Rater training for standardised assessment of Objective Structured Clinical Examinations in rural Tanzania.

Objectives: To describe a simulation-based rater training curriculum for Objective Structured Clinical Examinations (OSCEs) for clinician-based training for frontline staff caring for mothers and babies in rural Tanzania.

Background: Rater training for OSCE evaluation is widely embraced in high-income countries but not well described in low-income and middle-income countries. Helping Babies Breathe, Essential Care for Every Baby and Bleeding after Birth are standardised training programmes that encourage OSCE evaluations. Studies examining the reliability of assessments are rare.

Methods: Training of raters occurred over 3 days. Raters scored selected OSCEs role-played using standardised learners and low-fidelity mannikins, assigning proficiency levels a priori. Researchers used Zabar's criteria to critique rater agreement and mitigate measurement error during score review. Descriptive statistics, Fleiss' kappa and field notes were used to describe results.

Results: Six healthcare providers scored 42 training scenarios. There was moderate rater agreement across all OSCEs (κ=0.508). Kappa values increased with Helping Babies Breathe (κ=0.28-0.48) and Essential Care for Every Baby (κ=0.42-0.77) by day 3 of training, but not with Bleeding after Birth (κ=0.58-0.33). Raters identified average proficiency 50% of the time.

Conclusion: Our study shows that the in-country raters in this study had a hard time identifying average performance despite moderate rater agreement. Rater training is critical to ensure that the potential of training programmes translates to improved outcomes for mothers and babies; more research into the concepts and training for discernment of competence in this setting is necessary.