Mechanistic Models of Virus-Bacteria Co-Infections in Humans: A Systematic Review of Methods and Assumptions

Viral-bacterial co-infections can amplify disease severity through complex biological mechanisms. Mathematical models are critical tools for understanding these threats, but it is unclear how well they capture the underlying biology. This systematic review addresses a central question: to what extent does the current generation of models mechanistically represent co-infections, or do the mathematical assumptions underlying these models adequately represent the known biological mechanisms?

Following PRISMA guidelines, we systematically reviewed the literature on mechanistic models of human virus-bacteria co-infections. A systematic search of articles on the scientific literature repositories PubMed, Scopus, and Dimensions was conducted and data on study objectives, model structure, assumptions about biological interactions (e.g., susceptibility, mortality), control measures (if evaluated), and the empirical sources used for key parameters were extracted.

We identified 72 studies for inclusion in this analysis. The reviewed models are consistently built on the established premise that co-infection alters disease severity and host susceptibility. However, we found they incorporate these dynamics primarily through high-level mathematical shortcuts, such as applying static "multiplicative factors" to transmission or progression rates. Our quantitative analysis also revealed questionable approaches; for example, 79% (57) of these studies relied on non-empirical sources (assumed or borrowed values) for parameter values including interaction parameters (e.g., increased susceptibility to a secondary pathogen following primary infection, or elevated mortality rates in co-infected individuals).

An apparently unjustified practice exists in co-infection modeling, where complex biological processes are simplified to fixed numerical assumptions, often without empirical support. This practice limits the predictive reliability of current models. We identify an urgent need for data-driven parameterization and interdisciplinary collaboration to bridge the gap between biological complexity and modeling practice, thereby enhancing the public health relevance of co-infection modeling.