A metapopulation model for the spread of MRSA in correctional facilities

Abstract

The spread of methicillin-resistant strains of Staphylococcus aureus (MRSA) in health-care settings has become increasingly difficult to control and has since been able to spread in the general community. The prevalence of MRSA within the general public has caused outbreaks in groups of people in close quarters such as military barracks, gyms, daycare centres and correctional facilities. Correctional facilities are of particular importance for spreading MRSA, as inmates are often in close proximity and have limited access to hygienic products and clean clothing. Although these conditions are ideal for spreading MRSA, a recent study has suggested that recurrent epidemics are caused by the influx of colonized or infected individuals into the correctional facility. In this paper, we further investigate the effects of community dynamics on the spread of MRSA within the correctional facility and determine whether recidivism has a significant effect on disease dynamics. Using a simplified hotspot model ignoring disease dynamics within the correctional facility, as well as two metapopulation models, we demonstrate that outbreaks in correctional facilities can be driven by community dynamics even when spread between inmates is restricted. We also show that disease dynamics within the correctional facility and their effect on the outlying community may be ignored due to the smaller size of the incarcerated population. This will allow construction of simpler models that consider the effects of many MRSA hotspots interacting with the general community. It is suspected that the cumulative effects of hotspots for MRSA would have a stronger feedback effect in other community settings.

Keywords: Latin Hypercube Sampling; hotspots; mathematical model; metapopulation model; methicillin-resistant Staphylococcus aureus.

Fig. 1

Compartmental model of MRSA using an SIS model. Flow of the model is defined by system (1). All states and parameters are described in Table 1.

Fig. 2

Epidemic curves for varying values of p_S (probability of remaining susceptible) for (a) high and (b) low values of β. All other parameter values for simulations are listed in Table 1.

Fig. 3

Compartmental model for the two-patch SIS meta-population model. All states and parameters are listed in Table 1. Flow of the model is defined by system (3).

Fig. 4

PRCC sensitivity analysis with N = 1000 simulations for all parameters of system (3) using ranges from Table 2.

Fig. 5

Dependence of R₀ on the three most sensitive parameters (determined from the PRCCs) of system (3): (a) contact rate in the community, (b) recovery rate in the community, (c) contact rate in the correctional facility. Results were generated over N = 1000 simulations with parameter ranges shown in Table 2.

Fig. 6

Simulations of system (3) with varying values of β_J: (a) the number of infected individuals in the correctional facility with low disease transmission in the community (low β), and (b) the number of infected individuals in the correctional facility with high disease transmission in the community (high β). In (c), we see the number of infected individuals in the community with low disease transmission (β). All other parameter values are listed in Table 1.

Fig. 7

Compartmental model for MRSA transmission with susceptible and infectious classes for the general public, incarcerated and recidivist individuals. Flow of the model is defined by system (5).

Fig. 8

PRCC sensitivity analysis with N = 1000 simulations for all parameters of system (5) using ranges from Table 2.

Fig. 9

Dependence of R₀ on the most sensitive parameters (determined from the PRCCs) of system (5): (a) contact rate in the community, (b) contact rate in the correctional facility, (c) recovery rate in the community. Results generated over N = 1000 simulations with parameter ranges shown in Table 2.

Fig. 10

Simulations of system (5) of the number of infected incarcerated individuals with varying levels of (a) the probability of recidivism (p_R), (b) the general incarceration rate (ψ) and (c) the recidivist incarceration rate (ψ_R). All other parameter values are listed in Table 1.