Risk Management in the Clinical Pathology Laboratory: A Bayesian Network Approach

Abstract

The tests performed at clinical pathology laboratories play an important role in medical decision-making. This paper describes the application of a risk assessment framework to the clinical pathology laboratory of a hospital in the north of Portugal. The study involved a scoping literature review, focus groups and Bayesian Networks. Noisy-OR Canonical model was used to determine the conditional probabilities of Bayesian Networks. The approach can easily be adapted for other clinical laboratories. The study presents a new, simple and easy to understand alternative to the traditional Failure Mode Effects Analysis (FMEA) for risk management that: 1) facilitates the global visualization of the interdependencies between risk events; 2) obtains the likelihoods of the risks; and 3) allows simulations of risk mitigation strategies. The framework and its outcomes were well accepted by clinical pathology laboratory professionals since they considered it suitable to contextualize the risk network structure to the specific reality of the laboratory and the resulting model was useful to raise awareness about the latent risks at the laboratory. Moreover, some risks not referred in the literature were identified.