Development of the Diabetic Wound Assessment Learning Tool (DiWALT) and validity evidence.

ABSTRACT

OBJECTIVE: Diabetic foot wounds account for up to one-third of diabetes-related health care expenditure and are the greatest cause of extremity amputation in Canada. Physicians encounter patients with such wounds in all specialties, particularly as generalists in medical wards and emergency departments. However, there is a dearth of literature on the optimal way to teach and to assess the management of these patients. Given the importance of assessment for learning in the shift toward competency-based medical education, we aimed to develop an assessment tool and to build validity evidence for its use in this context. METHODS: A consensus process involving nine Canadian experts in diabetic wound management was used to develop the Diabetic Wound Assessment Learning Tool (DiWALT) items and two 10-minute simulation-based testing scenarios. The simulators used were modified from commercially available models to serve the testing scenarios. Validity evidence for the DiWALT was subsequently evaluated by assessing 24 physician participants' performance during the two scenarios. All participants were novices (<50 cases managed). Two assessors independently rated participants using the DiWALT. Evidence was organized using Kane's validity framework and included Cronbach α for interitem consistency as well as test-retest and inter-rater reliability using the intra-class correlation coefficient (ICC). RESULTS: Cronbach α was 0.92, implying high internal consistency. Test-retest reliability was also excellent with ICC of 0.89 (confidence interval [CI], 0.76-0.95) for single measures and ICC of 0.94 (CI, 0.86-0.98) for average measures. Inter-rater reliability was fair for single measures with ICC of 0.68 (CI, 0.65-0.71) and good for average measures with ICC of 0.81 (CI, 0.79-0.83). CONCLUSIONS: These results demonstrate that the DiWALT consistently and reliably evaluates competence in diabetic wound management during simulated cases using a small, homogeneous sample of physicians. Further work is necessary to quantify sources of error in the assessment scores, to establish validity evidence when it is used to assess larger and more heterogeneous participants, and to identify how well the DiWALT differentiates between different experience levels.