Risk of Spread

World Health Organization (2026) “Disease Outbreak News, 14 August 2026”: "Since the previous Disease Outbreak News was published on 1 August 2026, additional confirmed cases and deaths of BVD have been only reported in the Democratic Republic of the Congo. Cumulatively as of 12 August 2026, 4686 confirmed cases have been reported: 4665 in the Democratic Republic of the Congo (including two cases diagnosed in the Democratic Republic of the Congo and subsequently treated in Germany), 20 in Uganda and one in France. "

Epistorm (2026) “Methods Overview”: “Following the 2 June 2026 reopening of the Bunia International Airport, this update factors increased domestic and regional mobility from the affected area into the model. The full footprint covers Ituri, North Kivu, South Kivu, and Uganda, reflecting the laboratory-confirmed travel-related case in Bukavu (South Kivu) and the 19 laboratory-confirmed cases in Uganda.”

Fanelli et al. (2026a) “INTRODUCTION”: “As of 4 July 2026, the outbreak — previously declared a Public Health Emergency of International Concern (PHEIC) — has reached 1,561 confirmed cases across the provinces of Ituri, North Kivu, and South Kivu in eastern DRC, and 20 confirmed cases in Uganda.”

Chamla et al. (2026) “Summary”: “Cross-border spillover remained material: Uganda 94.2% importation probability (19 confirmed cases as of 4 June 2026, including five healthcare-worker infections and two deaths); South Sudan 69.3%; Rwanda 8.6%; and Burundi 2.0%.”

Fanelli et al. (2026a) “RESULTS”: “Across the seven scenarios tested, the ranking of high-risk countries remained largely unchanged and closely matched that estimated under the current epidemic spread. Specifically, South Sudan, Rwanda, Kenya, Tanzania, and Zambia remained among the 5 countries at highest risk of importation (Fig. 4a).”

Kinoshita, Suzuki, & Yoneoka (2026) “Abstract”: “Effective-distance analyses identified major international hub countries, including Belgium, France, South Africa, Kenya, and the United Arab Emirates, as higher-probability gateways until the end of June, 2026.”

Fanelli et al. (2026b) “Abstract”: “Relative exposure patterns remained largely unchanged under localized transmission in eastern DRC and border-spillover scenarios. Expansion into South Sudan generated a first structural increase in importation pressure to Europe through the connectivity associated with Juba, while hypothetical amplification in Kampala, Kigali, and Kinshasa substantially increased importation pressure and reshaped exposure patterns across Europe.”

Okingo, Iroungou, & Valdano (2026) “Body”: “These results separate importation from amplification risk: health zones near the affected area are more likely to receive cases, but not necessarily to sustain transmission. Our estimates therefore differ from Ref. 5 by identifying areas of concern farther from the affected zone.”

Okingo, Iroungou, & Valdano (2026) “Body”: “Our results warrant preparedness in health zones with high sustained-transmission potential but lower introduction risk, particularly in Kinshasa and other urban centers, despite their distance from the outbreak. This is relevant because several high-amplification areas are in Kinshasa, Kasaï, and Tshopo, where preparedness may be more feasible than in parts of eastern DRC despite heterogeneous healthcare access, accessibility constraints, and low-to-moderate insecurity outside Kinshasa.”

Fanelli et al. (2026a) “DISCUSSION”: “We also assumed homogeneous travel behaviour and exposure risk across travellers, although age, occupation, socioeconomic conditions, purpose of travel, and individual exposure may modify infection risk, as illustrated by the imported case in France involving a medical doctor working in the response.”

Bouhentala & Kadir (2026) “4. Discussion”: “The main limitation is dependence on public aggregate data. Reporting dates do not represent symptom onset, and the analysis could not estimate time-varying reproduction numbers, serial intervals, onset-to-isolation delays, household secondary-attack rates, health-zone-specific growth, or intervention effects.”

Verheyden and Mudogo (2026) “4 Discussion”: “The outcome modelled throughout is first reported confirmed case, not the unobserved date of viral introduction. These differ to the extent that introduction, local transmission, detection, testing access, and reporting capacity differ across zones, and we cannot separate these mechanisms with the data available. This matters specifically for the health-facility-density covariate, whose positive association with hazard may partly or wholly reflect greater detection capacity in better-served zones rather than, or in addition to, genuinely greater introduction risk.”

van Zandvoort et al. (2026) “Low overall risk of exportation”: “As a crude measure of risk, we calculated for Ebola disease outbreaks since the year 2000 the number of cases occurring outside Africa per 1,000 reported cases in the originating epidemic.”

Chamla et al. (2026) “Discussion”: “Future iterations should integrate line lists, contact-tracing data and dynamic mobility estimates.”

Bouhentala & Kadir (2026) “4. Discussion”: “Reporting dates do not represent symptom onset, and the analysis could not estimate time-varying reproduction numbers, serial intervals, onset-to-isolation delays, household secondary-attack rates, health-zone-specific growth, or intervention effects.”

Fanelli et al. (2026a) “DISCUSSION”: “First, reliable data on cross-border land mobility in Africa remain scarce.”

Walekhwa et al. (2026) “4. Discussion”: “The DTM data cover only a 10-day window and eight border crossing points; movements through unofficial crossings or after the outbreak escalated may not be captured, a limitation common to mobility-based epidemiological studies. While we imputed annual estimates and used time-dependent decay, real mobility patterns can change unpredictably.”

Walekhwa et al. (2026) “4. Discussion”: “Real-time updating, running the model weekly with new DTM data and case counts, would transform this retrospective analysis into a prospective decision support tool, as advocated by Kraemer et al. (2018) for real-time epidemic response.”

Chamla et al. (2026) “Discussion”: “Although South Sudan’s importation probability is lower than Uganda’s, documented deficits across all five WHO preparedness domains mean any importation would generate a substantially larger secondary outbreak. Closing these gaps within 4 to 8 weeks is the principal operational window.”

Fanelli et al. (2026a) “ABSTRACT”: “Anticipating where imported infections are most likely to occur, and whether recipient countries are prepared to detect and contain them, is essential for prioritizing surveillance and response.”

Fanelli et al. (2026a) “DISCUSSION”: “Our approach provides an operational framework for refining preparedness priorities, updating them as outbreaks evolve and medical countermeasures become available, and anticipating changes before regional dissemination accelerates.”

Fanelli et al. (2026a) “DISCUSSION”: “However, some individuals with early or mild symptoms may still travel before seeking care, particularly over shorter distances, potentially increasing importation risk to more distant destinations.”

Fanelli et al. (2026a) “DISCUSSION”: “Third, air-mediated importation risk only considered commercial passenger flights and did not account for humanitarian, military, or charter flights.”

Walekhwa et al. (2026) “4. Discussion”: “The practical implication is that early outbreak response should include rapid estimation of the infectious period and serial interval for the specific Ebola strain. Investing in field studies to characterise these biological parameters, through contact tracing data and serial interval analysis, could yield greater returns than refining operational protocols that are already performing adequately, a strategy advocated by Chowell et al. (2019) for Ebola preparedness.”

Gashema & Muvunyi (2026) “Considering epidemiological and contextual factors, the study estimated a high probability of regional spillover to Uganda (94·2%) and South Sudan (69·3%), identifying these countries as priority areas for preparedness. By contrast, the projected risks for Rwanda (8·6%) and Burundi (2·0%) remain comparatively low.”

Fanelli et al. (2026a) “DISCUSSION”: “Distinct readiness profiles emerged, ranging from countries combining high exposure with strong operational capacity, such as Zambia, Ethiopia and Tanzania, to countries where high importation risk coincides with substantial readiness gaps, including South Sudan, Burundi, and the Central African Republic. South Sudan, the country at highest importation risk, combined intermediate-low emergency preparedness and response with limited healthcare system capacity, raising concerns that imported cases could more readily generate secondary transmission.”

Gashema & Muvunyi (2026) “Importation risk alone does not determine epidemic outcomes. Countries with strong surveillance systems, diagnostic capacity, a trained health-care workforce, and effective infection prevention and control measures can successfully contain imported cases before widespread transmission occurs. Conversely, gaps in surveillance systems and health governance can facilitate epidemic propagation and delay effective response.”

Walekhwa et al. (2026) “Research in Context”: “Enhanced cross-border surveillance under International Health Regulations 2005 and the Africa CDC Public Health Emergency of Continental Security framework is crucial to harmonise detection and response across the DR Congo–Uganda–South Sudan tripoint.”

Bouhentala & Kadir (2026) “Abstract”: “Introduction likelihood was assessed as very low to low for North Africa and very low for the general European population; delayed recognition in routine healthcare was the principal scenario for limited secondary transmission.”

Bouhentala & Kadir (2026) “Abstract”: “Preparedness in North Africa and Europe should prioritise complete exposure histories, rapid isolation, validated diagnostics, protected clinical care, and contact management rather than reliance on border screening.”

Richard et al. (2026) “Results”: “CDC assessed the overall risk posed by the ongoing BVD outbreak to the U.S. population during the next 3 months as low. This assessment was made with moderate confidence, given the data available.”

van Zandvoort et al. (2026) “Conclusion”: “The risk of undetected Ebola virus transmission outside Africa is low. Nearly all confirmed Ebola disease cases reported outside Africa have been linked to known occupational exposures, with reasons for travel specific to outbreak response. We emphasise the role of local, community-based case management and infection prevention and control measures as the most effective strategy for managing outbreak risks both overall and outside the source region.”

van Zandvoort et al. (2026) “Discussion”: “Crucially, most Ebola disease cases reported outside Africa (27 of 28 cases) have been either (i) primary cases in individuals with a known responsive or occupational exposure to an Ebola disease epidemic (24 cases, including 20 healthcare workers, three UN employees and one journalist, among whom 20 were medically evacuated) or (ii) secondary cases in healthcare workers treating an Ebola disease patient outside Africa (3 cases). There was only one reported confirmed case outside Africa in a traveller with no responsive or occupational exposure.”

Verheyden and Mudogo (2026) “Abstract”: “We modelled the discrete-time hazard of a first reported confirmed case across 106 health zones in four provinces affected by the 2026 Bundibugyo virus disease outbreak (47 affected, 59 at risk, 26 July 2026), comparing four connectivity specifications - none, road-distance, a gravity score, and an incidence-weighted force-of-infection (FOI) term - fitted within an identical Bayesian hierarchical hazard architecture.”

Verheyden and Mudogo (2026) “Abstract”: “FOI had the highest top-10 hit rate (42.6%), approaching conventional significance against road-distance and no-connectivity comparators. On AUC-PR, a model with no connectivity term performed as well as or better than any connectivity specification (0.437 vs. 0.409 for FOI), a discrepancy we report rather than omit.”

Verheyden and Mudogo (2026) “3 Results”: “The top-ranked zone, Rethy (Ituri), had a posterior median daily hazard of 0.095 (95% CrI [0.018, 0.295]), corresponding to roughly a 75% chance of reporting a first case within 14 days if current conditions persisted; the 20th-ranked zone’s daily hazard corresponds to roughly 4% over the same window.”

Verheyden and Mudogo (2026) “3 Results”: “On the pre-specified primary metric (top-10 hit rate), FOI was highest (42.6%), followed by OSRM-distance and gravity (both 32.8%), then baseline (29.5%).”

Verheyden and Mudogo (2026) “3 Results”: “On AUC-PR, the secondary metric with the most divergent picture, the baseline model with no connectivity term at all performed as well as or better than every connectivity specification (baseline 0.437 vs. FOI 0.409, gravity 0.399, OSRM-distance 0.239), and FOI’s advantage over baseline on this metric was not distinguishable from zero (−0.027, 95% CI [−0.274, 0.200]).”

Verheyden and Mudogo (2026) “3 Results”: “Across a grid of 8 alternative combinations (φ ∈ {30,60,120} minutes × window ∈ {1,3,7} days), the resulting zone ranking correlated with the primary specification’s ranking at Spearman ρ = 0.90-0.99, with top-10 overlap of 9 or 10 out of 10 zones in every combination tested.”

