An Evaluation of the Internal Reliability, Construct Validity, and Predictive Validity of the Physical Therapist Clinical Performance Instrument (PT CPI)

Headnote

Background and Purpose. Despite its widespread use across the United States and Canada to assess physical therapist student performance, limited research has been published regarding the psychometric properties of the Physical Therapist Clinical Performance Instrument (PT CPI). The main purpose of this study was to evaluate the internal reliability, construct validity, and predictive validity of the PT CPI.

Subjects. Existing data from 7 graduating classes (2000-2006, n = 154) of a physical therapist education program were used in this retrospective design. Final PT CPI visual analogue scale (VAS) scores for 3 separate clinical experiences (CEs 1, 2, and 3), final cumulative GPAs, gender, prior degree, and National Physical Therapy Examination (NPTE) outcomes were available for 147 students.

Methods. Factor analysis of PT CPI VAS scores was used to identify possible underlying dimensions within the construct of entry-level performance. Factors identified became subscales for further analyses. To further evaluate construct validity of the PT CPI, Cronbach's alpha was used to assess internal reliability, ANOVAs were conducted to assess student progression over time, and logistic regression was performed to investigate the ability of the tool to predict NPTE outcomes.

Results. Three factors emerged from factor analysis of CE I data, explaining 72% of the variance in CE I scores. These factors were named as the following subscales: Integrated Patient Management (IPM), Professional Practice (PP), and Career Responsibilities (CR). To represent the tool as a whole, a fourth subscale was computed consisting of the total of all items scored for each subject, with the exclusion of item 17. Item 17, Consultation, was the least frequently marked item and was marked less than 50% of the time. Cronbach's alpha values for the subscales ranged from 0.75-0.96, indicating strong internal reliability. Repeated measure ANOVAs demonstrated a significant main effect for Time for all subscales indicating that scores improved across CEs. However, a significant Subscale-by-Time interaction demonstrated that subscales did not progress equally. PP mean scores increased 13.5% from CE 1 to CE 3, while IPM mean scores increased 51.4% from CE 1 to CE 3. PT CPI subscales were not predictive of NPTE outcome. GPA was the only significant predictor of NPTE outcome.

Discussion and Conclusion. The current PT CPI is a multidimensional, internally reliable instrument demonstrating construct validity....