrated the lack of impact that weekend CE courses have on patient outcomes. 6,10 Research on completing a residency has been limited, but the available 92 february 2015 volume 45 number 2 journal of orthopaedic & sports physical therapy

8 literature has shown increased individual perceptions of clinical skills among graduates. 33,39 It is unknown whether the perception of improved clinical skills translates into a meaningful benefit for patients. The results of this preliminary study do not support residency training as influencing patient outcomes. It is possible that residency programs are still in their relative infancy and that a potential lack of consistency in curricula between programs might have contributed to the results. It is likely that the therapists using FOTO who took the time to complete the survey were already a high-performing group, which could have increased selection bias. This bias could limit the ability to accurately assess changes that may occur in clinical practice during or after a residency program. From the data analyzed, it would appear that alternative or additional education is necessary to significantly impact patient outcomes. Further education, such as a fellowship program, has been reported to result in greater clinical outcomes. 13 The results of this study and the existing literature appear to support the positive influence that formal postprofessional education in the form of fellowship training may have on clinical practice. That influence may include greater improvements in outcome scores and greater efficiency of care. The differences observed in FS change among the 3 clinical groups are relatively small when assessed at face value (TABLE 4). With the large sample and, consequently, the high level of statistical power, small differences like these between groups would be expected to achieve statistical significance. It is unknown whether these statistical differences are clinically meaningful to patients. To further assess the differences, we also considered the magnitude of FS change achieved by each group. The odds ratios of achieving greater magnitudes of FS change (TABLE 6) demonstrated that residency-trained physical therapists were less likely to achieve higher FS changes than those without residency or fellowship training. In contrast, fellowship-trained physical FS Score * Intake FS Model 1 FS Discharge Model 2 FS Discharge No residency or fellowship Residency Fellowship Model 3 FS Discharge FIGURE 2. Baseline and discharge FS scores for each statistical model used (FS score range: minimal score of 0, completely disabled to maximal score of 100, no functional limitations). Model 1 controlled for intake FS. Model 2 controlled for intake FS and known patient characteristics. Model 3 controlled for intake FS, known patient characteristics, and known provider characteristics. *Residency and fellowship intake FS greater than no residency or fellowship intake FS (analysis of variance, P<.05). Fellowship greater than residency (P<.001) and no residency or fellowship (P<.001). No difference between residency and no residency or fellowship (P>.05). Abbreviation: FS, functional status. TABLE 5 Efficiency (FS Change per Number of Treatment Sessions) from ANCOVAs* Mean Treatment Group Sessions, n Raw Data Model 1 Model 2 Model 3 No residency or 11.1 (11.0, 11.2) 2.3 (2.3, 2.3) 2.3 (2.2, 2.3) 1.3 (0.8, 1.8) 1.0 (0.5, 1.5) fellowship Residency 12.0 (11.8, 12.3) 2.0 (1.9, 2.1) 2.0 (1.9, 2.1) 1.0 (0.5, 1.6) 0.8 (0.3, 1.4) Fellowship 9.7 (9.2, 10.2) 3.2 (3.0, 3.4) 3.3 (3.1, 3.4) 2.2 (1.6, 2.7) 1.9 (1.4, 2.5) Abbreviations: ANCOVA, analysis of covariance; FS, functional status. *Values are mean (95% confidence interval). Fellowship greater than no residency or fellowship (P<.001) and residency (P<.001). No residency or fellowship greater than residency (P<.001). Adjusted mean. Model 1 controlled for FS intake (covariate). Fellowship greater than no residency or fellowship (P<.001) and residency (P<.001). No difference between no residency or fellowship and residency (P =.06). Adjusted mean. Model 2 controlled for patient characteristics, with FS intake as a covariate. Fellowship greater than no residency or fellowship (P<.001) and residency (P<.001). No residency or fellowship greater than residency (P =.001). Adjusted mean. Model 3 controlled for patient and physical therapist characteristics, with FS intake as a covariate. Fellowship greater than no residency or fellowship (P<.001) and residency (P<.001). No residency or fellowship greater than residency (P =.019). journal of orthopaedic & sports physical therapy volume 45 number 2 february

