Hantavirus global briefing — August 12, 2026: Spain expands monitoring to six while a new mortality model stays unvalidated

Executive read

Spain’s imported Andes hantavirus response widened on 11 August: four additional close contacts were identified, bringing the monitored total to six. Two were passengers on the 18 July Santiago–Madrid flight, and two travelled on the 19 July Madrid–Bordeaux route. All six were asymptomatic and under home quarantine; microbiological testing for the four newly identified contacts was still pending in the health ministry report. 1

The same report says the index traveller and his two original household contacts had negative follow-up PCR tests. The traveller was discharged, while the two contacts continued isolation under the 42-day protocol counted from 28 July. Spain’s assessment put the general-population risk at "extremely low" because Andes-virus transmission generally requires close, prolonged contact and the virus is not known to circulate in Spanish reservoirs. 1

The change is therefore a wider contact net, not a wider transmission chain. Four newly identified contacts add follow-up work, but there is no reported symptomatic contact or positive secondary test in this update.

A new paper published on 11 August tests machine-learning models for hantavirus mortality risk. It reports an AUC of 0.858 for logistic regression and 0.857 for XGBoost on a 10,000-record dataset, but the authors describe the analysis as retrospective and internal. They also say the dataset is synthetic or repository-based and that external, temporal, and country-level validation is still required. The paper is a methods watchpoint for triage research, not a deployable clinical tool. 2

Signal	Latest position	Meaning for today’s watch
Spain	Six close contacts are monitored across Castilla y León, Madrid, and Catalonia; all are asymptomatic and quarantined at home. Four new contacts still awaited microbiological testing in the 10 August CCAES report. 1	The response has expanded geographically, but no secondary infection is reported.
Galicia household	The index traveller and two original household contacts had negative follow-up PCR tests. The contacts remain isolated under the 42-day protocol from 28 July. 1	A negative test narrows the immediate concern; it does not end exposure-based monitoring.
Chile	The latest readable balance remains 46 infections and 18 deaths through 28 July. A Rengo child’s Andes-virus diagnosis is reported at case level, but the accessible report does not reconcile it with the national denominator. 3	Keep the confirmed case and the national count as separate facts until Minsal or ISP publishes a revised table.
WHO / ECDC / CDC	WHO’s closed M/V Hondius event remains at 13 cases and three deaths. ECDC week 32, covering 1–7 August, lists no hantavirus topic. The CDC hantavirus-specific HAN update retrieved here is dated 18 May and is not a current US surveillance count. 4 5 6	No new escalation is verified in these official source boundaries; this does not prove zero infections elsewhere.

Spain: six contacts, three routes, no secondary case reported

The new Spanish figure is six close contacts in total. Four were added after French authorities notified Spain about people who shared travel segments with the traveller. Two were passengers on the Santiago–Madrid flight on 18 July. They live in Castilla y León and the Community of Madrid, and those regional public-health authorities are following them. Two others shared the Madrid–Bordeaux journey on 19 July and are quarantining at home in Catalonia. 1

The CCAES report was dated 10 August, and the four new contacts were still awaiting microbiological tests when the report was written. All six contacts were asymptomatic and under home quarantine. That wording matters: being counted as a close contact is an exposure classification, not a diagnosis. A positive result, compatible illness, or a documented transmission event would change the interpretation; this update reports none of those.

The earlier household response is also still active. The traveller was discharged after a negative follow-up PCR. His two family contacts were also negative and were discharged from the shared isolation setting, but they continued isolation because Spain’s protocol counts 42 days from 28 July. 1

The operational picture has two layers:

Case management: the traveller is no longer hospitalised, and the two initial household contacts tested negative.
Exposure management: six contacts remain under public-health follow-up, with four new microbiological assessments pending in the CCAES report.

Spain’s risk statement is correspondingly narrow. The report says the risk to healthcare workers is currently very low and the risk to the general population is extremely low outside the care setting. It attributes the low general risk to the need for close, prolonged contact for Andes-virus person-to-person transmission and to the absence of known Andes-virus reservoirs or circulation in Spain. 1

For the next update, the useful questions are not simply whether the contact count rises again. They are whether any of the four new contacts develops symptoms, whether tests return positive, and whether the 42-day clock is applied consistently from the last relevant exposure. A larger list with negative tests would mean broader tracing success; a positive contact would be a different epidemiological signal.

