National healthcare-associated infections surveillance programs – comparison between provincialand national case definitions in Canada

Abstract Background Provincial and national case definitions for healthcare-associated infections (HAI) differ. A key question is the extent to which these differences impact the estimated case proportions. Methods Analyses focused on two infections: Clostridioides difficile infection (CDI) and methicillin-resistant Staphylococcus aureus (MRSA). After identifying divergent elements in definitions, a parametric simulation was conducted using a multivariate Bernoulli model to generate elements, incorporating assumed relationships between them and necessary sample sizes. A reference simulation was performed using the original proportions of elements as reported in the literature. This reference served as a baseline for comparison with simulations in which the proportions of discordant elements varied between provincial and national case definitions. Results Nine discordant elements were identified for CDI and six for MRSA. Simulations showed comparability between provincial and national HAI proportions. The greatest differences between provincial and national proportions occurred for Newfoundland and Labrador (CDI) and Ontario (MRSA). These differences persisted even in the reference simulation. Conclusion Although adopting a standardized case definition would simplify comparisons, provincial case definitions are generally comparable to the national definition. Results should be interpreted with care, especially when estimating case proportions under extreme values for some characteristics, as the joint distributions we used for generating simulated datasets did not necessarily exist nor represent target marginal proportions or associations between elements in some scenarios. To improve comparability, hospitals should calculate both the number of patients with positive CDI or MRSA results and their proportion of all elements used in case definitions to have a better picture of their patient demographics. This information could further help the Canadian Nosocomial Infection Surveillance Program (CNISP) to describe comparability across provinces.