Verheyden and Mudogo (2026) “1 Introduction”: “This paper develops a discrete-time hazard model in which spatial connectivity is represented by the observed incidence in every currently affected zone, propagated to at-risk zones through a road-network distance-decay kernel, rather than by geometric distance to the nearest affected zone alone. We refer to this as an incidence-weighted spatial force-of-infection (FOI) hazard model.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Daily case data from the INRB-UMIE BDBV2026-Data repository (Mbulayi et al., 2026). Static covariates: WorldPop population estimates; OpenStreetMap/HOT health-facility points; Complex Crisis Vulnerability Index; province-level ACLED conflict events (Raleigh et al., 2010). Road-network travel times from a national OSRM graph.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “CDR/mobility data were not used, as DRC is not among the countries for which this is available (Wardle et al., 2023).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “The FOI kernel’s decay scale (φ, fixed at 60 minutes in the primary specification) and smoothing window (3 days) were not tuned against performance; we instead report ranking stability across a grid of plausible alternatives, refitting the full model at each combination and comparing the resulting zone ranking to the primary specification via Spearman rank correlation and top-10 overlap: φ ∈ {30, 60, 120} minutes × window ∈ {1, 3, 7} days (8 combinations in addition to the primary specification).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Four specifications of xz(t)'s connectivity term were compared, all embedded in the identical model described in Section 2.3, differing only in this one covariate: 1. Baseline: no connectivity term (four static covariates only). 2. Road-distance: log(1 + travel time to the nearest already-affected zone), the specification used in an earlier version of this analysis. 3. Gravity: a population-weighted score summed over all affected zones: - (iv) FOI (primary specification): incidence-weighted, kernel-summed force of infection,.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Fitting all four within the identical architecture, rather than comparing a fitted model against externally-computed rankings, isolates what each connectivity term contributes beyond the static covariates and beyond each other, and was added specifically in response to reviewer concern that an earlier version compared a fitted model to uncalibrated external scores.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Eight weekly origins, 21 May-9 July 2026, each refit using only data available through that origin, scored against zones reporting a first case within the following 14 days (Tashman, 2000). We designate top-10 hit rate - the proportion of true next-invading zones ranked in the model’s top 10 at that origin - as the primary evaluation metric, specified before examining comparative results across specifications, because the operational decision this model is built to support is which small number of zones warrant prioritised attention, not discrimination across the full 59-zone risk set.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “We used a cluster (block) bootstrap resampling whole origins with replacement (10,000 resamples), for each replicate pooling the zone-level rows (or origin-level AUC-PR values) belonging to the resampled origin set and recomputing each metric as in the point estimate.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “We report two further checks: leave-one-origin-out (recomputing pooled metrics excluding each origin in turn, to assess whether any single origin drives the result), and a non-overlapping-origin subset (4 origins spaced exactly 14 days apart: 21 May, 4 June, 18 June, 2 July), which removes horizon overlap entirely at the cost of halving the usable origins and events (61 to 35).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “A secondary, exploratory analysis - not a second validated model of comparable standing to the hazard model above - examined near-term case growth for the 47 zones already affected, using a hierarchical negative-binomial growth-rate model (an initial Poisson specification was found miscalibrated on rolling-origin backtesting and corrected).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “We summarise the headline finding here (Section 3.5) and do not present a combined invasion-plus-growth ranking as a validated decision tool in the main paper, since that combined ranking has not itself been backtested.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “with kernel K(z,z’) = exp(−d(z,z’)/φ), φ = 60 minutes (default), and Cz’(t) the 3-day trailing mean of new confirmed cases in zone z’ through day t − 2 (Section 2.4).”

Verheyden and Mudogo (2026) “3 Results”: “The FOI kernel’s decay scale and smoothing window were not tuned against performance. Across a grid of 8 alternative combinations (φ ∈ {30,60,120} minutes × window ∈ {1,3,7} days), the resulting zone ranking correlated with the primary specification’s ranking at Spearman ρ = 0.90-0.99, with top-10 overlap of 9 or 10 out of 10 zones in every combination tested.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Same-day leakage - using zone z’s incidence recorded on day t to predict zone z’s outcome on day t - is addressed directly: the FOI covariate (Section 2.5) uses only incidence through day t−2, never day t or t−1.”

Verheyden and Mudogo (2026) “3 Results”: “This is smaller than an earlier, same-day (non-lagged) specification’s estimate of +0.610 [0.43, 0.79]; fixing same-day information leakage (Section 2.4) reduced the coefficient’s magnitude by roughly 40%, confirming the leakage had inflated the apparent effect, though the corrected estimate still excludes zero.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “The outcome modelled is a zone’s first reported confirmed case, denoted Tz, with yz(t) = 1 on the day this occurs. We use this precise language deliberately rather than ‘invasion’: first confirmation reflects viral introduction, local transmission, detection, testing access, reporting capacity, and retrospective case reclassification jointly, and the model cannot separate these.”

Verheyden and Mudogo (2026) “3 Results”: “Health-facility density (β = +0.486, [0.245, 0.734]) and population (β = +0.400, [0.087, 0.720]) were also associated with hazard at the 95% credible level; socioeconomic deprivation (+0.326, [−0.074, 0.754]) and province-level conflict exposure (+0.156, [−0.640, 0.900]) were not.”

Verheyden and Mudogo (2026) “4 Discussion”: “Conflict exposure is measured only at province level across four provinces, alongside a province random effect estimated from the same four groups; this covariate’s estimate should be read as exploratory.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Data-vintage leakage - whether the case counts available to us for a historical date t already reflect corrections made after t, since we hold only the final, most recently corrected version of the dataset rather than dated historical snapshots - cannot be ruled out with the data available to us. We flag this explicitly rather than assume it away, and for this reason describe the rolling-origin evaluation throughout as pseudo-prospective: a reconstruction of what a real-time analysis would have produced, run retrospectively on the best data now available, not a true contemporaneous test.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Linear predictor: with a piecewise-constant weekly baseline evolving as a random walk: a non-centred province random effect (Betancourt and Girolami, 2015): weakly informative priors: and the discrete-time survival log-likelihood, summed over zones and at-risk days:.”

Verheyden and Mudogo (2026) “1 Introduction”: “Second, because historical snapshots of the surveillance dataset at each rolling-origin date were not available to us, we cannot rule out that some retrospective data revision (case reclassification, backfilled reports) is reflected in what our evaluation treats as contemporaneously known information. We therefore describe the rolling-origin evaluation as pseudo-prospective throughout, reserving the term prospective for the genuinely pre-registered future verification described in Section 2.10.”

Verheyden and Mudogo (2026) “1 Introduction”: “None of these efforts provides a validated, prospectively evaluated model of zone-level invasion risk specifically, and existing approaches that do address spatial spread typically rely on mobile-phone call detail record (CDR) data, which a systematic review found available in only a small number of African countries, not including the DRC (Wardle et al., 2023).”

Verheyden and Mudogo (2026) “4 Discussion”: “The top-10-vs-OSRM-distance comparison reaches only the boundary of conventional significance (CI lower bound 0.000); we would not describe this as a robust result on its own and rely on the convergent evidence across the primary metric, the sensitivity grid, and the leave-one-origin-out and non-overlapping-origin checks collectively, rather than on any single comparison.”

Verheyden and Mudogo (2026) “3 Results”: “On AUC-PR, the secondary metric with the most divergent picture, the baseline model with no connectivity term at all performed as well as or better than every connectivity specification (baseline 0.437 vs. FOI 0.409, gravity 0.399, OSRM-distance 0.239), and FOI’s advantage over baseline on this metric was not distinguishable from zero (−0.027, 95% CI [−0.274, 0.200]). We report this plainly rather than selectively: a global discrimination metric like AUC-PR rewards a model for correctly ranking the bulk of zones that will not invade soon, which the static covariates alone (population, facility density, deprivation) appear to do reasonably well without any connectivity information; the connectivity terms’ distinguishable contribution is specifically to sharpening the top of the ranking, which is what the primary metric measures and what the operational use case (which few zones get attention first) requires.”

Verheyden and Mudogo (2026) “4 Discussion”: “The growth-rate extension (Section 3.5) has not been backtested as a combined ranking with the hazard model, and we do not present one for that reason.”

Verheyden and Mudogo (2026) “4 Discussion”: “Fixed choices not evaluated here include the OSRM travel-time transformation and the province-level (rather than zone-level) conflict aggregation.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Daily case data from the INRB-UMIE BDBV2026-Data repository (Mbulayi et al., 2026).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Retrospective, pseudo-prospective observational study of health-zone-level case emergence during the 2026 BDBV outbreak, 14 May-26 July 2026 (74 days).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “This construction yields 5,727 zone-day observations, 47 events.”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Static covariates: WorldPop population estimates; OpenStreetMap/HOT health-facility points; Complex Crisis Vulnerability Index; province-level ACLED conflict events (Raleigh et al., 2010).”

Verheyden and Mudogo (2026) “2 Materials and Methods”: “Road-network travel times from a national OSRM graph.”

Verheyden and Mudogo (2026) “4 Discussion”: “An independent Flowminder-based mobility analysis found the outbreak’s three epicentre zones among the top 5% of DRC health zones nationally on mobility intensity (Bangelesa et al., 2026, as summarised in Walekhwa et al., 2026), using data and methodology with no overlap with ours.”

Epistorm (2026) “Executive Summary”: “No transmission model is used at this stage; the analysis describes where any exported case is most likely to be detected, conditional on at least one export occurring.”

Epistorm (2026) “Key Findings”: “**Proximity spillover concentrates in western Uganda and Juba. ** Across the neighbouring countries, 99.2 % of the city-level RR mass sits in five cities: Kasese (UGA, 56.3 %), Pakuba (UGA, 25.4 %), Arua (UGA, 12.8 %), Juba (SSD, 3.1 %), and Entebbe (UGA, 1.5 %), with Kigali (RWA), Dar Es Salaam (TZA), Bangui (CAF), and Bujumbura (BDI) capturing the remaining thin margin (each below 0.05 %).”

Epistorm (2026) “Key Findings”: “** With only Uganda excluded and the remainder renormalised, South Sudan takes 80.1 % of the tail (Juba alone 79.6 %).”

Epistorm (2026) “Neighbouring-Country Cities”: “Across DRC’s nine neighbouring countries, the mobility model concentrates 99.2 % of the non-DRC relative risk mass in a small number of cities on the eastern side of the country. Five cities (Kasese in Uganda’s Western Region, Pakuba in Northern Region, Arua in West Nile Region, Entebbe near Kampala; Juba in South Sudan’s Central Equatoria State) together account for 99.2 % of the neighbour total.”

Epistorm (2026) “Neighbouring-Country Cities”: “Uganda alone accounts for 96.1 % of the non-DRC mass and is the sole country removed from the tail in Sections 4 and 5.”

Epistorm (2026) “Regional Distribution · International Tail”: “With DRC and Uganda removed and the remainder renormalised to sum to one, the tail is dominated by South Sudan: Juba alone carries 79.6 %, and Rumbek adds a further 0.5 %. This reflects the strong overland connectivity from north-eastern DRC into South Sudan captured by the mobility model.”

Epistorm (2026) “Regional Distribution · International Tail”: “Africa carries 89.4 % of the tail (South Sudan alone 80.1 %; then Kenya 2.8 %, Tanzania 1.1 %, Rwanda 0.8 %, South Africa 0.8 %, Ethiopia 0.7 %). The Middle East accounts for 4.9 % (UAE 3.2 %, Saudi Arabia 0.9 %, Qatar 0.3 %), Europe for 3.0 % (United Kingdom 1.0 %, Germany 0.3 %, Belgium 0.3 %), Asia for 2.0 % (India 0.9 %, China 0.7 %), the Americas for 0.7 % (United States 0.5 %, Canada 0.2 %), and the “Other” residual sits at 0.1 %.”

Epistorm (2026) “Executive Summary”: “** The absolute probability of any single international export remains small.”

Epistorm (2026) “Executive Summary”: “They represent the share of international export probability that would land in each destination if an international export were to occur. They do not represent the absolute probability that an export will occur.”

Epistorm (2026) “Methods Overview”: “Global air travel patterns are derived from the International Air Transport Association (IATA) and the Official Airline Guide (OAG), providing passenger-level origin-destination flows between international airports. These data are complemented by short-scale commuting patterns capturing daily movements between adjacent subpopulations, representing the local spread of disease through routine human activity within and across health zones. The modeling approach follows the GLEAM (Global Epidemic and Mobility) computational framework [4,5].”

Epistorm (2026) “Methods Overview”: “For each potential destination Y, the model estimates the probability P(Y) that an infected individual originating in the affected area travels to Y, conditional on at least one exported case occurring. The output is a relative risk distribution across all candidate destinations, normalized so that values represent the share of importation probability assigned to each location.”

Epistorm (2026) “Methods Overview”: “The first (Section 3) reports the city-level RR across DRC’s neighbouring countries (before renormalisation), showing that Uganda alone accounts for 96.1 % of the non-DRC mass.”

Epistorm (2026) “Methods Overview”: “The second (Sections 4 and 5) removes only Uganda and renormalises the remaining mass over 117 international destinations reachable by air.”

