This morning I chaired the session “Which mathematical models for antimicrobial resistance?” at ECCMID2017. Three excellent talks, making one thing more crystal clear (to me, at least) than before. Antibiotic resistance epidemiology should be considered as a “complex system”. What does that mean?
Why complex? Some of the determinants relevant for its epidemiology: the gene, the plasmid, the bacterium, the other bacteria, the microbiome, the host (human or animal), the surroundings of that host, the network in which the host resides, the things the host receives (such as antibiotics), the things the host does (that may lead to transmission) and all we do to prevent (or facilitate) transmission. And even this list is far from exhaustive.
The global financial system and our climate are examples of complex systems, which are characterized by large unexpected events, also called tipping points. Think of the economic meltdown in October 2008. If you search google for AMR & tipping point you’ll find >80K hits: all predicting tipping points soon,…. over many years. Apparently it is difficult to predict the occurrence of a tipping point, which is characteristic for a complex system.
How can we ever understand our data if all of this (& more) matters. The honest answer is we hardly can. Making things more complex: most (if not all) of our data are imperfect. To me, this emphasizes the need of theoretical frameworks to help us understand what we (think to) see in our data. The unexpected and rapid emergence of plasmid-based carbapenemases in the last decade – to me – was a tipping point (not predicted by anyone), that still remains to be explained. We may need new theories and views from outside our own boxes for that.