The evolution of post-infection mortality

Abstract

COVID-19 infections have underlined that there can be substantial impacts on health after recovery, including elevated mortality. While such post-infection mortality (PIM) is clearly widespread, we do not yet have any understanding of its evolutionary dynamics. To address this gap, we use an eco-evolutionary model to determine conditions where PIM is evolutionarily favoured. Importantly, from a pathogen perspective, there are two potential 'resources': never-infected susceptibles and previously infected susceptibles (provided some reinfection is possible), and PIM only occurs in the latter. A key insight is that unlike classic virulence (i.e. during-infection mortality, DIM) PIM is neutral and not selected against in the absence of other trade-offs. However, PIM modulates characteristics of endemicity, and may also vary with other pathogen-specific components. If PIM is only correlated with transmission, recovery or DIM, it simply acts to modulate their impacts on the evolutionary outcome. On the other hand, if PIM trades off with the relative susceptibility to reinfection, there are important evolutionary implications that contrast with DIM. We find settings where a susceptibility-mortality trade-off (i.e. an increase in mortality leads to higher relative susceptibility to reinfection) can select against DIM but favour PIM. This provides a potential explanation for the ubiquity of PIM. Overall, our work illustrates that PIM can readily evolve in certain settings and highlights the importance of considering different sources of mortality.

Keywords: eco-evolutionary model; pathogen evolution; post-infection mortality.

Figure 1.

Schematic of the eco-evolutionary model for the evolution of PIM, with the epidemiological model (for the endemic pathogen) as in [6].

Figure 2.

Conditions for the evolution of PIM as a function of without-PIM relative susceptibility to reinfection, assuming transmission increases with PIM. For values of ${\epsilon }^{\prime }[0]$ above this threshold $\tau$, mutants with PIM can invade. On the other hand, for values of ${\epsilon }^{\prime }[0]$ below it, no PIM is an ESS. Other parameter values are $\mathrm{\Lambda }=\mu =\frac{1}{50(52)}$ and $\gamma =1$.

Figure 3.

Condition for the evolution of PIM (i.e. the value of the threshold $\tau$) as a function of the transmission rate with no PIM. In particular, for ${\epsilon }^{\prime }[0]>\tau$ such that mutants with PIM can invade an endemic pathogen with no PIM, whereas ${\epsilon }^{\prime }[0]<\tau$ implies no PIM is an ESS. The rows depict values of ${\beta }^{\prime }[0]$, i.e. the ‘initial' strength of the transmission–mortality trade-off, and the columns depict values of DIM. Note that other parameter values are as in figure 2.

Figure 4.

Comparison of the evolution of DIM and PIM under a relationship between relative susceptibility to reinfection and mortality, when transmission increases with mortality (i.e. ${\beta }^{\prime }[0]>0$). The rows denote different scenarios for the change in transmission with mortality, and the columns are different scenarios of baseline relative susceptibility to reinfection without mortality. As before, for values of ${\epsilon }^{\prime }[0]$ above each curve, mutants with mortality can invade, whereas no mortality is an ESS for values of ${\epsilon }^{\prime }[0]$ beneath them. Below the dashed red line at ${\epsilon }^{\prime }[0]=0$, a mutant with either PIM or DIM pays a cost as it elicits lower relative susceptibility to reinfection. On the other hand, above this line, a mutant with either PIM or DIM elicits higher relative susceptibility to reinfection. Note that for PIM evolution, we set ${\alpha }_I=0$, and for DIM evolution, we set ${\alpha }_S=0$, and other parameter values are as in figure 2.