Chile: keep the case signal separate from the denominator

Chile remains the main severity watch in the Americas, but today’s evidence does not support rewriting its national count. A 9 August report says the Chilean Institute of Public Health confirmed Andes virus in an 11-year-old from Rengo who later died in Rancagua. The same report cites the latest Ministry of Health balance available to it: 46 confirmed cases and 18 deaths through 28 July, with a fatality rate near 39%. It does not say whether the Rengo death was already included or has been added. 3

A separate UPI report describes Chile’s first preventive hantavirus health alert across 13 of 16 regions. It reports the same 46-infection, 18-death balance, describes hantavirus as endemic rather than an epidemic, and lists stronger surveillance, resource management, specialist hiring, health controls, and direct procurement as response measures. 7

That leaves a familiar but important denominator problem: the Rengo diagnosis is a confirmed case-level development, while 46/18 is still the latest readable national table cited by the sources retrieved here. The defensible daily position is confirmed Rengo case; national denominator unreconciled. A new national total should wait for an explicit Minsal or ISP revision.

UPI’s reporting also keeps the exposure interpretation grounded. Chilean specialists linked risk to the long-tailed pygmy rice rat and to interacting environmental and human factors, including food availability, rainfall and vegetation cycles, outdoor activity, and settlement expansion. Those are plausible contributors described by the specialists; they are not a measurement of a new rodent outbreak or proof of a single cause for the 2026 fatality pattern. 7

Official status: closure, bulletin boundary, and an old US advisory

WHO’s 2 July statement remains the closure anchor for the M/V Hondius event. WHO said the final exposed contact completed quarantine and tested negative, no further cases had been reported since 25 May, and the outbreak was over at 13 cases including three deaths. More than 650 contacts had been followed in 33 countries and territories. 4

ECDC’s latest readable CDTR page was published on 7 August and covers 1–7 August. Its summary names Ebola, chikungunya, dengue, West Nile virus infection, Crimean-Congo haemorrhagic fever, Vibrio, respiratory-virus epidemiology, World Pride, MERS, and expert deployments; hantavirus is not named. This is a statement about that bulletin, not a claim that Europe had no hantavirus infections. 5

CDC HAN-00529, dated 18 May, says no confirmed US Andes-virus cases associated with the cruise outbreak had been reported in the United States as of that date. It also tells clinicians to consider Andes virus after exposure to the ship or its contacts, and other hantaviruses after rodent exposure. Because the notice is historical and no newer hantavirus-specific CDC HAN notice was verified in this pass, it should not be read as a current US case count. 6

Taken together, the official status has not escalated to a new multinational event. Spain’s contact tracing is active, Chile’s alert remains an operational response to an endemic high-severity signal, and the ship-linked outbreak remains closed.

Research digest: a promising model that is not ready for a ward

Interpretable Machine Learning for Mortality Risk Stratification in Hantavirus Infection

Authors and source: Yetunde O. Enigbokan, Akeem A. Abiona, and Micheal O. Ajinaja; Cureus Journal of Computer Science; published 11 August 2026; DOI 10.7759/s44389-026-00217-5. The journal page labels it an original, peer-reviewed article. 2
Design and data: Retrospective supervised-learning analysis of 10,000 records combining HFRS and HCPS. The dataset contained 8,938 recovered and 1,062 deceased records, with 6,460 HFRS and 3,540 HCPS cases. The authors describe the data as synthetic or repository-based rather than as a direct hospital registry. 2
Models and outcome: Logistic regression, random forest, and XGBoost were compared on a held-out test set. The outcome was mortality, and the paper reports ROC-AUC, accuracy, precision, recall, F1 score, Brier score, and SHAP-based interpretation. 2

The headline metric is not a single winner. The models make different trade-offs:

Model	ROC-AUC (95% CI)	Recall for death (95% CI)	Precision (95% CI)	Brier score (95% CI)
Logistic regression	0.858 (0.834–0.882)	0.741 (0.692–0.788)	0.321 (0.282–0.362)	0.119 (0.108–0.131)
XGBoost	0.857 (0.832–0.881)	0.321 (0.278–0.368)	0.511 (0.468–0.556)	0.076 (0.068–0.085)
Random forest	0.847 (0.821–0.872)	0.085 (0.058–0.118)	0.462 (0.418–0.508)	0.076 (0.068–0.085)

The paper’s interpretation is clinically familiar. Logistic regression catches more of the fatal cases but produces more false positives. XGBoost gives a similar ranking of risk with better precision and calibration, but misses more deaths at the reported threshold. Random forest’s 0.893 accuracy looks strong until the 10.62% mortality prevalence and its 0.085 recall are considered. The paper used 1,000 bootstrap resamples for the confidence intervals and reports that the AUC difference between logistic regression and XGBoost was not statistically significant. 2

ROC curves comparing logistic regression, random forest, and gradient boosting in the study — *The original ROC figure shows close discrimination among the three models; the larger differences are in recall, precision, and calibration. 2*

What it adds to this channel: the paper gives a concrete way to discuss care-capacity signals without pretending that a case count alone predicts who will deteriorate. A future validated model could help prioritise monitoring or transfer when HFRS and HCPS patients present with different severity patterns. That is a research direction, not an instruction to use these coefficients in current Chilean or Spanish care.

What prevents deployment: the evaluation used one source dataset and one held-out split, not independent hospitals, countries, or future outbreaks. The authors flag possible leakage-prone variables and say the final model should use only information available at presentation or admission. They call for external, temporal, country-stratified, and prospective validation, plus threshold optimisation. The dataset’s synthetic or repository-based character further limits how directly the performance can be transported to clinical records. 2

The practical reading is therefore modest: this is a useful methods paper for deciding what a hantavirus triage model should report, especially recall and calibration. It is not evidence that a model can yet forecast the outcome of the Rengo case, the Aysén patients, or any of Spain’s contacts.

Reservoir and response watch

No new rodent-abundance survey or reservoir-infection estimate was verified in the 11–12 August monitoring window. The current actionable evidence remains the public-health response: Spain is tracing travel-linked contacts and keeping them in home quarantine, while Chile is using its alert to strengthen surveillance and health-system readiness.

For individual exposure prevention in Chile’s endemic areas, the latest readable regional guidance advised sealing rodent entry points, storing food in closed containers, ventilating rodent-contaminated rooms for 20–30 minutes, and using wet chlorine disinfection rather than dry sweeping. Those measures reduce contact with contaminated material; they do not imply that a person-to-person chain is active. 8

Watchpoints for the next briefing

Spain’s four new contacts: pending microbiological results, symptoms, and whether any additional travel-linked contact is added.
The 42-day protocol: whether the two original household contacts complete monitoring without symptoms or a positive test.
Chile’s denominator: an explicit Minsal or ISP table that includes or excludes the confirmed Rengo death.
Aysén and O’Higgins care demand: new cases, outcomes, transfers, and any change in regional severity.
Official escalation: a new WHO Disease Outbreak News item, an ECDC bulletin that names hantavirus, or a current CDC Andes-virus notice.
Model validation: independent clinical data, baseline-only predictors, calibration testing, and prospective evaluation of the new mortality-risk approach.

Bottom line

Spain’s hantavirus response is larger than yesterday’s contact count suggested, but its epidemiological signal is still containment work rather than documented secondary spread: six asymptomatic contacts are under home quarantine, with testing for four newly identified contacts pending in the CCAES report.

The new mortality-model paper is worth opening for its metrics and its warnings. Logistic regression prioritised recall; XGBoost improved precision and calibration; neither was externally validated. Chile’s 46/18 balance therefore remains the latest readable national denominator, the Rengo diagnosis remains a confirmed case-level signal, and the WHO ship-linked outbreak remains closed at 13 cases and three deaths.