Epistorm (2026) “Executive Summary”: “The framework is unchanged from Reports #1 to #4: air-travel and short-scale commuting flows are combined into a mobility-only relative importation risk (RR), evaluated over a two-week horizon.”

Epistorm (2026) “Limitations &; Assumptions”: “This report uses the mobility framework only. No transmission-dynamics model is applied, and the RR values should not be read as forecasts of future case counts at any destination.”

Epistorm (2026) “Methods Overview”: “The mobility model does not simulate country-level border-control interventions such as entry screening, restrictions on travellers from the affected area, or air-service suspensions.”

Epistorm (2026) “Limitations &; Assumptions”: “The model does not include country-level border-control interventions such as entry screening, restrictions on travellers from the affected area, or air-service suspensions.”

Epistorm (2026) “Methods Overview”: “Readers should treat the RR values here as the mobility-only baseline.”

Epistorm (2026) “Limitations &; Assumptions”: “The RR values here should be interpreted as the mobility-only baseline.”

Epistorm (2026) “Limitations &; Assumptions”: “The incubation period of Bundibugyo virus (2-21 days; average ~10 days) constrains the window within which an exported case could reasonably travel while infectious or pre-symptomatic.”

Epistorm (2026) “Limitations &; Assumptions”: “Case importations are modelled as statistically independent events.”

Epistorm (2026) “Limitations &; Assumptions”: “The mobility model does not incorporate sociodemographic attributes of travellers (age, economic status, occupation, or pre-existing medical conditions), any of which may modulate both exposure risk at origin and care-seeking behaviour at destination. Travel probability within a catchment area is treated as independent of these attributes and of specific location within the catchment.”

Epistorm (2026) “Limitations &; Assumptions”: “These estimates apply to traffic from the general population.”

Epistorm (2026) “Limitations &; Assumptions”: “The importation risk associated with repatriation flights, responder evacuations, or other ad-hoc operational movements is not part of this analysis.”

Epistorm (2026) “Limitations &; Assumptions”: “Mobility data may not fully capture informal cross-border movement, which is substantial within the immediate cross-border region but is not the focus of the international-tail analysis in Sections 4 and 5.”

Epistorm (2026) “Limitations &; Assumptions”: “The model does not predict downstream transmission chains following an importation event: risk estimates reflect the probability of a single importation and do not account for the probability of sustained local transmission in the receiving location.”

Epistorm (2026) “Limitations &; Assumptions”: “Real-world events involving multiple related cases (for example, a family cluster travelling together) are counted as a single importation event. To the extent that such clusters are common, the model will under-count the number of distinct exposure events while correctly counting the number of distinct geographic seedings.”

Epistorm (2026) “Limitations &; Assumptions”: “The mobility model does not incorporate sociodemographic attributes of travellers (age, economic status, occupation, or pre-existing medical conditions), any of which may modulate both exposure risk at origin and care-seeking behaviour at destination.”

Epistorm (2026) “Executive Summary”: “This update refreshes the international-dissemination assessment using the 23 July 2026 DRC data.”

Epistorm (2026) “Methods Overview”: “Global air travel patterns are derived from the International Air Transport Association (IATA) and the Official Airline Guide (OAG), providing passenger-level origin-destination flows between international airports.”

Epistorm (2026) “Methods Overview”: “These data are complemented by short-scale commuting patterns capturing daily movements between adjacent subpopulations, representing the local spread of disease through routine human activity within and across health zones.”

Epistorm (2026) “Situation at a glance”: “Incubation period 2-21 days · average ~10 days (from 2007 outbreak).”

Epistorm (2026) “Situation at a glance”: “Pathogen Bundibugyo virus, second Ebola species; first isolated 2007, Uganda; confirmed CFR ~14% (DRC/UGA combined).”

Bouhentala & Kadir (2026) “Abstract”: “Risk was assessed across introduction likelihood, conditional onward-transmission likelihood, impact, and confidence.”

Bouhentala & Kadir (2026) “3. Results”: “As of 15 July 2026, the DRC had reported 2,124 laboratory-confirmed cases, 828 deaths among confirmed cases, 390 reported recoveries, and 725 patients hospitalised in isolation. Confirmed cases had been reported from 46 of 140 health zones across five provinces.”

Bouhentala & Kadir (2026) “3. Results”: “Excluding the 29 June – 1 July interval, the aggregate rate was 35.9 per day during 2–29 June (970 notifications over 27 days; exact counting interval 33.7–38.3) and 47.4 per day during 1–15 July (664 over 14 days; 43.9–51.2). The July-to-June notification-rate ratio was 1.32 (counting-model 95% interval 1.20–1.46).”

Bouhentala & Kadir (2026) “3. Results”: “ECDC estimated approximately one importation per 23,000–24,000 travellers from the principal outbreak region, with a 90% uncertainty interval of one per 13,000–54,000. Under a hypothetical scenario of 100 travellers during 11–25 June, ECDC estimated a 0.45% probability of at least one importation (90% uncertainty interval 0.20%–0.85%).”

Bouhentala & Kadir (2026) “2. Methods”: “This was a rapid public-health analysis of publicly available aggregate surveillance data and official risk assessments. The principal sources were WHO Disease Outbreak News, WHO/IHR statements and technical guidance, ECDC surveillance and modelling reports, WHO regional offices, national ministry-of-health dashboards and updates, and peer-reviewed primary studies for historical BDBV outbreaks, viral persistence, and candidate countermeasures.”

Bouhentala & Kadir (2026) “2. Methods”: “Airline sources were used only to establish the existence of plausible routes and not to estimate passenger volumes.”

Bouhentala & Kadir (2026) “2. Methods”: “Cumulative confirmed DRC counts were indexed by reporting date rather than symptom-onset date. Interval-specific average notification rates were calculated as the change in cumulative confirmed cases divided by the number of days between reports. Exact 95% Poisson counting intervals were calculated using the Garwood method.”

Bouhentala & Kadir (2026) “2. Methods”: “The notification-rate ratio comparing 1–15 July with 2–29 June was calculated as (C₁/t₁)/(C₀/t₀). The 95% interval was estimated on the log scale using SE[log(RR)] = √(1/C₁ + 1/C₀).”

Bouhentala & Kadir (2026) “2. Methods”: “No reproduction number was estimated because onset-date line-list data were unavailable.”

Bouhentala & Kadir (2026) “2. Methods”: “Risk was separated into four dimensions: introduction likelihood, conditional onward-transmission likelihood after an importation, impact if detection was delayed, and confidence in the judgement. Categories were comparative reference classes rather than calibrated probabilities. Boundary ratings such as ‘very low to low’ were retained when the evidence did not support a more precise classification.”

Bouhentala & Kadir (2026) “2. Methods”: “The ratings were calibrated against WHO’s 6 June assessment, which classified risk as very high in the DRC, high in Uganda and bordering countries, low in the remainder of the WHO African Region, and low globally.”

Bouhentala & Kadir (2026) “2. Methods”: “These intervals describe counting variation only and do not incorporate systematic uncertainty from backlogs, under-ascertainment, changing testing, or reporting delays.”

Bouhentala & Kadir (2026) “3. Results”: “This increase cannot be interpreted as a direct measure of biological acceleration. WHO noted that expanded surveillance, testing, and diagnostic capacity contributed to rising reported counts, and newly reported cases could include backlogs.”

Bouhentala & Kadir (2026) “2. Methods”: “Risk was separated into four dimensions: introduction likelihood, conditional onward-transmission likelihood after an importation, impact if detection was delayed, and confidence in the judgement. Categories were comparative reference classes rather than calibrated probabilities.”

Bouhentala & Kadir (2026) “3. Results”: “Passenger-volume and confidential readiness data were unavailable.”

Bouhentala & Kadir (2026) “4. Discussion”: “Conversely, a recognised symptomatic case can usually be contained because transmission requires direct contact with blood or other body fluids, contaminated materials, or the body of a deceased case rather than ordinary casual proximity.”

Bouhentala & Kadir (2026) “4. Discussion”: “An incubating traveller can cross a border while asymptomatic because people are not considered infectious before symptom onset and the incubation period can extend to 21 days. Temperature screening may detect some symptomatic travellers but cannot exclude infection during incubation.”

Bouhentala & Kadir (2026) “3. Results”: “A final departure airport may conceal earlier travel to eastern DRC; triage therefore requires a complete 21-day itinerary and exposure history.”

Bouhentala & Kadir (2026) “2. Methods”: “Arithmetic cumulative doubling times were treated as descriptive properties of the cumulative curve and not as evidence of changes in transmission.”

Bouhentala & Kadir (2026) “4. Discussion”: “Passenger volumes, traveller occupations, deployment rosters, medevac activity, and country-specific laboratory and referral readiness were incomplete. Country ratings were therefore comparative judgements rather than calibrated probabilities.”

Bouhentala & Kadir (2026) “4. Discussion”: “Nevertheless, notification rates remain composite measures of infection, care seeking, alert investigation, specimen transport, laboratory capacity, retrospective death investigation, and reporting. The exact Poisson intervals are therefore deliberately described as counting-model intervals rather than complete epidemiological uncertainty intervals.”

Bouhentala & Kadir (2026) “3. Results”: “This rise is compatible with maturation of outcomes among previously active cases, retrospective confirmation of deaths, delayed presentation, preferential detection of severe disease, delayed recovery reporting, and changes in testing or death investigation. It should not be interpreted as a time-resolved measure of clinical virulence.”

Bouhentala & Kadir (2026) “3. Results”: “The published status totals did not fully reconcile: 828 deaths, 390 recoveries, and 725 hospitalised patients summed to 1,943, leaving 181 of 2,124 confirmed cases outside those categories. Public reports did not define the status of these cases. Potential explanations include pending admission or outcome reconciliation, transfers, and reporting lags, but no category could be assigned without line-list data.”

Bouhentala & Kadir (2026) “4. Discussion”: “The analysis was time-stamped and may be superseded by subsequent official reports.”

Bouhentala & Kadir (2026) “Abstract”: “** Public surveillance reports from the World Health Organization (WHO), European Centre for Disease Prevention and Control (ECDC), Africa CDC, national ministries of health, and peer-reviewed sources were synthesised through 15–17 July 2026.”

Bouhentala & Kadir (2026) “2. Methods”: “The main DRC trajectory used laboratory-confirmed cases and deaths among confirmed cases. Suspected, probable, and confirmed categories were not summed across reports because individuals and deaths can be reclassified as investigations progress.”

Bouhentala & Kadir (2026) “3. Results”: “As of 17 July, Uganda’s official dashboard reported 20 confirmed cases, two confirmed deaths, 18 recoveries, and no current admissions.”

Bouhentala & Kadir (2026) “3. Results”: “WHO’s jurisdictional accounting on 17 July comprised 2,145 confirmed cases globally: 2,124 in the DRC, 20 in Uganda, and one in France, with 830 confirmed-case deaths.”

Bouhentala & Kadir (2026) “2. Methods”: “The principal sources were WHO Disease Outbreak News, WHO/IHR statements and technical guidance, ECDC surveillance and modelling reports, WHO regional offices, national ministry-of-health dashboards and updates, and peer-reviewed primary studies for historical BDBV outbreaks, viral persistence, and candidate countermeasures.”

Bouhentala & Kadir (2026) “3. Results”: “ECDC estimated approximately one importation per 23,000–24,000 travellers from the principal outbreak region, with a 90% uncertainty interval of one per 13,000–54,000.”

Bouhentala & Kadir (2026) “3. Results”: “These were separately generated Monte Carlo summaries and should not be forced into exact reciprocal equivalence.”

Fanelli et al. (2026a) “ABSTRACT”: “We developed an integrated spatial framework combining high-resolution epidemic and population data, air and land mobility, geographical accessibility, conflict-adjusted travel times to estimate the risk of Ebola importation across African countries. We contextualized importation risk using different national readiness indicators.”

Fanelli et al. (2026a) “ABSTRACT”: “Importation risk was highly concentrated geographically, with neighbouring countries accounting for 85% of the estimated continental risk. However, high-risk destinations extended beyond the immediate neighbours, whereas some neighbouring countries remained at low risk, reflecting the complementary roles of land and air mobility in shaping regional dissemination. South Sudan, Rwanda, Kenya, Tanzania, Zambia, Ethiopia, Burundi, Angola, South Africa, the Central African Republic, and Nigeria were estimated at high risk, but differed markedly in healthcare system capacity and emergency preparedness and response. Readiness gaps were largest in South Sudan, Burundi, and the Central African Republic.”

