Fig 1.
Population and health care system diversity in Southeast Asia and the Western Pacific.
(A) Population density ranges from less than 1 person per square kilometre in parts of Papua New Guinea and Indonesia to over 6,800 people per square kilometre in Singapore. Data obtained from Center for International Earth Science Information Network (CIESIN), Columbia University (http://sedac.ciesin.columbia.edu/gpw). (B) Number of health care workers per thousand population ranges from 1.2 in Cambodia to 11.5 in Brunei Darussalam. Data obtained from Global Health Observatory data repository (http://apps.who.int/gho/data/).
Fig 2.
The natural history of infection comprises susceptibility (S), exposure (E) at a rate determined by the force of infection β and the current prevalence of infectious individuals and unburied dead bodies. Exposed individuals progress to mild infectiousness prior to developing symptoms (I0) at rate σ, and symptomatic infection (I) at rate γ0, followed by either death (D) or recovery (R) at rate γ1. The proportion of infections leading to death or recovery is informed by estimates of the case fatality ratio (CFR). Dead bodies remain infectious prior to burial (B) at rate τ. Ascertainment of cases (with probability pasc) allows symptomatic individuals to be hospitalised in isolation wards (H), which reduces their contribution to transmission and increases their probability of recovery. Full equations describing the model are provided in Model description in S1 Methods.
Fig 3.
Time series of disease prevalence for three example simulations.
These three examples demonstrate the three possible categories of epidemic outcome: stochastic fade-out, a controlled outbreak, and an uncontrolled outbreak.
Fig 4.
The classification of epidemic outcomes.
In the absence of a health care response, EVD importation may result in zero or very few secondary cases (green, left panel) or may result in a large, uncontrolled epidemic (red, left panel). A health care response may mitigate some of these uncontrolled epidemics by greatly reducing their final size (yellow, right panel).
Fig 5.
Effect of local response capacity.
Epidemic outcomes as a function of case ascertainment (x-axis), subject to the first detected case (FD) and the health care system capacity (small: 0.1% of population are HCWs; medium: 0.2% of population are HCWs; large: 0.3% of population are HCWs; 1:15 ratio of HCWs to bed capacity, 1:50 ratio of HCWs to contact-tracing capacity.
Fig 6.
Effect of increased case ascertainment.
Starting with a baseline case ascertainment of 20%, this figure shows the effect of boosting ascertainment to 100% (i.e., perfect detection) at different times after the first detected case. This clearly shows the simultaneous importance of early detection and high ascertainment; provision of one is not a substitute for lack of the other.
Fig 7.
Effect of improved health care system capacity.
Starting with a baseline case ascertainment of 80% and a small health care system (0.1% of population are HCWs, 1:15 and 1:50 ratio of bed and contact-tracing capacities to HCWs, as above), this figure shows the effect of doubling both the health care capacity and the health care workforce at different times after the first detected case. This increase in capacity represents the transition from a small health care system to a medium health care system, as defined in Fig 5. This clearly shows the importance of early detection; the time delay in delivering the additional capacity is less important over this timescale of 0–8 weeks, because the existing health care system is capable of accommodating patients in the early stage of the outbreak.
Fig 8.
Effect of reducing force of infection from dead bodies.
Starting with a baseline case ascertainment of 70% and a small health care system (0.1% of population are HCWs, 1:15 and 1:50 ratio of bed and contact-tracing capacities to HCWs, as above), this figure shows the effect of reducing the force of infection from dead bodies (βD) by 25%. This represents a sociocultural intervention that changes burial practices, as conducted in previous Ebola outbreaks and the temporary storage of dead bodies in the 2009 cholera outbreak in Papua New Guinea. Reducing the force of infection (whether directly or indirectly) will decrease the epidemic burden, but the likelihood of controlling the epidemic is strongly affected by the speed with which such an intervention can be delivered. Success requires both early detection and a rapid response from the community.
Fig 9.
Effect of reducing force of infection from community transmission.
Starting with a baseline case ascertainment of 70% and a small health care system (0.1% of population are HCWs, 1:15 and 1:50 ratio of bed and contact-tracing capacities to HCWs, as above), this figure shows the effect of reducing the force of infection from dead bodies (βD) and in the community (βI) by 25%. This represents a sociocultural intervention that changes both burial practices and social mixing in the community. Similar to Fig 7, success of such an intervention relies upon early detection and a rapid response from the community.
Table 1.
The administrative regions of Papua New Guinea.
We separate Port Moresby (the largest city and national capital) from the rest of the Southern (“Papua”) region on the grounds that Port Moresby comprises a much more urbanised population than the rest of the region.
Fig 10.
Effect of behavioural interventions in Port Moresby region.
The effect of reducing the force of infection in the community and/or from dead bodies by 25%, in the urban population of Port Moresby where community transmission is high. When case ascertainment is 40% or higher, reducing transmission in both settings has a synergistic effect on outbreak control.
Fig 11.
Effect of behavioural interventions in Southern region.
The effect of reducing the force of infection in the community and/or from dead bodies by 25%, in the rural population of the Southern region where community transmission is low. When transmission is reduced in both settings, the overall force of infection is sufficiently low that uncontrolled outbreaks never occur.