Dynamics and control of Ebola virus transmission in Montserrado, Liberia: a mathematical modelling analysis

Abstract

Background: A substantial scale-up in public health response is needed to control the unprecedented Ebola virus disease (EVD) epidemic in west Africa. Current international commitments seek to expand intervention capacity in three areas: new EVD treatment centres, case ascertainment through contact tracing, and household protective kit allocation. We aimed to assess how these interventions could be applied individually and in combination to avert future EVD cases and deaths.

Methods: We developed a transmission model of Ebola virus that we fitted to reported EVD cases and deaths in Montserrado County, Liberia. We used this model to assess the effectiveness of expanding EVD treatment centres, increasing case ascertainment, and allocating protective kits for controlling the outbreak in Montserrado. We varied the efficacy of protective kits from 10% to 50%. We compared intervention initiation on Oct 15, 2014, Oct 31, 2014, and Nov 15, 2014. The status quo intervention was defined in terms of case ascertainment and capacity of EVD treatment centres on Sept 23, 2014, and all behaviour and contact patterns relevant to transmission as they were occurring at that time. The primary outcome measure was the expected number of cases averted by Dec 15, 2014.

Findings: We estimated the basic reproductive number for EVD in Montserrado to be 2·49 (95% CI 2·38-2·60). We expect that allocating 4800 additional beds at EVD treatment centres and increasing case ascertainment five-fold in November, 2014, can avert 77 312 (95% CI 68 400-85 870) cases of EVD relative to the status quo by Dec 15, 2014. Complementing these measures with protective kit allocation raises the expectation as high as 97 940 (90 096-105 606) EVD cases. If deployed by Oct 15, 2014, equivalent interventions would have been expected to avert 137 432 (129 736-145 874) cases of EVD. If delayed to Nov 15, 2014, we expect the interventions will at best avert 53 957 (46 963-60 490) EVD cases.

Interpretation: The number of beds at EVD treatment centres needed to effectively control EVD in Montserrado substantially exceeds the 1700 pledged by the USA to west Africa. Accelerated case ascertainment is needed to maximise effectiveness of expanding the capacity of EVD treatment centres. Distributing protective kits can further augment prevention of EVD, but it is not an adequate stand-alone measure for controlling the outbreak. Our findings highlight the rapidly closing window of opportunity for controlling the outbreak and averting a catastrophic toll of EVD cases and deaths.

Funding: US National Institutes of Health.

Figure 1. Model calibration

A: Reported and model-predicted cases, with black points showing observed data used in model calibration, blue points showing observations outside the fitting period, and shaded areas showing 95% credible intervals around model predictions; B: Reported deaths, as in (A). Credible intervals are computed based on 5,000 simulations. The superimposed line indicates median predicted cases and deaths.

Figure 2. Impact of adding new Ebola treatment centers and increasing case ascertainment

Intervention effects for programs initiated 31 and 15 October, respectively, considering expansions in ETCs and case ascertainment. Roman numerals I, II, and III describe ETC construction at the rates of 3, 6, and 12 ETCs per week, respectively. Ascertainment labels are defined in Table 1. We expand the figure in the online supplemental materials to illustrate effects under all modeled ETC deployment schedules.

Figure 3. Impact of augmenting interventions with protective kit allocation

A: Effectiveness of programs beginning 31 October, considering all possible expansions in ETCs and case ascertainment, and varying efficacy levels for protective kits. Roman numerals II and III describe ETC construction at the rates of 6 and 12 ETCs per week, respectively. Ascertainment labels are defined in Table 1. We expand the figure in the online supplemental materials to illustrate effects under all modeled ETC deployment schedules. B: Effects of equivalent interventions initiated 15 October. C: Effects of equivalent interventions initiated 15 November.