Fanelli et al. (2026a) “RESULTS”: “The 11 countries in the high-risk quartile — South Sudan, Rwanda, Kenya, Zambia, Ethiopia, Tanzania, Burundi, Angola, South Africa, Central African Republic, and Nigeria — accounted for 93% of the total estimated importation risk, and formed a largely contiguous regional belt surrounding the affected area, with the exception of Ethiopia, South Africa, and Nigeria, which were geographically more distant (at border distance 2, 3, and 3 from the outbreak source, respectively).”

Fanelli et al. (2026a) “RESULTS”: “Neighboring countries alone accounted for 85% of the total estimated risk, while countries located two border crossings away contributed an additional 9%, highlighting the strong spatial concentration of potential spread around the outbreak source (Fig. 2f).”

Fanelli et al. (2026a) “RESULTS”: “Ethiopia, South Africa, and Nigeria were the only non-neighbouring countries classified as high risk overall, and each ranked substantially higher when considering air importation risk than land importation risk: Ethiopia ranked first for air importation risk but fifth for land importation risk, while South Africa ranked third and ninth, respectively, and Nigeria seventh and eleventh.”

Fanelli et al. (2026a) “RESULTS”: “Epidemic amplification in Nairobi and Lusaka produced the largest increase in continental importation (4-fold increase), and especially in southern Africa (up to 15-fold). Amplification in Kampala and Kigali produced a moderate increase in importation risk across the continent (2.5-fold and 2.8-fold, respectively), whereas expansion to Kinshasa, Bujumbura, or Juba produced comparatively limited continental effects.”

Fanelli et al. (2026a) “INTRODUCTION”: “We developed an integrated analytical framework to estimate the risk of Ebola Bundibugyo dissemination in Africa in relation to national readiness, by combining epidemiological and high-resolution spatial population data, commercial air transportation, land mobility, geographical accessibility, conflict-adjusted travel-times, and different readiness scores.”

Fanelli et al. (2026a) “METHODS”: “The importation risk for each African country was estimated separately for air and land mobility, as the probability that an infected traveller from the affected areas would reach that country. We assumed that Ebola infected cases could travel only in their latent phase, as symptomatic patients would be expected to either remain in the affected country for clinical management or, if transferred internationally, travel under medically supervised and controlled conditions. Therefore, for land mobility, travel times were explicitly incorporated to estimate the destination reached before the onset of symptoms. For air travel, journey times were considered negligible relative to the incubation period.”

Fanelli et al. (2026a) “METHODS”: “Risks were weighted by the cumulative attack rate in each affected administrative unit. Air and land importation risks were then combined to obtain an overall estimate of relative importation risk for each destination country.”

Fanelli et al. (2026a) “METHODS”: “Air mobility was derived from the International Air Transport Association (IATA) data on origin–destination passenger flows between any pair of commercial airports and accounting for connections at intermediate airports (Fig. 1c).”

Fanelli et al. (2026a) “METHODS”: “For air travel, it was assigned using weights proportional to population size and decreasing with the distance to each airport of the country.”

Fanelli et al. (2026a) “METHODS”: “Land mobility was then obtained as residual mobility by subtracting air mobility from the Global Transnational Mobility Dataset 2.0 (GTMD2), which provides estimates of overall population movements between countries (Fig. 1d).”

Fanelli et al. (2026a) “METHODS”: “To characterize spatial accessibility from the outbreak area and the potential for land-based spread within the incubation period, we computed travel-time surfaces using a global motorized-travel friction surface map integrating road networks, land cover, rivers, topography, and transport infrastructure. Travel times were estimated from each affected health district (DRC) and ADM3 units (Uganda) to all destination countries in Africa. As armed conflict can substantially alter travel time, our estimates were adjusted for conflict-related disruptions using the most recent conflict data at ADM2 level obtained from the Armed Conflict Location & Event Data Project (ACLED), and calibrated against Google Maps driving-time estimates.”

Fanelli et al. (2026a) “METHODS”: “For land travel, the share of the population in each administrative unit travelling to each destination country was estimated through a gravity model calibrated on observed land mobility patterns.”

Fanelli et al. (2026a) “DISCUSSION”: “Although uncertainty remains, sensitivity analyses reducing land mobility by 66% and 90% produced only limited changes in country rankings, supporting the robustness of our findings. Second, land-mediated importation risk was estimated assuming an average travel duration of 8h per day, and analyses were restricted to contiguous Africa because comparable data on maritime mobility are not available. Alternative travel durations produced no changes in the estimated geography of risk.”

Fanelli et al. (2026a) “METHODS”: “National epidemic readiness was assessed using several complementary frameworks, including the WHO State Parties Self-Assessment Annual Reporting (SPAR, 2025), the Global Health Security Index (GHS, 2021), the Infectious Disease Vulnerability Index (IDVI, 2016), and the Epidemic Risk Index (INFORM-WHO, 2020). From each framework, we extracted indicators describing Healthcare system capacity and Emergency preparedness and response.”

Fanelli et al. (2026a) “METHODS”: “Given its complete coverage of African countries, recent update, and direct relevance to the current WHO Framework for Health Emergency Preparedness and Response Capabilities for national public health agencies, SPAR was selected as the primary epidemic readiness metric for the main analysis (Fig. 1fg).”

Fanelli et al. (2026a) “METHODS”: “Corresponding GHS indicators, complementary to SPAR indicators, were used in sensitivity analyses.”

Fanelli et al. (2026a) “METHODS”: “Amplification scenarios were designed to assess the potential impact of geographical expansion of the epidemic beyond the currently affected area into highly connected urban centers. Tested locations included the capital cities of the two affected countries, Kinshasa (DRC) and Kampala (Uganda), and the capitals of neighboring countries scoring highest in importation risk: Juba (South Sudan), Kigali (Rwanda), Bujumbura (Burundi), Nairobi (Kenya), and Lusaka (Zambia).”

Fanelli et al. (2026a) “METHODS”: “To account for geographically heterogeneous reporting rates, we used the proportion of alerts investigated among reported alerts in each DRC province, averaged over the two weeks preceding 5 July 2026, relative to 100% reporting in Uganda. We explored alternative values in the sensitivity analysis.”

Fanelli et al. (2026a) “METHODS”: “We assumed that Ebola infected cases could travel only in their latent phase, as symptomatic patients would be expected to either remain in the affected country for clinical management or, if transferred internationally, travel under medically supervised and controlled conditions.”

Fanelli et al. (2026a) “DISCUSSION”: “We assumed that infected individuals travelled only during the latent phase, consistent with the rapid onset of symptoms and progressive clinical deterioration characteristic of Ebola virus disease. However, some individuals with early or mild symptoms may still travel before seeking care, particularly over shorter distances, potentially increasing importation risk to more distant destinations.”

Fanelli et al. (2026a) “DISCUSSION”: “Similarly, geographically heterogeneous under-ascertainment is difficult to quantify. We therefore used the proportion of investigated alerts as a proxy for reporting rates by health district, and found the results to be robust to alternative assumptions.”

Fanelli et al. (2026a) “DISCUSSION”: “Second, land-mediated importation risk was estimated assuming an average travel duration of 8h per day, and analyses were restricted to contiguous Africa because comparable data on maritime mobility are not available. Alternative travel durations produced no changes in the estimated geography of risk.”

Fanelli et al. (2026a) “METHODS”: “Border restrictions and closures related to the current outbreak were not considered, because the implemented measures were partial, temporary and largely limited to official points of entry. Also, available data on cross-country mobility suggested no meaningful reduction in cross-border movement occurred during such restrictions.”

Fanelli et al. (2026a) “METHODS”: “Risks were weighted by the cumulative attack rate in each affected administrative unit.”

Fanelli et al. (2026a) “DISCUSSION”: “First, reliable data on cross-border land mobility in Africa remain scarce. We therefore estimated land mobility from the best available continental dataset, combining multiple mobility sources and deriving land movements as the residual after discounting commercial air travel.”

Fanelli et al. (2026a) “DISCUSSION”: “Although uncertainty remains, sensitivity analyses reducing land mobility by 66% and 90% produced only limited changes in country rankings, supporting the robustness of our findings.”

Fanelli et al. (2026a) “DISCUSSION”: “Second, land-mediated importation risk was estimated assuming an average travel duration of 8h per day, and analyses were restricted to contiguous Africa because comparable data on maritime mobility are not available.”

Fanelli et al. (2026a) “DISCUSSION”: “Finally, we did not explicitly model behavioural adaptations or the impact of border control measures. While some border restrictions were implemented during the current outbreak, independent mobility data suggested no measurable change in cross-border movements. Moreover, evidence from previous Ebola outbreaks highlights that travel restrictions have limited impact on spatial dissemination while potentially hampering response activities.”

Fanelli et al. (2026a) “DISCUSSION”: “At the same time, recent outbreak experience demonstrates that operational preparedness is not solely determined by structural readiness indicators.”

Fanelli et al. (2026a) “METHODS”: “Cumulative confirmed cases were obtained from the Institut National de Santé Publique (INSP) of the DRC in the daily situation report number 51 published on 5 July 2026, and by the Ministry of Health of Uganda on the update of the same date (Fig. 1a).”

Fanelli et al. (2026a) “METHODS”: “Population data were obtained from the 2024 health-zone population projections produced by the DRC Information Management Working Group and disseminated by the United Nations Office for the Coordination of Humanitarian Affairs (OCHA) through the Humanitarian Data Exchange Platform.”

Fanelli et al. (2026a) “METHODS”: “For Uganda, population estimates at administrative levels 0–3 were obtained from the United Nations Population Fund (UNFPA), based on Uganda Bureau of Statistics data.”

Fanelli et al. (2026a) “METHODS”: “As armed conflict can substantially alter travel time, our estimates were adjusted for conflict-related disruptions using the most recent conflict data at ADM2 level obtained from the Armed Conflict Location & Event Data Project (ACLED), and calibrated against Google Maps driving-time estimates.”

Fanelli et al. (2026a) “METHODS”: “National epidemic readiness was assessed using several complementary frameworks, including the WHO State Parties Self-Assessment Annual Reporting (SPAR, 2025), the Global Health Security Index (GHS, 2021), the Infectious Disease Vulnerability Index (IDVI, 2016), and the Epidemic Risk Index (INFORM-WHO, 2020).”

Fanelli et al. (2026a) “METHODS”: “To account for geographically heterogeneous reporting rates, we used the proportion of alerts investigated among reported alerts in each DRC province, averaged over the two weeks preceding 5 July 2026, relative to 100% reporting in Uganda.”

Gashema & Muvunyi (2026) “A modelling study by Dick Chamla and colleagues, published in The Lancet Infectious Diseases shows how fitted stochastic susceptible, exposed, infectious, recovered and deceased (SEIRD) ensemble modelling can leverage limited data to predict cross-border outbreak spread and inform regional preparedness.”

Gashema & Muvunyi (2026) “Linking outbreak dynamics with spillover risk, Chamla and colleagues move beyond forecasting disease burden toward informing operational preparedness.”

Gashema & Muvunyi (2026) “Stochastic models assume homogeneous population mixing and uniform intervention effectiveness, which potentially overlooking important spatial, health-care, and socioeconomic heterogeneities that shape transmission dynamics.”

Gashema & Muvunyi (2026) “Projections are further constrained by the quality of early surveillance data and approximated mobility patterns, particularly in low or limited resource settings where data gaps are common.”

Gashema & Muvunyi (2026) “Additionally, these models remain restricted in their ability to capture behavioural adaptation and community engagement, including changes in risk perception, health-care seeking behaviour, intervention uptake, and public trust, all of which can substantially influence outbreak trajectories.”

Gashema & Muvunyi (2026) “The estimated 6-week interval between the probable index case and formal outbreak declaration created a window during which transmission chains expanded undetected.”

Gashema & Muvunyi (2026) “The study also demonstrates the limitations of real-time epidemic modelling during rapidly evolving outbreaks. Models developed during unfolding outbreaks inevitably operate under conditions of incomplete information, evolving surveillance systems, and changing response capacities which can affect their ability to accurately predict transmission dynamics. Accordingly, these models are not intended to provide a precise forecast, but rather to identify reasonable transmission trajectories and inform preparedness decisions.”

Gashema & Muvunyi (2026) “Collectively, these limitations underscore the importance of interpreting model outputs as scenario-based estimates that require continuous refinement as more comprehensive epidemiological, mobility, and behavioural data become available.”

Gashema & Muvunyi (2026) “The study also underscores the growing importance of epidemic intelligence systems that integrate epidemiological, laboratory, and mobility data into real-time risk assessments.”

