assessment

ABSTRACT:

Background:

Objective structured clinical examinations (OSCEs) are widely used to assess clinical competence, but their implementation assumes substantial institutional capacity: trained examiners, standardized patients, simulation space, station banks, administrative support, and psychometric expertise. These assumptions are problematic for small-cohort Caribbean medical schools. The accreditation and certification environment creates a concrete assessment mandate: CAAM-HP’s 2024 standards require a centralised assessment system using a variety of measures, including direct observation, to assess achievement of programme objectives including core clinical skills, behaviours and attitudes, while ECFMG Certification requires international medical graduates to meet medical school, examination, clinical skills, communication skills and credential requirements. These requirements do not mandate OSCEs as the only acceptable format, but they make directly observed, defensible clinical-skills assessment unavoidable.

Methods:

This article presents a narrative review and proposed framework for resource-constrained OSCE design in small-cohort Caribbean medical schools. The argument is framed, following established assessment-utility theory, as a budget-constrained problem in which each design choice carries both psychometric and economic consequences. Evidence from OSCE reliability reviews, standard-setting error studies, cost-effectiveness analyses, global rating versus checklist studies, peer role-play and standardized-patient comparisons, student standardized-patient feasibility work, AI-assisted station generation, Caribbean OSCE perception literature, and psychometric quality-control literature was synthesised.

Results:

The review supports a cost-aware “minimum viable OSCE” model with seven linked components: a station architecture centred on the reliability/cost frontier; explicit attention to standard-setting error in small cohorts; selective substitution of professional standardized patients with peer role-play or trained student standardized patients where validity risks are acceptable; use of global ratings alongside essential checklist items; examiner calibration using clear scoring guidance and anchor descriptors; AI-assisted station drafting under faculty governance; and iterative validation using post-examination quality-control methods rather than overinterpretation of single small-cohort psychometric estimates.

Conclusions:

The proposed framework does not lower standards; it prioritises the configuration most likely to preserve validity, reliability and decision defensibility per unit of scarce institutional resource. The framework requires local validation and should be developed through regional collaboration across Caribbean medical schools.