Validation of the SPICE-R instrument among a diverse interprofessional cohort: A cautionary tale

ObjectiveThe Student Perceptions of Interprofessional Clinical Education-Revised (SPICE-R) instrument was developed to measure interprofessional teamwork in a clinical setting. This study investigated whether the SPICE-R could adequately measure teamwork in a classroom-based setting, while also evaluating function of the instrument׳s rating scale as used among these learners, and psychometric responsiveness to change.MethodsThe SPICE-R instrument was administered to students at start and end of an interprofessional education course for first-year students from eight health care professions. The SPICE-R׳s rating scale functionality was characterized using Rasch Measurement. Without statistical adjustment, pre/post data were analyzed for total score change. With adjustment, a complex multiple regression modeled pre-SPICE-R, professional program, and gender on post-SPICE-R.ResultsA total of 553 students completed the pre-SPICE-R while 185 students also completed the post-SPICE-R. The 5-point rating scale did not function while a 4-point scale did. Overall, both SPICE-Rs were reliable (Cronbach׳s-alpha: 0.83 and 0.87). On unadjusted paired analysis, there was no mean improvement with course [pre-SPICE-R = 32.4 ± 4.4; post-SPICE-R = 31.8 ± 4.6; P = 0.10; and Cohen׳s d = −0.1 (trivial-effect-size)]. Separately, pharmacy improved modestly [pre-SPICE-R = 31.7 ± 5.7; post-SPICE-R = 33.5 ± 4.1; P = 0.12; and Cohen׳s d = 0.3 (small-effect-size)]. Regression showed that medicine and physician assistant did not have as much improvement as pharmacy (both <0.01), though students from all other professions responded less favorably than pharmacy.ImplicationsThis classroom-based setting among first-year students appeared to be outside the SPICE-R instrument׳s utility. This investigation showed that a 4-point rating scale allowed the SPICE-R to better function instead of a 5-point rating scale. Ultimately, the SPICE-R was reliable though showed only modest responsiveness to change after statistical adjustment.