Psychometric analysis of cardiopulmonary resuscitation knowledge exams for nursing professionals

Abstract

Introduction

Brain injury following cardiac arrest is the primary disabling sequela, significantly impacting the quality of life of those affected.

Given this problem, the American Heart Association (AHA) recommends continuing education for healthcare professionals as a strategy to improve survival rates and reduce long-term consequences. To ensure knowledge acquisition during this training process, assessment tools are required, including exams with adequate psychometric properties to determine the quality and reliability of their results.

Therefore, the following question arises: Do the cardiopulmonary resuscitation (CPR) exams administered during nursing professional training exhibit adequate psychometric properties in terms of internal consistency, difficulty, and item discrimination?

Objective

To analyze the psychometric properties of three CPR knowledge exams administered to nursing professionals by evaluating the internal consistency, difficulty index, and item discrimination.

Methodology

A quantitative, observational, and cross-sectional study was conducted using the psychometric evaluation of three CPR protocols (basic, advanced adult, and advanced pediatric) based on AHA reference literature and with a multiple-choice response structure.

A non-probabilistic census sample of 101 nursing professionals was administered during a scheduled training course at a tertiary care hospital, maintaining the confidentiality of the respondents at all times.

Dichotomous data analysis assigned the number 1 to correct answers and 0 to incorrect answers. The difficulty index was calculated using the formula Df=AS+AI/N+N, and the item discrimination index for extreme groups was calculated using the formula DC2=AS/AS+AI.

Internal consistency was calculated using the Kuder-Richardson formula, which measures the proportion of correct responses, errors, and variance.

KR20 = (K/K-1)(1-(sum(pq)/variance))

The relationship between difficulty and discrimination was then analyzed using Pearson's coefficient of compensation, the coefficient of determination (R²), and simple linear regression.

Results

Basic CPR: 16.6% of items had an adequate difficulty index, 30% were relatively easy, and 53% were very easy, with 86.6% of items having a discrimination index greater than 0.5. KR20 = 0.68, r = (-)0.95, R² = 90%, calculated using the linear regression equation Y = (-)0.5761X + 1.0611.

Advanced Adult CPR: 3% of items had an adequate difficulty index, 45.4% were relatively easy, and 51.5% were very easy, with 100% of items having a discrimination index greater than 0.5. KR20 = 0.79, r = -0.88, R² of 77% with linear regression equation Y = -0.6473X + 1.1289.

Advanced Pediatric CPR: 2.7% of items had an adequate difficulty index, 2.7% were relatively easy, and 94.59% were very easy, with a discrimination of 89.18% greater than 0.5, KR20 = 0.74, r = -0.98, R² of 96% with linear regression equation Y = -0.6898X + 1.1824.

Conclusions

The three tests showed acceptable psychometric properties for assessing CPR knowledge. Although easy and very easy items predominated, these had an adequate level of discrimination with a strong inverse association with difficulty, which is characteristic of classical test theory. Furthermore, they had an acceptable KR20 internal consistency. This supports the usefulness of these instruments for assessing knowledge after training.