Chamla et al. (2026) “Summary”: “The study aimed to establish a short-term trajectory of the BDBV outbreak and probability of cross-border spillover into countries with elevated risk of importation to guide preparedness priorities. We calibrated a stochastic SEIRD (susceptible, exposed, infectious, recovered, dead) ensemble model to the laboratory-confirmed case series, anchoring on 598 cumulative confirmed cases on 8 June 2026 (day 68) using simulation filtering (calibration window plus or minus 30%; reporting fraction 1.0 for laboratory-confirmed cases).”

Chamla et al. (2026) “Summary”: “A linked daily-hazard spillover model estimated importation probability for Uganda, South Sudan, Rwanda and Burundi over a 12-week horizon.”

Chamla et al. (2026) “Summary”: “Calibrated to the confirmed-case anchor (598 on 8 June 2026; central basic reproduction number [R0] = 1.71), the confirmed-case trajectory is most consistent with the central scenario. Under the central scenario the ensemble projected a median of 990 cumulative confirmed cases by week 12 (24 June 2026; 90% prediction interval [PI] 709 to 1,293) and 174 deaths; the low scenario projected 870 confirmed cases (90% PI 641 to 1,133) and 160 deaths.”

Chamla et al. (2026) “Results”: “Under the central transmissibility scenario (mean accepted R0 = 1.71), the ensemble calibrated to the confirmed-case anchor projected a median of 648 cumulative confirmed cases by week 10 (8 June 2026; 50% prediction interval [PI] 560 to 732), rising to 990 by week 12 (24 June 2026; 50% PI 858 to 1,141 and 90% PI 709 to 1,293), with a projected median of 174 deaths at week 12 (Table 1).”

Chamla et al. (2026) “Results”: “By week 24, the central scenario projected a median of 8,210 cumulative confirmed cases (50% PI 6,636 to 10,287 and 90% PI 5,063 to 13,498) and approximately 1,440 deaths if transmission were merely sustained.”

Chamla et al. (2026) “Results”: “Under the low-transmissibility scenario (mean accepted R0 = 1.42), cumulative confirmed cases reached a median of 870 by week 12 (50% PI 752 to 1,004 and 90% PI 641 to 1,133; 160 deaths) and 3,662 by week 24 (50% PI 2,884 to 4,484 and 90% PI 2,094 to 6,037; 673 deaths; appendix 3 p 10).”

Chamla et al. (2026) “Results”: “Under the high-transmissibility scenario (mean accepted R0 = 2.08), confirmed cases reached a median of 1,364 (50% PI 1,154 to 1,579 and 90% PI 975 to 1,807; 202 deaths) by week 12, rising to 66,420 (50% PI 56,560 to 78,937 and 90% PI 41,718 to 95,881; 10,644 deaths) by week 24 (appendix 3 p 9).”

Chamla et al. (2026) “Results”: “Because the observed confirmed-case trajectory is consistent with the central scenario, the high scenario is not the operative trajectory; it is retained as an upper-bound contingency should control measures lapse.”

Chamla et al. (2026) “Methods”: “We conducted a stochastic modelling study using a discrete-time SEIRD (susceptible, exposed, infectious, recovered, dead) framework, recalibrated to cumulative reported cases in Ituri Province on 18 May 2026 (day 48) and projected forward over a 24-week horizon to 16 September 2026.”

Chamla et al. (2026) “Methods”: “Daily transitions were implemented as binomial draws using standard mass-action incidence (full mathematical specification in appendix 3 p 2). No vaccinated compartment was included, as there is no licensed BDBV-specific vaccine.”

Chamla et al. (2026) “Methods”: “Three transmissibility scenarios (low basic reproduction number [R0] = 1.42; central R0 = 1.71; high R0 = 2.08) spanned published BDBV estimates.”

Chamla et al. (2026) “Methods”: “Response-intensity reductions applied from 15 May 2026 were 20% (low), 25% (central) and 15% (high).”

Chamla et al. (2026) “Methods”: “Calibration was performed by simulation filtering (approximate rejection sampling; not formal Bayesian inference). Replicates were accepted if cumulative laboratory-confirmed cases at day 68 (8 June 2026) fell within plus or minus 30% of 598 (range 419 to 777). We retained 400 calibration-passing replicates per scenario.”

Chamla et al. (2026) “Methods”: “Importation risk to Uganda, South Sudan, Rwanda and Burundi was modelled as a daily Poisson hazard driven by calibrated infectious prevalence, country-specific cross-border mobility, point-of-entry detection sensitivity (from WHO Joint External Evaluation assessments) and travel-while-infectious parameters (Uganda 12%; South Sudan 15%; Rwanda and Burundi 8%). Full equations and parameter values are in appendix 3 (pp 2–3).”

Chamla et al. (2026) “Methods”: “Importation probability = 1 − exp(cumulative 12-week hazard). Probabilities of at least one importation were estimated for Uganda, South Sudan, Rwanda and Burundi.”

Chamla et al. (2026) “Methods”: “Sensitivity analyses examined travel-while-infectious (5 to 25%), reporting completeness (0.60 to 0.90) and catchment size (250,000 to 500,000).”

Chamla et al. (2026) “Methods”: “These distributions were chosen to span the range of published BDBV estimates from the 2007 Uganda and 2012 DR Congo outbreaks.”

Chamla et al. (2026) “Summary”: “Calibrated to the confirmed-case anchor (598 on 8 June 2026; central basic reproduction number [R0] = 1.71), the confirmed-case trajectory is most consistent with the central scenario.”

Chamla et al. (2026) “Methods”: “The latent period was fixed at 9.7 days and the infectious period at 5.0 days, in line with the broader filovirus literature.”

Chamla et al. (2026) “Methods”: “The case-fatality ratio was drawn from a uniform distribution (0.15 to 0.25; appendix 3 pp 3–4), centred on the case-fatality ratio observed among laboratory-confirmed cases in the current outbreak (115 [approximately 19%] of 598 as of 8 June 2026) and consistent with the historical BDBV range.”

Chamla et al. (2026) “Methods”: “Because the analysis is anchored on laboratory-confirmed cases, the reporting fraction was set at 1.0 (laboratory-confirmed cases are, by definition, ascertained); the earlier suspected-case reporting fraction assumption is no longer required.”

Chamla et al. (2026) “Methods”: “The model incorporated a transmission reduction factor representing the combined effects of infection prevention and control measures, community engagement, safe and dignified burial and contact tracing, with reduction factors varying by scenario to reflect differing operational effectiveness from the date of outbreak declaration onward.”

Chamla et al. (2026) “Discussion”: “The response-intensity factor was applied uniformly from 15 May, but real-world response effectiveness varies by place, time, access and community acceptance.”

Chamla et al. (2026) “Methods”: “We modelled transmission within a homogeneously mixing catchment of 350,000 (seven affected health zones, including Ituri and North Kivu). Spatial heterogeneity is not captured; this is a stated limitation.”

Chamla et al. (2026) “Methods”: “Importation risk to Uganda, South Sudan, Rwanda and Burundi was modelled as a daily Poisson hazard driven by calibrated infectious prevalence, country-specific cross-border mobility, point-of-entry detection sensitivity (from WHO Joint External Evaluation assessments) and travel-while-infectious parameters (Uganda 12%; South Sudan 15%; Rwanda and Burundi 8%).”

Chamla et al. (2026) “Methods”: “These parameters were not empirically derived for this outbreak and represent the principal sources of uncertainty in spillover projections.”

Chamla et al. (2026) “Results”: “Varying the travel-while-infectious parameter from 5% to 25% changed Uganda’s importation probability by less than 5 percentage points (now saturating at the upper end of the distribution) but altered South Sudan’s projected importation probability by approximately 20 percentage points (appendix 3 pp 7–8).”

Chamla et al. (2026) “Methods”: “Calibration was performed by simulation filtering (approximate rejection sampling; not formal Bayesian inference).”

Chamla et al. (2026) “Methods”: “The plus or minus 30% tolerance was chosen pragmatically given residual data uncertainty.”

Chamla et al. (2026) “Discussion”: “Homogeneous mixing oversimplifies spatial heterogeneity across health zones; healthcare-worker transmission was not separately modelled; response intensity was applied uniformly; the model is anchored on a single recent confirmed-case datum (598 on 8 June 2026), so the fit is sensitive to that anchor; simulation-filtering is not formal inference (appendix 3 p 6); and informal cross-border mobility estimates are imprecise.”

Chamla et al. (2026) “Discussion”: “The response-intensity factor was applied uniformly from 15 May, but real-world response effectiveness varies by place, time, access and community acceptance. Live mobility data and behavioural adaptation were not incorporated.”

Chamla et al. (2026) “Discussion”: “Surveillance dynamics, particularly the post-declaration case-finding surge, probably contributed to the rapid apparent growth between 15 and 19 May; we addressed this by lowering the reporting fraction and widening the calibration window, but residual confounding between transmission and surveillance intensity remains.”

Chamla et al. (2026) “Discussion”: “Most importantly, the calibration relied on early-phase surveillance counts captured in the first weekly WHO External Situation Report (data as of 18 May); the cumulative suspected-case figure of 528 was only partially resolved into 35 laboratory-confirmed cases at the time of analysis, and the associated death count of 132 reflected surveillance signals across the affected health zones rather than laboratory-confirmed mortality alone (of which there were five). Misclassification, delayed laboratory throughput and ongoing active case finding could each shift the calibration anchor in either direction.”

Chamla et al. (2026) “Discussion”: “These projections require iterative recalibration as new data emerge.”

Chamla et al. (2026) “Discussion”: “The projections presented here should therefore be interpreted as an early situational assessment rather than a definitive forecast and will be updated as verified line-list data become available.”

Chamla et al. (2026) “Discussion”: “Individual-level sex and gender data were not available in the aggregate surveillance reports used for this modelling study; disaggregated analysis by sex or gender was therefore not possible.”

Chamla et al. (2026) “Methods”: “Case data were taken from the WHO AFRO External Situation Reports and the WHO AFRO internal tracker through to 8 June 2026. The laboratory-confirmed case series was used as the primary calibration metric: confirmed cases rose from 33 on 18 May 2026 to 105 on 24 May 2026, and 598 on 8 June 2026.”

Chamla et al. (2026) “Summary”: “Uganda reported seven confirmed cases (three imported; four locally acquired including three healthcare workers; case-fatality ratio 14%).”

Chamla et al. (2026) “Results”: “Cross-border spillover projections (Figure 3 and Table 2) indicated very high importation probability for Uganda (94.2%; expected imports 2.85), where the outbreak is now firmly established: as of 4 June 2026, Uganda reported 19 confirmed cases (including five healthcare-worker infections and two deaths), reflecting sustained importation and onward local transmission since the first imported case.”

Walekhwa et al. (2026) “Abstract”: “** We constructed a data-driven directed weighted mobility network using IOM Displacement Tracking Matrix (DTM) flows collected from 15–24 May 2026 (11,245 observed movements) and the 2024 Uganda census (45.9 million people). A stochastic metapopulation SEIR model, incorporating pre-symptomatic transmission and movement of exposed and infectious individuals, was simulated over 90 days across 135 Ugandan districts and two DRC provinces.”

Walekhwa et al. (2026) “Abstract”: “Under the containment scenario (effective R = 0.35), the model projected a median of 22 cumulative cases (95% CrI: 22–24) and 2 deaths (95% CrI: 2–4) in Kampala over 90 days. All other 134 districts had a median of 0 cases. Non-pharmaceutical interventions at high, moderate, and low compliance produced no statistically significant reduction in cases (22 cases across all scenarios).”

Walekhwa et al. (2026) “3. Results”: “Under the containment scenario (effective R = 0.35), the model projected a median of 22 cumulative cases (95% CrI: 22–24) and 2 deaths (95% CrI: 2–4) over 90 days, all concentrated in Kampala (Table 3). The median total cases ranged from 22 to 22 across compliance scenarios, with 2 deaths, 3 HCW infections, and 0 HCW deaths. The 95% credible intervals overlapped substantially, indicating no statistically significant difference between compliance levels.”

Walekhwa et al. (2026) “3. Results”: “Under the containment scenario, Kampala was the only district with a non-zero median projection, bearing a predicted burden of 22 cases (95% CrI: 22–24) and a 100% probability of a local outbreak (>10 cases). All other 134 districts had a median of 0 predicted cases (upper credible intervals extending to 1 in a small fraction of simulations).”

Walekhwa et al. (2026) “2. Methods”: “The containment scenario (used for district-level predictions) had effective R₀ = 0.35, mobility scale 0.05 (95% reduction), and suppression strength 100. Moderate compliance used R₀ = 0.40, mobility scale 0.08, and suppression strength 50. Low compliance used R₀ = 0.45, mobility scale 0.12, and suppression strength 30.”

