Effective programmatic change at the GME level requires deliberate program implementation and rigorous program evaluation. In this study, we identified critical elements of the Stanford Neurology and Surgery Communication Coaching Program, considering program inputs, outputs, outcomes, and evaluation metrics, all within the context of our unique environment and individual stakeholder needs. We considered the implementation and evaluation of the coaching program in parallel by combining a commonly used implementation science framework, CFIR, with a common program evaluation method, the logic model.
While program implementation and evaluation are distinct entities, the two go hand-in-hand and should ultimately build on each other in a cyclical fashion to make programs more effective over time and as community and stakeholder needs change.9 Mapping the key themes identified in our analysis onto a logic model offered a more holistic description of all critical elements of the intervention and exposed areas where the program may not sufficiently meet implementation goals, and even offered suggestions for improvement. Themes that emerged from only one specific stakeholder group or one portion of the logic model may not present the full story of the program; however, in our study, multiple different perspectives contributed to the comprehensive nature of this evaluation, an essential feature of program evaluation.9
One of the advantages of using the logic model in this way was its emphasis on the relationship of other domains to the program’s context or environment.13 Our qualitative findings demonstrated a shared perception of a gap in communication education and an emphasis on the importance of communication from a patient perspective. These themes served as a foundation for program implementation, providing common ground for all stakeholder groups. Our findings were consistent with the known importance of a needs assessment in identifying programmatic priorities and specifically seeking to address the needs of the community.20 The analysis also demonstrated extensive early program investment in time, funding, resources, and personnel. Although the inputs were robust, the evaluation revealed a wider range of participant experiences related to perceived program value, suggesting key differences in the degree of perceived benefit, engagement, and experience in the program. While the linear nature of a logic model has been cited as one of its limitations,21 clear links between different elements of the model help illuminate discrepancies. Thus, the inequalities between inputs and outputs highlight a potential area for programmatic improvement to better align participant experiences with program objectives and inputs.
The findings of our study also exposed a unique interplay between definitions of program success, outcomes, and challenges with evaluation. The highly varied descriptions of program success suggested distinct perceptions and experiences both by individual and stakeholder groups. This also introduced potential unintended or unexpected consequences of the program, which are essential to consider in any program evaluation.13 Although the foundation for the program was firmly rooted in patient-centeredness and a gap in communication education, program participants described successful outcomes much more broadly – at the level of the patient, the resident, the faculty, and even the culture of the institution. We found that participants recognized outcomes and evaluation strategies at multiple Kirkpatrick levels and for various stakeholder groups (i.e., resident perceptions at Level 1, knowledge of communication strategies at Level 2, better non-coach faculty utilization of the coaching program at Level 3, and patient outcomes at Level 4).22 Stakeholders also recognized challenges in the measurement of success according to established metrics, such as patient satisfaction scores and resident milestones. These findings ultimately informed a framework from which to consider interwoven concepts of program success, outcomes, and challenges with evaluation to align in medical education interventions (see Fig. 3).
The challenges with existing mechanisms of evaluation further exposed the invaluable nature of the qualitative approach to participant-described outcomes. While it is understandably challenging to see an observed change in patient-level outcomes data, for instance, it brings depth and meaning to the program when participants describe their experience with change, such as the way the intervention has impacted their patient-level communication or interactions with their peers. The perceived definitions of success also indicate that there is room to consider other types of program evaluation metrics, such as perceptions of non-coach faculty, feedback culture, and other patient-level data.
There are several study limitations that warrant further discussion. While 35 separate interviews were conducted, it is possible that some concepts and themes were not represented in this cohort or that findings may be specific to our institution. Participants also had varying degrees of involvement in the program; thus, their experiences may be specific to only some domains of the logic model. However, within each group of interview participants, the researchers felt that thematic saturation was adequately achieved. As well, we believe that our comprehensive program evaluation would be incomplete without input from key stakeholders across departments who can speak to different aspects of the program and various domains of the logic model. Additionally, while patients are a key stakeholder group in the program implementation, they were not included in the qualitative study. Patients have varying levels of contact with the coaching program and many interactions beyond those directly involved in the program. Thus, we determined that it would be too difficult to parse out the impact of the communication coaching program at the level of individual patients.