9 therapists were more likely to achieve higher FS changes, with more patients improving 2 to 4 times greater than the MCID, than the therapists without residency or fellowship training (TABLE 6). These data may further support the argument that the observed differences in FS change between groups are valid. However, caution in the interpretation of the odds ratio data is warranted, because on occasion the confidence intervals approached and even included 1. Increased efficiency in the fellowship group may be specific to the type of fellowship program completed. All physical therapists with fellowship training who were included in the final sample were graduates of orthopaedic manual physical therapy fellowship programs. There were no physical therapists who completed any other type of fellowship program among the respondents. The increased efficiency in the fellowship group may be specific to the manual physical therapy approach in managing patients with musculoskeletal conditions. In addition to potentially enhanced manual therapy examination and intervention skills, this approach may also provide a greater focus on a hypothetico-deductive reasoning in patient care, as discussed in the AAOMPT educational standards. 1 This approach involves the therapist putting an unstructured problem into a structured one by generating a list of potential hypotheses and then using them to guide or direct further testing and manual physical therapy based interventions. 14,32,34 Pattern recognition in the form of evolving classification-based approaches for the use of specific interventions, including manual physical therapy, has been encouraged and studied. 8,15,31 Physical therapy is strongly focused on movement impairment and its relationship to functional impairment. This fundamental construct in physical therapy is introduced in entry-level education, with refinement and improvement likely achieved over years of experience or formally facilitated through postprofessional programs such as a manual physical TABLE 6 therapy fellowship. Having an approach that has a central focus on movement assessment linked to patient function has also been identified as an important component of expert practice in physical therapy. 32 In a musculoskeletal setting, manual physical therapy fellowships may provide additional benefits beyond those of residency programs. Though the therapists in this study do not likely reflect those of all residency or fellowship programs, the data do allow for dialog on the potential for varying quality between programs. In the changing environment of health care, achieving better outcomes in fewer visits or demonstrating the efficacy of additional visits will likely be important for physical therapists to demonstrate the value of treatment for the care of common musculoskeletal conditions and to validate reimbursement for services. Residency and fellowship programs should ideally be focused on quality outcome measures of their programs, and patient outcome data should be fundamental to the assessment of each program. A common motivator for pursuing advanced clinical education for a physical therapist is to achieve improved patient outcomes; therefore, changes in actual patient outcomes over the duration of a program and upon graduation from a program may arguably be the best arbiter of the quality of education provided. Further, participation in an independent outcome data management program could be valuable for each program to pool data for further analyses. This will help researchers to further investigate the efficacy of residency and fellowship educational programs and contribute to Odds Ratios for Achieving Increasing Levels of Functional Status Change* Group MCID MCID 2 MCID 3 MCID 4 MCID 5 Residency 0.97 (0.90, 1.05) 0.93 (0.87, 0.99) 0.89 (0.83, 0.96) 0.85 (0.78, 0.92) 0.87 (0.78, 0.96) Fellowship 1.18 (1.03, 1.36) 1.16 (1.02, 1.31) 1.20 (1.06, 1.36) 1.20 (1.04, 1.38) 1.17 (0.99, 1.40) Abbreviation: MCID, minimal clinically important difference. *Reference standard: no residency or fellowship. Values in parentheses are 95% confidence interval. the continued evolution of the physical therapy profession. Limitations There are inherent limitations associated with the use of retrospective data from a commercial database Furthermore, the smaller number of patients treated by the residency- and fellowship-trained groups, selection bias, low survey response rates, use of retrospective data from a third-party database that not all therapists participate in, and unidentified confounding variables are threats to the validity of this study. The sample of physical therapists included in this study was strictly chosen based on survey responses. The response rate of 21% (based on total s delivered) in this study is consistent with the rate traditionally seen in web-based and organizational surveys, 5,11 though only 7% of the total users registered in the FOTO system responded to the survey. In our opinion, the sample of patient data analyzed was extremely large (roughly patients from 363 physical therapists), which strengthens the results, despite representing only a small percentage of FOTO database users. The patient sample was composed of patients in the FOTO database from the physical therapists who responded to the survey. No effort was made to contact other residency- or fellowship-trained physical therapists not using the FOTO database. Use of a commercial database does present threats to both internal and external validity. Only patients with both intake and discharge FS scores were considered for the analyses. Patients who did not complete an episode of care 94 february 2015 volume 45 number 2 journal of orthopaedic & sports physical therapy

