400 Section - The Contribution of Team Functioning to Patient-Centered Measurement: An Exploratory Study Using Random Forests Methods

RF models were employed to examine the relative contribution of a range of predictors. Model fit was limited, with relatively low R² values (.11 to .15), indicating that the overall explanatory power of the models was modest. Variable importance metrics (percent increase in Mean Statistical Error [%IncMSE]  provided insight into the relative influence of individual predictors. The strongest contributors across the RAS domains were TF measures of Use of Quality Information (%IncMSE, 6.87 -11.68) and Satisfaction as Team Member (%IncMSE,  6.04 - 8.55) while three RA measures (Support, Involvement, and Patient-centered) were also high on several RAS domains.   The findings suggest that broad provider measures of TF and selected RA attributes play larger roles in prediction of patient experiences than more detailed, process-centered measures. 

 


Conclusion:

While this research design does not allow for correlational nor causative interpretations, the study acts as a probe into the potential uses of social environmental measures in the development of PCM.  Provider measures predict patients’ experiences, though with only modest explanatory power, across domains of the RAS and support the utility of incorporating RA and TF constructs into PCM consistent with the American Institutes of Research principles of Patient-Driven, Holistic, Transparent, and Co-creation – all of which are in line with basic tenets of team-based rehabilitation services.