Walekhwa et al. (2026) “2. Methods”: “We built both undirected and directed weighted networks from the DTM flow data using the igraph package in R. For the undirected network, we summed reciprocal flows for each unordered pair of locations to obtain a single undirected edge weight. For the directed network, we preserved the original flow directions without collapsing reciprocal edges, allowing computation of in-strength (import risk) and out-strength (export risk) separately. We then added all 135 Ugandan districts and the two DRC provinces as nodes. Districts that did not appear in any flow record remained as isolated nodes with no edges.”

Walekhwa et al. (2026) “2. Methods”: “To impute 12-month movement estimates from the 10-day observed flows, we multiplied the observed weights by 36.5. To assess uncertainty due to the limited 10-day sampling period, we performed bootstrap resampling with 1,000 replicates.”

Walekhwa et al. (2026) “2. Methods”: “We developed a discrete-time, stochastic SEIR model at the node level. Each node maintained five compartments: Susceptible (S), Exposed (E), Infectious (I), Recovered (R), and Dead (D). The latent period was set to 7.5 days and the infectious period to 8.5 days, giving daily progression probabilities σ = 1/7.5 and γ = 1/8.5.”

Walekhwa et al. (2026) “2. Methods”: “Pre-symptomatic transmission was modelled by setting the relative infectiousness of exposed individuals to 0.5 compared to infectious individuals, based on estimates from early EVD outbreaks and the empirical observation that six of the secondary cases in Kampala tested positive while under quarantine.”

Walekhwa et al. (2026) “2. Methods”: “The baseline transmission rate β was derived from the basic reproduction number R₀ = 2.76, calibrated to the early growth of the outbreak (16 cases in 21 days), using β = R₀ / infectious period.”

Walekhwa et al. (2026) “2. Methods”: “The DRC provinces were modelled as explicit nodes with their own SEIR dynamics. We recalibrated the DRC model using actual reported data as of May 25, 2026: 381 confirmed cases, 2,987 exposed contacts, and 64 deaths. Nord-Kivu was initialised with 4,000 exposed and 400 infectious individuals, and Ituri with 500 exposed and 50 infectious, to reproduce these numbers.”

Walekhwa et al. (2026) “Abstract”: “The 7-1-7 agile response was explicitly modelled with dynamic contact tracing coverage (40% → 70% → 85%).”

Walekhwa et al. (2026) “2. Methods”: “To quantify the influence of parameter uncertainty on the total cumulative case count, we performed a global sensitivity analysis using the Sobol method. The first-order sensitivity index Sᵢ measures the fraction of output variance caused by parameter i alone, while the total-order index Tᵢ measures the fraction caused by parameter i and all its interactions with other parameters:.”

Walekhwa et al. (2026) “2. Methods”: “The design used quasi-random Sobol sequences with N = 500 samples, leading to 500 × (11 + 2) = 6,500 model evaluations. For each parameter set, we averaged the output over 3 stochastic replicates to reduce noise. The Sobol indices were then estimated with 200 bootstrap replicates to obtain 95% confidence intervals.”

Walekhwa et al. (2026) “2. Methods”: “We made the following key assumptions: The DTM flow data (10-day period) are representative of average daily movement patterns after scaling by 36.5.”

Walekhwa et al. (2026) “2. Methods”: “Movement is independent each day (no memory); only exposed and infectious individuals move, and the probability of moving is the same for both compartments.”

Walekhwa et al. (2026) “4. Discussion”: “The DTM data cover only a 10-day window and eight border crossing points; movements through unofficial crossings or after the outbreak escalated may not be captured, a limitation common to mobility-based epidemiological studies.”

Walekhwa et al. (2026) “2. Methods”: “The latent period was set to 7.5 days and the infectious period to 8.5 days, giving daily progression probabilities σ = 1/7.5 and γ = 1/8.5.”

Walekhwa et al. (2026) “2. Methods”: “The case fatality ratio (2/20 = 0.10) and HCW attack rate (3/20 = 0.15) remain constant.”

Walekhwa et al. (2026) “2. Methods”: “The DRC case fatality ratio was set to 64/381 = 0.168, based on reported outbreak data.”

Walekhwa et al. (2026) “2. Methods”: “The detection delay of 3 days applies uniformly to all infected individuals, and quarantine after detection is perfect.”

Walekhwa et al. (2026) “2. Methods”: “Reported incidence was the number of new cases on the previous day multiplied by the detection probability pdetect = 1.0 (assuming all symptomatic cases are detected).”

Walekhwa et al. (2026) “2. Methods”: “The DRC epidemic curve is independent of Uganda’s interventions and drives importations solely through the movement matrix.”

Walekhwa et al. (2026) “2. Methods”: “DRC response parameters were set to represent a moderately delayed response: detection delay 7 days (based on the 2018–2020 DRC outbreak), notification delay 3 days, response delay 5 days, contact tracing coverage 30%, detection probability 0.7, and no behavioural response (k = 1).”

Walekhwa et al. (2026) “2. Methods”: “Compliance with NPIs is constant over the 90 days and homogeneous across districts.”

Walekhwa et al. (2026) “2. Methods”: “No importations from countries other than DRC are considered.”

Walekhwa et al. (2026) “2. Methods”: “At each daily time step, the algorithm: computed the time-dependent mobility decay factor based on cumulative Ugandan cases; moved exposed and infectious individuals according to Mₜ using binomial draws; calculated the force of infection for each node, incorporating the behavioural response and the NPI multipliers; drew new exposures (binomial), new infections (negative binomial for superspreading, with dispersion parameter kdisp = 0.3), natural recoveries, and deaths (CFR = 2/20 = 0.10 for Uganda, 64/381 = 0.168 for DRC); applied dynamic contact tracing (coverage based on cumulative cases); and updated all compartments.”

Walekhwa et al. (2026) “4. Discussion”: “The homogeneous mixing assumption within districts likely overestimates transmission in low-density areas and underestimates it in high-density slums.”

Walekhwa et al. (2026) “4. Discussion”: “The Sobol confidence intervals for some parameters remained wide because the small outbreak size (median 22 cases) produces low output variance, making precise quantitative ranking unstable, hence our qualitative interpretation.”

Walekhwa et al. (2026) “4. Discussion”: “Hospital and funeral transmission are not explicitly modelled, a limitation given that Legrand et al. (2007) identified nosocomial transmission and unsafe burials as major amplifiers in Ebola outbreaks.”

Walekhwa et al. (2026) “4. Discussion”: “Importations from neighbouring countries other than DRC were not considered, though the 79.9% importation risk estimated by Chalma et al. (2026) focused specifically on DRC-origin introductions.”

Walekhwa et al. (2026) “2. Methods”: “First, the Ministry of Health EVD situation report as of June 5, 2026, provided the initial outbreak conditions. The report confirmed 15 cumulative cases (plus 1 probable), 2 deaths, 5 recoveries, and 720 cumulative contacts listed (455 active under follow-up). All initial cases were located in Kampala and were under institutional quarantine.”

Walekhwa et al. (2026) “2. Methods”: “We recalibrated the DRC model using actual reported data as of May 25, 2026: 381 confirmed cases, 2,987 exposed contacts, and 64 deaths. Nord-Kivu was initialised with 4,000 exposed and 400 infectious individuals, and Ituri with 500 exposed and 50 infectious, to reproduce these numbers.”

Walekhwa et al. (2026) “2. Methods”: “Second, the IOM DTM Flow Monitoring report covering 15–24 May 2026 summarised cross-border movements from DRC into Uganda and internal movements among Ugandan districts. From the report we extracted the total observed individuals (11,245) and the percentage distribution across eight Flow Monitoring Points (FMPs): Ntoroko Main (28%), Mpondwe (18%), Busunga (18%), Bunagana (12%), Cyanika (8%), Goli (8%), Vurra (3%), and Busanza (3%).”

Walekhwa et al. (2026) “2. Methods”: “Third, the official 2024 Uganda Population and Housing Census final report (UBOS, December 2024) gave the de facto population for each of the 135 districts. The total national population was 45,905,417.”

Walekhwa et al. (2026) “Research in Context”: “Previous outbreaks in Uganda (2007–08; 131 confirmed cases, 42 deaths) and DR Congo (2012; 38 confirmed cases, 13 deaths) suggested lower transmissibility and case-fatality ratio than Zaire ebolavirus, with reproduction number estimates ranging from 1.2 to 2.6.”

Walekhwa et al. (2026) “Research in Context”: “Concurrently, Bangelesa and colleagues conducted a population mobility mapping exercise in Ituri Province using monthly data from Flowminder (March 2025–March 2026) and computed a normalised mobility intensity index (MII) for all 340 health zones. Their analysis revealed that Ituri Province had the second highest average MII nationally (25.3), behind only Kinshasa (26.6), and 2.4 times the national median (10.7). Critically, the three outbreak epicentre health zones, Rwampara (MII 68.6, rank 2), Bunia (MII 62.6, rank 3), and Mongbwalu (MII 49.9, rank 11), all exceeded the 95th national percentile (41.4), underscoring the exceptional connectivity of the affected zones.”

Okingo, Iroungou, & Valdano (2026) “Abstract”: “During the ongoing Bundibugyo ebolavirus disease outbreak in the Democratic Republic of the Congo, we quantify introduction risk and sustained-transmission potential across the country. This identifies priority zones far from currently affected areas, where rapid amplification could follow introduction and response efforts should be focused.”

Okingo, Iroungou, & Valdano (2026) “Body”: “We used the method developed in Ref. 2 to estimate, for each DRC health zone (HZ), the potential for an introduced case to generate sustained transmission (hereafter, sustained-transmission potential). We used available population and mobility data to parameterize spatial transmission, testing a broad range of assumptions on disease transmissibility and on how local and between-community contact patterns can be inferred from these data. We ranked HZs by sustained-transmission potential and compared them with introduction-risk ranks estimated from mobility data.”

Okingo, Iroungou, & Valdano (2026) “Body”: “Crossing the two rankings identified eight HZs that combined high introduction risk and high sustained-transmission potential (Fig. 1E, F): Binza Ozone and Binza Meteo in Kinshasa; Karisimbi in Nord-Kivu; Kadutu and Ibanda in Sud-Kivu; and Makiso-Kisangani, Kabondo, and Mangobo in Tshopo.”

Okingo, Iroungou, & Valdano (2026) “Body”: “Introduction risk was highest in HZs close to the reported foci, specifically in Ituri, Nord-Kivu, and Haut-Uele, consistent with expected mobility patterns (Fig. 1A,B). Non-negligible risk of introduction also appeared in distant urban centers, including the Kinshasa area.”

Okingo, Iroungou, & Valdano (2026) “Body”: “By contrast, high sustained-transmission potential was distributed across DRC, including Kinshasa, Kasaï-Oriental, Kongo-Central, Tshopo, Nord-Kivu, Haut-Katanga, Sud-Kivu, Kasaï, and Kasaï-Central (Fig. 1C,D). Notably, several are far from the current outbreak.”

Okingo, Iroungou, & Valdano (2026) “Body”: “We used the method developed in Ref. 2 to estimate, for each DRC health zone (HZ), the potential for an introduced case to generate sustained transmission (hereafter, sustained-transmission potential).”

Okingo, Iroungou, & Valdano (2026) “Body”: “We used available population and mobility data to parameterize spatial transmission, testing a broad range of assumptions on disease transmissibility and on how local and between-community contact patterns can be inferred from these data.”

Okingo, Iroungou, & Valdano (2026) “Figure 1: Introduction risk and sustained-transmission potential across health zones.”: “Population data are from worldpop.org. Mobility data are from flowminder.org.”

Okingo, Iroungou, & Valdano (2026) “Body”: “We ranked HZs by sustained-transmission potential and compared them with introduction-risk ranks estimated from mobility data.”

Okingo, Iroungou, & Valdano (2026) “Body”: “Unlike risk estimates, risk rankings were robust across scenarios and assumptions and can therefore inform action despite limited knowledge of transmissibility and contact patterns.”

Okingo, Iroungou, & Valdano (2026) “Body”: “This requires estimating introduction risk and, more importantly, the potential for sustained onward transmission, despite limited data on human mobility, contact patterns, disease transmissibility, tracing and isolation capacity.”

Okingo, Iroungou, & Valdano (2026) “Body”: “Insecurity, high population mobility, limited healthcare capacity, and the absence of a vaccine pose challenges to response.”

Okingo, Iroungou, & Valdano (2026) “Body”: “This is relevant because several high-amplification areas are in Kinshasa, Kasaï, and Tshopo, where preparedness may be more feasible than in parts of eastern DRC despite heterogeneous healthcare access, accessibility constraints, and low-to-moderate insecurity outside Kinshasa.”

