[PurelyApplied]

> > it is recommended not to ... compare values based on different models > i.e. R0 numbers are worthless. You can't compare these values between pathogens

That conclusion doesn't follow from the predicate. The model isn't a single simulation run. To mathematicians, the model is the entire framework. It's the SIR model, or the SIS model, or the SIER model, or my favorite model of immunity decay against pertussis... _those_ are the models. The virulence of the infection being modeled, resilience of the population, morbidity and mortality rates, rate and timing of population quarantine, vaccine adoption rate, vaccine efficacy, whether any of the aforementioned values are themselves functions of time... those are just parameters into your model. We absolutely compare the values between pathogens _within the same model_ when discussing the utility of the model.

But typically, the utility in reproduction ratio is as a point of comparison as you tune other parameters. This is part of why a comparison of reproduction ratio across models isn't recommended. It's not just that it's hard to draw meaning from the comparison, it's that the base assumptions of the models might be too different for the comparison to hold any meaning. The reproductive ratio between a continuous model with uniform population mixing is fundamentally different from the reproductive ratio approximated from discrete simulation. They may both broadly speak to "if everyone they saw was susceptible, how many people do you expect to get sick per sick person?" But what that ratio means is dictated by the context of the model.

A frequent utility in comparing ratios is to discuss intervention impact. One might write "Our model found that overall reproduction was reduced to a factor of 0.XYZ (comprehensive infectivity parameters a=nnn b=mmm c=ppp) when the population undertook vaccination schedule foo. Conversely, reproduction was reduced to a factor of 0.JKL for the same infection under vaccination schedule bar. See Figure 14 for complete population state levels over time."

That's all that is. It's a number that has some meaning in context, and one that is often removed from that context and unreasonably expected to retain its meaning. Averages don't mean as much if you don't know if your distribution is multimodal. It's the same thing - a summary stat that can give you a glimpse of the whole, but still just a summary stat.