The development of the IPA was extensive and well executed from 2006-2015 over three phases: (1) Construct development and generation of observable behaviors and response scales; (2) Content expert review and cognitive interviews with typical raters; and (3) a two-year pilot study. The process began with literature review, construct definition, and the organization of 200 potential behaviors into categories by the Interprofessional Professionalism Collaborative (IPC). The number of behaviors was reduced to 43 after the IPC applied explicit inclusion criteria (e.g., observable in practice, applicable across the professions, not redundant). Members of the IPC then made national and international presentations about the tool, documented oral feedback from audience members, and collected follow-up online survey feedback from 205 individuals representing 11 professions . This feedback led to the formatting of a 39-item instrument, which was then reviewed by a panel of 23 expert reviewers from the U.S. and Canada. The panel responded to structured survey questions about the tool’s content, fit of 39 behavioral items within and across six domains, overall organization, format, and length.
Twenty-four preceptors, two from each of the 12 IPC member health professions representing “typical” preceptors that would use the tool, were involved in the cognitive interviews. Based on their feedback, the IPA was reduced to 26 items. It was this version of the IPA which was used in a large, multi-institution and multi-profession pilot study. A total of 67 academic institutions were invited to participate in the pilot; 30 agreed to do so (44.8%). Using a key contact method, nearly 3,000 preceptors (estimated) were invited into the study; 376 agreed and 233 provided data (62% of enrolled, 7.9% of potential population).
Exploratory factor analysis (EFA) was conducted on preceptors’ ratings of their learners assuming ordered categorical factor indicators. To determine the number of factors to retain, eigenvalues and measures of fit were examined [i.e., the root mean square error of approximation (RMSEA), the comparative fit index (CFI), and the standardized root mean square residual (SRMR)]. Prior to factor analysis, the extent of missing data for each of the IPA items was examined (i.e., an item either left blank or the respondent recorded N/O – No opportunity to observe in this environment). Internal consistency reliability of the factors suggested by the EFA was calculated using coefficient alpha. The initial EFA using 21 items (excluding five items with extensive missing data) suggested retaining four factors. With eigenvalues of 12.670, 1.229, 0.888, and 0.787, the four factors together accounted for 86.5% of the variance in the set of variables, and the fit indices indicated good model fit (RMSEA = 0.064, 90% CI: 0.055 – 0.078; CFI = 0.991; SRMR = 0.027). The four factors loaded well on the following domains: Communication, Respect, Excellence, Altruisim and Caring. Internal reliability consistency coefficients were high (alpha >0.94) for each of the factors. Despite the psychometric results, and based on other considerations, the study authors decided to keep the 5 excluded items and the two other domains (Ethics and Accountability) in the final instrument.
Relation to Other Variables:
During the pilot study, responding preceptors were also asked to complete two global items for each learner they evaluated: one was a global rating of the learner's interprofessional professionalism, and the other a global rating of the learner’s overall performance on the practice experience. These ratings were made using a 5-point Likert-type scale (1 = "poor," 5 = "excellent"). Given the results of the factor analysis, items within each domain were averaged to create subscale scores and factor scores were also estimated from the final factor model. These scores were all positively and significantly correlated with the two global performance items described above.