Development and validation of a multiple-choice test for head and neck ultrasound certification

Abstract

Purpose: This prospective trial aimed to develop and gather validity evidence for a theoretical test in head and neck ultrasound (HNUS).

Methods: Seven HNUS experts from Europe, North America, and Asia participated in a Delphi study to reach consensus on multiple choice test (MCT) items. Novices (n=56) and experienced HNUS operators (n=22) were then invited to take the full MCT. Based on their answers, an item-response analysis selected the MCT items with the best performance. Generalizability theory determined the number of MCT items sufficient for certification. An ANOVA test examined the MCT's ability to distinguish novices from experts and contrasting groups' standard setting was used to establish a cut-off test score. A group of physicians (n=23) was tested at a formal ultrasound course, and pass-fail consequences for the final test were evaluated.

Results: Over three Delphi rounds, 64 items were revised and 21 were excluded, yielding 106 items. The item-response analysis found nine items with low discrimination that were excluded based on the MCQ answers from 78 novices and experienced physicians who had taken the test. The final 97 test items had a high internal consistency reliability of 0.97, and an MCT with 15 items was found sufficient for certification purposes. The MCT could significantly discriminate between novices (mean 51.1, SD 13.8) and experienced participants (mean 92.0, SD 3.1), p < 0.001. A pass-fail score of 83 was established. At a formal introduction ultrasound course, 57% of the participants passed the MCT at the established pass-fail score.

Conclusion: The developed MCT for HNUS, based on international expert consensus, has multiple sources of validity evidence to support its use as part of a thorough certification process.

Fig. 1

Flowchart of the Delphi-like process. Flowchart of the Delphi-like process with excluded, revised, new, and included test items. 97 items were included after item response analysis based on data from the experimental study with novices and experienced HNUS operators

Fig. 2

Multiple-choice test score distribution of novices and experienced. Multiple-choice test (MCT) score(x-axis) distribution of novices(red) and experienced(blue). The pass-fail score was established by the contrasting groups’ method. The intersect(green) between the score curves of the novices and the experienced visualizes the pass-fail score of 83

Fig. 3

Generalizability coefficient for different number of test items. Generalizability coefficient for different number of test items, achieved through a Decision study. The red line indicates that sampling 15 test items achieves a Generalizability coefficient above 0.8 which suffices for certification purposes

Fig. 4

Differences in multiple-choice test scores between groups of head and neck ultrasound operators. Boxplot visualizing the differences in multiple-choice test scores between groups of novice head and neck ultrasound (HNUS) operators, physician course participants at a formal HNUS course, and experienced HNUS operators. Highlighted in green is the Pass-fail score of 83. MCT: Multiple-choice test