Okingo, Iroungou, & Valdano (2026) “Figure 1: Introduction risk and sustained-transmission potential across health zones.”: “Methodology, additional data sources, additional results and tabled values are available in the Appendix.”

Okingo, Iroungou, & Valdano (2026) “Body”: “An outbreak of Bundibugyo virus disease is ongoing in Ituri province, Democratic Republic of the Congo (DRC).”

Okingo, Iroungou, & Valdano (2026) “Body”: “As of June 4, confirmed cases in Nord-Kivu and Sud-Kivu indicate that the spread has gone beyond the initial focus.”

Kinoshita, Suzuki, & Yoneoka (2026) “Abstract”: “During the 2026 Bundibugyo virus disease outbreak in the Democratic Republic of the Congo and Uganda, we projected potential airline-mediated importation risk using contemporary airline network and an externally calibrated Ebola importation hazard.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Airport-level arrival probabilities, interpreted as the probability of at least one airline-mediated importation until the end of June, 2026 ( t = 44), varied substantially across the global airline network (Figure 2). The probability ranged from 0.247 to 0.937 (mean of 0.431).”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Higher arrival probabilities were concentrated among airports with shorter effective distances from FIH and stronger connectivity through major transit hubs: for example, 0.889 in Brussels Airport, 0.888 in Paris Charles de Gaulle Airport, and 0.871 in Johannesburg O.R. Tambo International Airport (Supplementary Table 1).”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “The highest 30-day country-level arrival probabilities were observed in major international hub countries including Belgium, France, South Africa, Kenya, and the United Arab Emirates, with the top-ranked countries all exceeding 0.77, whose probaility ranged from 0.38 to 0.98 in the sensitivity analysis.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “The sensitivity analysis shows that the probability had a similar geographical tendency, even if the starting airport is assumed to be EBB: the mean (standard deviation) ranged from 0.170 (0.049) to 0.616 (0.099) (Supplementary Figures 1-5 and Supplementary Table 2).”

Kinoshita, Suzuki, & Yoneoka (2026) “Abstract”: “We projected the potential airline-mediated importation risk during the 2026 DRC-Uganda outbreak using a hazard-based model informed by contemporary airline transportation networks and a hazard parameter calibrated from the 2013-2016 West African Ebola virus disease (EVD) epidemic, one of the few EBOD outbreaks with documented long-distance airline-mediated international exportations.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “To model the global spread on airline network, we first constructed a directed, weighted air-traffic network using airport-level origin-destination passenger-flow data from OAG for May 2024. OAG is a global travel data provider listed in the IATA Strategic Partners Directory and provides aviation schedules, status data, and analytics products for the air-travel ecosystem. The network includes 11,264 airports (nodes), 569,741 routes (edges) and 329,326,078 passengers.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Given a single airport-level adjacency matrix A , where A ij denotes the number of passenger volume travelling from airport i to j during May 2024, the shortest-path effective distance from Kinshasa/N’djili International Airport (IATA code: FIH) to airport j was defined by , where Γ ij indicates the set of all directed paths from airport i to j and γ is one path. Shortest paths were calculated on the directed network using Dijkstra’s algorithm.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Then, to estimate the arrival probability using the effective distance, we modeled the arrival hazard at airport j as λ j = θ/ D j , where θ = 0.19 is used as the primary value based on a previous EBOD importation hazard model. The arrival probability that infection reaches airport j within t days is calculated by p j ( t ) = 1 - exp(- tλ j ).”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Further, the country-level arrival risk was summarized as the maximum airport-level probability within each country: , where A c denotes the set of airports located in country c .”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “As the sensitivity analysis, we checked θ = {0.06, 0.32} based on the 95% confidence interval of.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “In the sensitivity analysis, δ can be also estimated from, where r is the epidemic growth rate ( r = {0.034, 0.05, 0.1} estimated as of May 17, 2026, γ e is the Euler-Mascheroni constant ( γ e ≈ 0.58), and α FIH is the outbound passenger flow from FIH divided by the assumed FIH catchment population α FIH ≈ 0.0001.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Here, we assume that the relationship between effective distance and EVD importation observed in 2013-2016 was transferable to the current BVD outbreak. The estimated effective distance distribution was very similar to that in the previous study, supporting the use of same θ value on a comparable hazard scale.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “In the main analyses, disease-specific values of δ was set to 1, which assumes that the each additional travel step contributes one unit of distance, and thus corresponds to the Brockmann-Helbing scale. This scale has been empirically shown to effectively predict the global spread patterns of infectious diseases.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “In the sensitivity analysis, δ can be also estimated from, where r is the epidemic growth rate ( r = {0.034, 0.05, 0.1} estimated as of May 17, 2026, γ e is the Euler-Mascheroni constant ( γ e ≈ 0.58), and α FIH is the outbound passenger flow from FIH divided by the assumed FIH catchment population α FIH ≈ 0.0001. While this δ formulation allows us to incorporate information about the current epidemic situation by adding more parameters, it also increases uncertainty.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Sensitivity analyses using the lower and upper bounds of the historical parameter changed absolute arrival probabilities but broadly preserved the overall pattern of higher-probability airports. This is expected because effective distance determines the spatial pattern, whereas the hazard parameter mainly scales the probabilies.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Our estimates also relied on 2024 airline network data and therefore may not capture subsequent changes in routes, passenger volumes, airspace restrictions, or travel advisories, including those affecting routes through Russia and the Middle East.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “The projections also did not explicitly account for post-PHEIC interventions, enhanced screening or behavioural changes.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “More importantly, our analysis focused on global air travel originationg from FIH, with EBB considered in sensitivity analysis, and did not explicitly account for land-based spread within the DRC or cross-border movements to neighboring countries.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Estimated arrival probability should be interpreted as importation risk rather than the risk of local transmission after arrival. Importation does not necessarily lead to onward spread because EBOD requires direct contact with infectious body fluids or contaminated materials, and timely detection, isolation, contact tracing and infection prevention and control can reduce transmission.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Because no airline-mediated importation had been reported so far as of 28 May 2026, we used a hazard parameter estimated using the 2013-2016 West African EVD epidemic. Here, we assume that the relationship between effective distance and EVD importation observed in 2013-2016 was transferable to the current BVD outbreak.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Our estimates also relied on 2024 airline network data and therefore may not capture subsequent changes in routes, passenger volumes, airspace restrictions, or travel advisories, including those affecting routes through Russia and the Middle East. The projections also did not explicitly account for post-PHEIC interventions, enhanced screening or behavioural changes.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Estimated arrival probability should be interpreted as importation risk rather than the risk of local transmission after arrival.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Thus, countries with high-probability airport gateways should be interpreted as priorities for awareness and preparedness, rather than as locations where local outbreaks are expected.”

Kinoshita, Suzuki, & Yoneoka (2026) “Abstract”: “As of 4 June 2026, the DRC had reported 363 confirmed cases, including 62 deaths. Uganda had reported 15 confirmed cases, including one death.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Because no airline-mediated importation had been reported so far as of 28 May 2026, we used a hazard parameter estimated using the 2013-2016 West African EVD epidemic.”

Kinoshita, Suzuki, & Yoneoka (2026) “Airline-flow-based network and effective distance for hazard estimation”: “Then, to estimate the arrival probability using the effective distance, we modeled the arrival hazard at airport j as λ j = θ/ D j , where θ = 0.19 is used as the primary value based on a previous EBOD importation hazard model.”

Richard et al. (2026) “Abstract”: “In response to reports of high numbers of suspected cases and deaths in the affected countries, CDC assessed the risk posed by the BVD outbreak to the U.S. population during the next 3 months. This analysis used a standardized risk assessment approach that included epidemiologic data from the ongoing outbreak and historical data from previous Ebola outbreaks; the overall risk was determined by taking into account independent assessments of the likelihood of infection and the impact of infection.”

Richard et al. (2026) “Methods”: “Overall risk to the U.S. population was determined by independently assessing two factors: (1) the likelihood of infection and (2) the impact of infection.”

Richard et al. (2026) “Abstract”: “The assessment found that the overall risk to the U.S. population posed by the current BVD outbreak during the next 3 months is low, based on the extremely low likelihood of transmission, despite the high impact that potential infection could have and the resources that would be required to respond to the outbreak.”

Richard et al. (2026) “Results”: “This overall risk was determined based on a combination of extremely low likelihood of infection, but high impact of infection for the U.S. population, were it to occur. The likelihood of Bundibugyo virus infection for the U.S. population was assessed as extremely low.”

Richard et al. (2026) “Results”: “These modeling results suggested that the relative risk of importation to the United States compared with other locations was 1.3%.”

Richard et al. (2026) “Results”: “The impact of infection, based on the standardized framework, was assessed as high, primarily based on the severity of the illness, lack of available medications and vaccines, and resources required to respond to the current outbreak. In the two previously identified outbreaks of BVD in Uganda in 2007 and DRC in 2012, case-fatality rates ranged from 25% to 50%.”

Richard et al. (2026) “Methods”: “CDC subject matter experts in risk assessment methodology, infectious disease modeling, global health, and Ebola disease and viral hemorrhagic fevers collaborated to develop this assessment. These experts used a standardized risk assessment approach that has been applied to previous viral hemorrhagic fever outbreaks. They considered available evidence including epidemiologic data from the ongoing BVD outbreak and historical data on Bundibugyo virus and other Ebola disease outbreaks.”

Richard et al. (2026) “Methods”: “Overall risk to the U.S. population was determined by independently assessing two factors: (1) the likelihood of infection and (2) the impact of infection. † The likelihood of infection refers to the probability that members of the U.S. population would acquire Bundibugyo virus infection during the next 3 months; this, in turn, depends on the likelihood of exposure, infectiousness of the virus, and susceptibility of the population. The impact of infection refers to the consequences of infection in this population. Factors include the severity of disease, level of population immunity to severe disease, availability of medications and vaccines, and necessary public health response resources.”

Richard et al. (2026) “Methods”: “Experts then assigned a degree of confidence to the assessment, taking into account evidence quality, extent, and corroboration of information.”

Richard et al. (2026) “Results”: “Despite the large number of cases identified at the time the outbreak was reported, the current likelihood for potential spread of BVD from DRC to the United States, via travelers from DRC who might be infected, is considered very low based on modeling results that consider population movement. These modeling results suggested that the relative risk of importation to the United States compared with other locations was 1.3%.”

Richard et al. (2026) “Results”: “Despite the large number of cases identified at the time the outbreak was reported, the current likelihood for potential spread of BVD from DRC to the United States, via travelers from DRC who might be infected, is considered very low based on modeling results that consider population movement.”

Richard et al. (2026) “Results”: “If BVD were to be introduced into the United States, based on historical observation and the known epidemiology of BVD, secondary transmission would likely be minimal.”

Richard et al. (2026) “Results”: “Although BVD symptoms can appear suddenly and might be nonspecific, these public health measures are highly effective against Ebola disease, in part because the average interval between cases is long (10–16 days), and because persons are not known to be infectious before the onset of symptoms.”

Richard et al. (2026) “Methods”: “Risk was assessed only for the general U.S. population; however, subpopulations that might have different assessments of likelihood are noted, based on varying risk factors.”

Richard et al. (2026) “Results”: “Although the likelihood of infection for the general U.S. population is low, the likelihood of infection might be higher among U.S. health care workers practicing in or who have recently returned from affected regions in DRC and Uganda based on possible exposure risks.”

Richard et al. (2026) “Results”: “In the two previously identified outbreaks of BVD in Uganda in 2007 and DRC in 2012, case-fatality rates ranged from 25% to 50%. However, many of the deaths in these outbreaks occurred in locations where health resources are limited; clinical outcomes might improve with the specialized care available in the United States.”

Richard et al. (2026) “Results”: “However, this assessment also recognizes uncertainties about the epidemiology of BVD, the scope and geographic spread of the outbreak, and the potential timelines for implementation of interventions.”

Richard et al. (2026) “Discussion”: “First, given that only two previous BVD outbreaks have occurred, less is known about Bundibugyo virus than other types of orthoebolaviruses that cause human illness (e.g., O. zairense, which was responsible for the 2014–2016 Ebola virus disease outbreak in West Africa).”

Richard et al. (2026) “Discussion”: “Second, limited confirmatory diagnostic testing and challenges in contact tracing that likely resulted in undetected transmission mean that the full scope of the current outbreak is unclear, making it difficult to assess the potential geographic spread of the outbreak and the subsequent potential for spread to the United States.”

