Abstract : Parasite interactions have been widely evidenced experimentally but field studies remain rare. Such studies are essential to detect interactions of interest and access (co)infection probabilities but face methodological obstacles. Confounding factors can create statistical associations, i.e. false parasite interactions. Among them, host age is a crucial covariate. It influences host exposition and susceptibility to many infections, and has a mechanical effect, older individuals being more at risk because of a longer exposure time. However, age is difficult to estimate in natural populations. Hence, one should be able to deal at least with its cumulative effect. Using a SI type dynamic model, we showed that the cumulative effect of age can generate false interactions theoretically (deterministic modeling) and with a real dataset of feline viruses (stochastic modeling). The risk to wrongly conclude to an association was maximal when parasites induced long-lasting antibodies and had similar forces of infection. We then proposed a method to correct for this effect (and for other potentially confounding shared risk factors) and made it available in a new R package, Interatrix. We also applied the correction to the feline viruses. It offers a way to account for an often neglected confounding factor and should help identifying parasite interactions in the field, a necessary step towards a better understanding of their mechanisms and consequences.