10 might have discontinued treatment due to limited improvement or worsening of symptoms. Conversely, patients might have elected to discontinue physical therapy due to significant improvement and the sense that treatment was no longer needed. In either case, a discharge FS score would not likely be collected, and these patients would not be reflected in the results. There was no attempt to collect and analyze the amount of missing data for each group, and imputation of missing dependent variables would result in additional, possibly greater, challenges to validity. Previous research has demonstrated consistent amounts of missing data between different patient groups. 9,37 Similar data would likely be missing in the sample of this study, but the amount of missing data is not known because only completed episodes were extracted. This increases the possibility of selection bias. It would be premature to generalize the results of this study to all physical therapists. There was no attempt to identify which specific residency or fellowship programs the FOTO users completed and, therefore, attributes of specific programs are a potential confounding factor. Individuals who pursue fellowship training may inherently be stronger clinicians due to their personal motivation and professional desire to excel. It is arguable that this type of individual might otherwise achieve greater outcomes with patients, regardless of formal postprofessional education. The self-selection of these physical therapists to complete formal fellowship training does introduce additional layers of bias that need to be recognized. Generalizability of findings would be improved with a prospective design that included all residency and fellowship programs with patient outcome data that were not limited to 1 database. A project of this magnitude would not likely be feasible in the current physical therapy environment. Therefore, analyzing existing data is an initial step in the process of assessing the efficacy of postprofessional residency and fellowship training. CONCLUSION The results of this study appear to support the benefit that fellowship training may have on patient outcomes. Conversely, the results appear to challenge the benefit that completing a residency has on patient outcomes. After controlling for potential confounding variables, patients treated by fellowshiptrained therapists demonstrated larger effect sizes, achieved with slightly more efficiency, than those treated by other therapists. Patients treated by residency-trained physical therapists did not achieve greater outcomes than the group without residency training. The data indicate that the therapists without residency training achieved equivalent outcomes to residency-trained therapists, but did so in a more efficient manner. The findings of this study warrant further research using larger and more diverse samples to determine the impact of residency and fellowship training on improved physical therapy care delivery and outcomes. t KEY POINTS FINDINGS: Physical therapists with fellowship training achieved greater FS outcomes and had greater efficiency than therapists with residency training or therapists without residency or fellowship training. There was no difference in adjusted FS outcomes between the group of therapists with residency training and the group without residency or fellowship training. The therapists without residency or fellowship training had greater adjusted efficiency than the residency-trained therapists. IMPLICATIONS: There may be benefit from manual physical therapy fellowship training that contributes to greater functional outcomes in fewer treatment sessions for patients with musculoskeletal conditions. The clinical benefit (patient outcomes) of residency training was not reflected in the patient outcome data analyzed in this study. CAUTION: The groups were determined by a relatively small amount of survey responses from physical therapists using FOTO. There was no attempt to identify specific residency or fellowship programs, introducing uncontrolled variance. No attempt was made to include data from residency or fellowship programs not using FOTO. Selection bias should be considered when interpreting the results. ACKNOWLEDGEMENTS: This project was completed with the support of the 2012 OPTP Research Grant issued by the AAOMPT. REFERENCES 1. American Academy of Orthopaedic Manual Physical Therapists. Educational Standards in Orthopaedic Manual Physical Therapy. Baton Rouge, LA: American Academy of Orthopaedic Manual Physical Therapists; American Physical Therapy Association. Conference Dispatch: The 18th Maley Lecturer: Kornelia Kulig, PT, PhD, FAPTA, FAAOMPT. Available at: watch?v=qczhb68kfnk. Accessed August 9, Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36: Austin TM, Graber KC. Physical therapists perspectives on the role and effectiveness of continuing education. J Allied Health. 2007;36: Baruch Y, Holtom BC. Survey response rate levels and trends in organizational research. Hum Relations. 2008;61: org/ / Brennan GP, Fritz JM, Hunter SJ. Impact of continuing education interventions on clinical outcomes of patients with neck pain who received physical therapy. Phys Ther. 2006;86: Brosky JA, Jr., Scott R. Professional competence in physical therapy. J Allied Health. 2007;36: 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: ; discussion dx.doi.org/ /jospt Childs JD, Harman JS, Rodeghero JR, Horn M, George SZ. Implications of practice setting on clinical outcomes and efficiency of care in the delivery of physical therapy services. J Orthop Sports Phys Ther. 2014;44: dx.doi.org/ /jospt Cleland JA, Fritz JM, Brennan GP, Magel J. Does journal of orthopaedic & sports physical therapy volume 45 number 2 february