Richard et al. (2026) “Discussion”: “Finally, although DRC and neighboring countries have extensive experience responding to outbreaks of viral hemorrhagic fevers, several features of this outbreak pose challenges to assessing its future trajectory, including that the outbreak is occurring in a region with ongoing conflict and unpredictable infrastructure, and the initial large size of the outbreak combined with limited confirmatory diagnostic testing might lead to challenges in infection prevention and control efforts.”

Richard et al. (2026) “Abstract”: “Limitations to this assessment included uncertainties around the epidemiology of BVD as well as the current and future scope and geographic spread of the outbreak.”

Richard et al. (2026) “Introduction”: “As of June 2, 2026, a total of 378 confirmed BVD cases and 63 deaths have been reported in the Democratic Republic of the Congo (DRC) and Uganda.”

Richard et al. (2026) “Methods”: “They considered available evidence including epidemiologic data from the ongoing BVD outbreak and historical data on Bundibugyo virus and other Ebola disease outbreaks.”

Richard et al. (2026) “Results”: “In the two previously identified outbreaks of BVD in Uganda in 2007 and DRC in 2012, case-fatality rates ranged from 25% to 50%.”

Richard et al. (2026) “Results”: “Only 11 persons infected with Ebola disease have ever been treated in the United States; all were associated with the 2014–2016 Ebola virus disease outbreak in West Africa. Despite two instances of secondary transmission to U.S. health care workers during that outbreak, no community spread occurred in the United States.”

Fanelli et al. (2026b) “Abstract”: “Using International Air Transport Association origin-destination passenger flows, we assessed relative exposure to Ebola virus disease importation into Europe under six outbreak expansion scenarios reflecting plausible pathways of geographical spread, including cross-border transmission and amplification in highly connected regional capitals.”

Fanelli et al. (2026b) “Abstract”: “Across all scenarios, France, Italy, and the United Kingdom remained among the most exposed countries.”

Fanelli et al. (2026b) “Text”: “Potential spread across the Ugandan border produced no change in either the relative exposure of European countries (Fig. 1b) or their overall importation pressure (Fig. 1c), compared with the currently reported localized transmission in eastern DRC, with Italy, France, Belgium, and the UK being the most exposed.”

Fanelli et al. (2026b) “Text”: “Expansion across the border to South Sudan produced a first structural increase in importation pressure through the connectivity associated to Juba (5-fold increase for Northern Europe and 6-fold increase for Eastern Europe relative to localized transmission in DRC), while still leaving UK, France, and Italy among the highest-risk countries.”

Fanelli et al. (2026b) “Text”: “Kampala and Kigali amplification increased exposure in Northern Europe, with Kampala producing a >120-fold increase in importation pressure relative to localized transmission in DRC. Kinshasa amplification instead primarily reinforced exposure in Western Europe (~80-fold increase).”

Fanelli et al. (2026b) “Appendix”: “Results do not show notable differences with those reported in the main text (Figure S2, S3), with the exception of the Kampala amplification scenario, in which the increase in importation pressure for Northern Europe rises to approximately 190-fold when using monthly averages of 2024 travel data.”

Fanelli et al. (2026b) “Appendix”: “We use origin-destination (O&D) passenger flows derived from the 2024 International Air Transport Association (IATA) database to reconstruct intercontinental and regional travel patterns at the airport level. The IATA O&D data capture complete passenger itineraries, including traffic travelling via connecting flights rather than only direct flight segments.”

Fanelli et al. (2026b) “Appendix”: “The relative importation risk to country c was estimated as the proportion of total outgoing passenger flow from the selected seed airports directed to the country, for each scenario: where $w_{ij}$ denotes the origin-destination passenger flow from airport $i$ , among the set $O_S$ of origin airports included in scenario $S$ , and airport $j$ in the set $D_c$ of airports of the destination destination country $c$. The relative risk of importation for each European country was normalized to the total passenger flow from the source airports to Europe.”

Fanelli et al. (2026b) “Appendix”: “Risks were then stratified by subregion.”

Fanelli et al. (2026b) “Appendix”: “To account for potential seasonal effects in travel patterns, we conducted the analysis using average monthly data from the full 2024. Results do not show notable differences with those reported in the main text (Figure S2, S3), with the exception of the Kampala amplification scenario, in which the increase in importation pressure for Northern Europe rises to approximately 190-fold when using monthly averages of 2024 travel data.”

Fanelli et al. (2026b) “Text”: “We considered six outbreak expansion scenarios designed to reflect plausible pathways of geographical spread from the currently affected area through porous borders (Fig. 1a, Appendix).”

Fanelli et al. (2026b) “Appendix”: “All analyses presented in this report are based on the cumulative passenger traffic reported for May, June, and July 2024. Data from 2024 showed moderate seasonal variability in passenger traffic to Europe. Relative to the annual traffic to Europe, mean passenger volumes during MayJuly ranged from 3% below to 16% above the monthly 2024 average across scenarios.”

Fanelli et al. (2026b) “Text”: “Scenarios 4-6 explored hypothetical amplification assuming community transmission in major urban hubs-Kampala, Kigali, and Kinshasa.”

Fanelli et al. (2026b) “Text”: “Patterns of Ebola virus importation risk to Europe may rapidly shift under uncertainty surrounding the geographical extent and evolution of the outbreak, particularly if transmission expands into major urban hubs with strong international connectivity near the affected area.”

Fanelli et al. (2026b) “Text”: “As of 29 May 2026, the Bundibugyo Ebola outbreak, a Public Health Emergency of International Concern, involved 125 confirmed cases (17 deaths), and 906 suspected cases (223 deaths) across Ituri, North Kivu, and South Kivu provinces in Eastern DRC, and 9 confirmed cases (1 death) in Uganda.”

Fanelli et al. (2026b) “Appendix”: “All analyses presented in this report are based on the cumulative passenger traffic reported for May, June, and July 2024.”

Fanelli et al. (2026b) “Appendix”: “Relative to the annual traffic to Europe, mean passenger volumes during MayJuly ranged from 3% below to 16% above the monthly 2024 average across scenarios.”

van Zandvoort et al. (2026) “Ebola disease cases outside Africa”: “We conducted internet searches for all laboratory-confirmed Ebola disease cases presenting outside Africa, performing both manual and artificial intelligence (AI)-assisted internet searches to identify scientific articles, public health bulletins and news reports without any time limit.”

van Zandvoort et al. (2026) “Summary”: “Following the Bundibugyo virus disease outbreak reported in the Democratic Republic of the Congo in May 2026, we reviewed all known Ebola disease cases outside Africa and found that intercontinental transmission risk remains low.”

van Zandvoort et al. (2026) “Low overall risk of exportation”: “This gives a crude risk estimate of 0.81 cases outside Africa, including medically evacuated, latent and secondary cases, per 1,000 Ebola disease cases in a source outbreak (Table 2). Focusing on latent cases and their secondary cases only, there have been 0.17 cases outside Africa per 1,000 Ebola disease cases in the source outbreak.”

van Zandvoort et al. (2026) “Low overall risk of exportation”: “Of note, this represents the empirically observed risk of importation (and onward transmission) across all countries outside Africa; per-country risks are smaller. For example, focusing only on the UK, there has been one latent Ebola disease case since 2000, giving a risk estimate of 0.03 latent UK cases for every 1,000 Ebola disease cases in a source outbreak.”

van Zandvoort et al. (2026) “Decrease in exportation risk over the 2014–16 epidemic”: “Over the course of the epidemic, the case exportation risk per reported source outbreak case decreased by 4.6% per week (95% CI: 1.7–7.4%; p = 0.0097). More generally, the case exportation risk was substantially lower after the epidemic peak (0.46; 95% CI: 0.17–0.73 exported cases per 1,000 source cases from 27 October 2014 onwards) than it was before the peak (1.7; 95% CI: 0.95–2.7 exported cases per 1,000 source cases).”

van Zandvoort et al. (2026) “Summary”: “We identified and analysed all known Ebola disease cases outside Africa to contextualise intercontinental spread and assess the risk of undetected Orthoebolavirus transmission outside Africa. We conducted internet searches for all laboratory-confirmed Ebola disease cases presenting outside Africa, performing both manual and artificial intelligence (AI)-assisted internet searches to identify scientific articles, public health bulletins and news reports without any time limit.”

van Zandvoort et al. (2026) “Low overall risk of exportation”: “As a crude measure of risk, we calculated for Ebola disease outbreaks since the year 2000 the number of cases occurring outside Africa per 1,000 reported cases in the originating epidemic. This gives a crude risk estimate of 0.81 cases outside Africa, including medically evacuated, latent and secondary cases, per 1,000 Ebola disease cases in a source outbreak (Table 2). Focusing on latent cases and their secondary cases only, there have been 0.17 cases outside Africa per 1,000 Ebola disease cases in the source outbreak.”

van Zandvoort et al. (2026) “Decrease in exportation risk over the 2014–16 epidemic”: “For cases exported from the 2014–16 epidemic in Western Africa, we carried out a time-varying risk analysis using Poisson regression of the number of exported cases for a given epidemiological week and country (across Guinea, Liberia and Sierra Leone, including weeks with no exported cases, from 6 January 2014 to 20 December 2015) against (i) the number of cases reported for that epidemiological week and country and (ii) the number of weeks since 6 January 2014 as a linear predictor of time-varying risk.”

van Zandvoort et al. (2026) “Decrease in exportation risk over the 2014–16 epidemic”: “The number of exported cases (both medically evacuated and latent) closely scaled with the number of cases in the source country in that week (Figure 2), with no significant deviation from linear scaling (estimated scaling: 0.98; 95% CI (confidence interval): 0.60–1.36; p = 0.86).”

van Zandvoort et al. (2026) “Discussion”: “A key distinction in our analysis is between medically evacuated and latent Ebola disease cases. While medically evacuated cases represent known risks that can be mitigated with strict biosecurity measures, latent cases arriving via commercial flights need to be diagnosed before they can be isolated.”

van Zandvoort et al. (2026) “Discussion”: “Agreement among search methods and with a clinical review of exported cases from 2014 to 2016 support that our list is likely complete. Nonetheless, potential sources of under-ascertainment include that public health bulletins published not in English were underrepresented despite targeted native language searches, and web-based search tools may have returned geographically biased results.”

van Zandvoort et al. (2026) “Discussion”: “Nonetheless, potential sources of under-ascertainment include that public health bulletins published not in English were underrepresented despite targeted native language searches, and web-based search tools may have returned geographically biased results.”

van Zandvoort et al. (2026) “Decrease in exportation risk over the 2014–16 epidemic”: “As most exported primary cases during this epidemic (20 of 24) were medical evacuations, this is likely to be most consistent with improvement in infection prevention and control measures coinciding with the stepping-up of response efforts in late 2014, as further evidenced by decreasing transmission after the peak of the epidemic.”

van Zandvoort et al. (2026) “Low overall risk of exportation”: “Of note, this represents the empirically observed risk of importation (and onward transmission) across all countries outside Africa; per-country risks are smaller.”

van Zandvoort et al. (2026) “Discussion”: “Notably, the four latent cases we identified occurred among 300,000 travellers screened by the US Centers for Disease Control and Prevention (CDC)-supported exit screening programmes in Guinea, Liberia and Sierra Leone, and all were asymptomatic (and hence undetectable) at the point of both exit screening and entry screening. Screening protocols included clear guidance on procedures for self-monitoring of symptoms, which ultimately led to rapid confirmation and isolation among the three latent cases in medical workers.”

van Zandvoort et al. (2026) “Ebola disease cases outside Africa”: “We identified 28 confirmed Ebola disease cases outside Africa: 25 primary imported cases and three secondary cases infected by another patient in the United States (US) or Europe (Table 1, Figure 1). Of these, 27 occurred during the 2014–16 Ebola virus epidemic in Western Africa and one has occurred so far during the ongoing 2026 Bundibugyo virus epidemic.”

van Zandvoort et al. (2026) “Ebola disease cases outside Africa”: “In total, we included 27 scientific articles, 9 reports in public health bulletins, 82 news articles and 9 other sites in gathering data.”

van Zandvoort et al. (2026) “Decrease in exportation risk over the 2014–16 epidemic”: “Weekly count of new confirmed, probable and suspected cases, by week and country for Guinea, Liberia and Sierra Leone (total 28,592).”

van Zandvoort et al. (2026) “Discussion”: “Notably, the four latent cases we identified occurred among 300,000 travellers screened by the US Centers for Disease Control and Prevention (CDC)-supported exit screening programmes in Guinea, Liberia and Sierra Leone, and all were asymptomatic (and hence undetectable) at the point of both exit screening and entry screening.”

van Zandvoort et al. (2026) “Ebola disease cases outside Africa”: “We checked our findings against existing lists of Ebola disease cases.”