A metapopulation model to assess the capacity of spread of meticillin-resistant Staphylococcus aureus ST398 in humans

Abstract

The emergence of the livestock-associated clone of meticillin-resistant Staphylococcus aureus (MRSA) ST398 is a serious public health issue throughout Europe. In The Netherlands a stringent 'search-and-destroy' policy has been adopted, keeping low the level of MRSA prevalence. However, reports have recently emerged of transmission events between humans showing no links to livestock, contradicting belief that MRSA ST398 is poorly transmissible in humans. The question regarding the transmissibility of MRSA ST398 in humans therefore remains of great interest. Here, we investigated the capacity of MRSA ST398 to spread into an entirely susceptible human population subject to the effect of a single MRSA-positive commercial pig farm. Using a stochastic, discrete-time metapopulation model, we explored the effect of varying both the probability of persistent carriage and that of acquiring MRSA due to contact with pigs on the transmission dynamics of MRSA ST398 in humans. In particular, we assessed the value and key determinants of the basic reproduction ratio (R(0)) for MRSA ST398. Simulations showed that the presence of recurrent exposures with pigs in risky populations allows MRSA ST398 to persist in the metapopulation and transmission events to occur beyond the farming community, even when the probability of persistent carriage is low. We further showed that persistent carriage should occur in less than 10% of the time for MRSA ST398 to conserve epidemiological characteristics similar to what has been previously reported. These results indicate that implementing control policy that only targets human carriers may not be sufficient to control MRSA ST398 in the community if it remains in pigs. We argue that farm-level control measures should be implemented if an eradication programme is to be considered.

Figure 1. Model structure.

(A) Representation of the structure of the metapopulation model for transmission of MRSA ST398 from a single pig farm to the community. The metapopulation comprises eight populations (squares) of various size, species and exposure to the pigs (grey circle) present in a single pig farm. Among these populations, two involve dogs and cats (CA: companion animals, FCA: companion animals in the pig farm). The remaining six populations involve humans with varying level of exposures to LA-MRSA and their families (F: workers of the pig farm, VP: pig veterinarians, T: transporters, SHW: slaughterhouse workers in dirty zone, V: small animal veterinarians, GH: general human population). Solid arrows represent contacts between individuals of different populations. Dotted arrows represent the exposure to pigs, with their width indicating the level of this exposure. (B) Schematic for transmission dynamics of MRSA ST398 between carriage states. Individuals are grouped into three carriage states: non-carriers and those susceptible to be colonised by LA-MRSA, showing either transient or persistent carriage. Straight arrows represent transition between carriage states with solid lines representing transmission and dotted lines representing natural clearance of carriage (see Table 1 for parameters values and definition).

Figure 2. Contact matrix.

Contact matrix showing the average daily number of contacts p_i,j between individuals of the metapopulation (see Figure 1 for details).

Figure 3. Influence of parameter variations on R ₀.

(A) Line plot showing the evolution of the critical values of the transmission rate (β*) that would reverse the net reproduction number (R ₀) for increasing values of duration of persistent carriage (1/µ_P) and for each considered scenario of the probability of persistent carriage q. (B) Line plot showing the deviation in R ₀ estimates attributable to changes in the number of contacts within the SHW (p _SHW,SHW) and/or the GH (p _GH,GH) populations. Insert in A shows the deviation in R ₀ estimates attributable to changes in β. Large dots in the inset of A and in B mark the required changes in parameters value for R ₀ = 1 for each considered scenario of the probability of persistent carriage q. Triangle dots in A mark baseline values for parameters.

Figure 4. Weekly evolution of the probability of MRSA ST398 occurring in the metapopulation.

Weekly evolution of the probability of MRSA ST398 presence in the metapopulation after: (A) a single persistent acquisition in humans working in the pig farm, and (B) multiple incursions in humans due to frequent contacts with pigs. Solid line represents outputs for each considered scenario of the probability of persistent carriage q. Curves in (B) were computed with the probability of acquiring LA-MRSA from contact with pigs δ = δ ₀.

Figure 5. Temporal evolution of the prevalence of MRSA ST398 carriage in humans.

Temporal evolution of the prevalence of LA-MRSA carriage in humans when recurrent contacts with pigs occur. Curves were computed for each considered scenario of the probability of persistent carriage q and probability acquisition from pigs δ: (A) q = 0.05, (B) q = 0.10, (C) q = 0.20 and (D) q = 0.35.

Figure 6. Role of exposure to pigs on the evolution of MRSA ST398 carriage, δ = δ ₀.

Role of exposure to pigs on the evolution of MRSA ST398 carriage when LA-MRSA is recurrently acquired of from pigs with a probability δ = δ ₀ (similar for δ = 5δ ₀, see Figure S1). (A) The proportion of exposed carriers among all carriers. (B) Ratio of the incidence risk of LA-MRSA carriage in pig-exposed individuals with that in individuals with no direct contact with pigs. (C) The proportion of LA-MRSA carriage in the exposed populations attributable to exposure with pigs (i.e. the attributable fraction). (D) The proportion of LA-MRSA carriage in the population that is attributable to exposure to pigs (i.e. the population attributable fraction). The horizon dashed grey lines in (B) represents the risk ratio = 1.

Figure 7. Influence of q_nc/q_c on MRSA ST398 carriage.

(A) Evolution of the endemic prevalence when varying the probability of persistent carriage for individuals that show no contacts with live pigs, q_nc, while fixing the probability of persistent carriage in individuals that show contacts with livestock q_c, such as q_c = q. (B) Evolution of the proportion of carriers that are persistent among the unexposed populations as a function of q_nc/q_c. Note that, for clarity, no confidence intervals for the scenario q = 0.05 were plotted in (A).