agriculture · health · family: success's shadow
the vaccine hid the signal
France vaccinated 35 million ducks and bird flu outbreaks fell from 1,374 to 10 — now it must prove absence of a virus the vaccine has made invisible
Problem statement
France began preventive vaccination of domestic ducks against highly pathogenic avian influenza in October 2023; by 1 July 2024 more than 35 million ducks had received two doses and 1.5 million a third. Reported H5 poultry outbreaks fell from 1,374 in 2021–22 and 396 in 2022–23 to 10 in 2023–24, against a model-predicted 487 outbreaks (95% CI 314–756) in the absence of vaccination. The trouble is what the protection does to the detection system. HPAI outbreaks in unvaccinated flocks are generally detected by passive surveillance — a rise in deaths that a farmer reports — and the absence of mortality or overt clinical disease is, in the EFSA opinion's description, the basic effect of vaccination, so vaccination can remove that signal while infection can still occur (two of the 10 outbreaks were in vaccinated flocks). EU law therefore requires enhanced surveillance of vaccinated poultry both to detect outbreaks early and to demonstrate that the virus is absent from vaccinated areas. EFSA's 2024 opinion assessed those schemes — species by species, sample type by sample type, interval by interval — and its plain-language summary states the problem plainly: "Field studies are scarce on the effectiveness of different surveillance schemes to be applied after vaccination for the early detection of the HPAI virus and to demonstrate freedom from disease."
Why this matters
Most countries rely on poultry depopulation and movement restrictions to control HPAI; France's campaign was followed by a fall in outbreaks that the authors of the field evaluation judge likely attributable to vaccination, and they conclude that the potential of vaccination to reduce HPAI incidence warrants consideration. What a country considering it is actually deciding is whether it can afford, and defend, the surveillance that has to come with it: an active testing programme in vaccinated flocks, sized on modelled assumptions rather than field evidence, whose output is the assurance on which movement of vaccinated poultry and their products depends. If the schemes are less sensitive in practice than the models assume, the failure mode is silent circulation — a virus evolving unobserved in a vaccinated population, with the risk of escape variants and of a late, large detection. If they are set conservatively enough to be safe, the recurring sampling effort may itself become an obstacle to vaccination, as the opinion says of individual animal sampling. Either way the bottleneck on a promising HPAI control tool is not the vaccine; it is the ability to see through it.
What’s been tried and why it hasn’t worked
The design work has been done carefully and is still an argument from models rather than from fields. EFSA assessed surveillance effectiveness using mathematical modelling built from four inputs: its own earlier vaccination opinion, records of primary HPAI introductions in Dutch poultry, serology from experimental trials in Belgium, France, Italy and the Netherlands, and poultry mortality data from France and the Netherlands. From these it derived workable options — for emergency vaccination, molecular testing of dead birds at defined intervals for chicken layers, ducks and turkeys, with molecular plus serological testing of live birds effective in ducks; for preventive vaccination, molecular testing of dead birds with variations in sampling interval and in the proportion of flocks covered (100%, 50%, 25%). It also flagged the load-bearing weaknesses. Passive surveillance in vaccinated flocks still applies as a trigger — testing "if, for example, an unusual increased mortality is observed between two consecutive sampling events" — which is precisely the signal vaccination suppresses, so EFSA recommends keeping passive surveillance in unvaccinated farms inside vaccinated areas "to enhance the overall sensitivity of the surveillance." Uncertainties arise, in EFSA's words, "due to the use of values (real or assumed) retrieved from previously published studies that may not be representative of all scenarios and poultry production systems in Europe" — the full opinion adds that its transmission and clinical parameters may not be applicable to certain chicken layer, duck and turkey production systems in Europe. Serological differentiation of infected from vaccinated animals depends on the vaccine used, so the diagnostic toolkit is not uniform across programmes. And the French field record cannot settle the question either: the authors concede that "the extent to which vaccination contributed to this reduction remains unclear," and although their counterfactual model attributes most of the reduction to vaccination (a 96–99% reduction in epizootic size, "likely attributable to vaccination"), that is an outbreak-count evaluation, not a measurement of surveillance sensitivity in vaccinated flocks. So the schemes rest on parameters drawn from previously published studies, and field studies of how surveillance schemes perform after vaccination are, in the summary's words, scarce.
What would unlock progress
Two unlocks. First, a cheaper, more sensitive sampling substrate than dead birds. Detecting a pathogen in a population whose clinical signal has been suppressed resembles the problem wastewater surveillance takes on for human respiratory viruses: sample the environment rather than the individual. The full opinion names such substrates — swabs from surfaces, drinking water, air samples — and says samples collected from the environment could be a valid, time- and labour-saving alternative to animal samples, but that validated protocols are currently lacking and that Delegated Regulation (EU) No 2023/361 requires clinical sample material in vaccination surveillance programmes; molecular testing of them could raise flock-level sensitivity per euro above the current dead-bird protocols, if the sensitivity were characterised. Second, field validation by design rather than by outbreak: a vaccinated-flock challenge or sentinel study, or a natural-experiment analysis of the French programme's own testing records, would replace assumed parameters with measured ones — how much virus a vaccinated flock sheds, for how long, and what fraction of a scheme's tests would actually catch it. EFSA's opinion itself ends with recommendations for future scientific studies to fill exactly these gaps.
Entry points for student teams
A team could build a flock-level surveillance simulator — vaccinated-flock infection dynamics against candidate sampling schemes — and produce the sensitivity-versus-cost frontier that ministries currently lack, using EFSA's published scheme parameters as the baseline. An instrumentation team could prototype and bench-test an environmental sampler for a poultry house (exhaust dust or drinker-line concentrate) and characterise its limit of detection against a surrogate virus, an unambiguous semester-scale laboratory project. A data team could design the observational study that would validate a scheme from routine records, specifying what a national programme would have to log for the analysis to be possible — a design contribution that is worth more than another model. An economics team could cost the surveillance obligation per bird across production systems and identify at what price vaccination stops being adoptable. Relevant skills: epidemiological modelling, veterinary microbiology, sensor and sampling design, agricultural economics.
Genes
Structural cousins: stuck the same way, in other fields
Sources
EFSA Panel on Animal Health and Animal Welfare (AHAW), European Union Reference Laboratory for Avian Influenza, et al., "Vaccination of poultry against highly pathogenic avian influenza – Part 2. Surveillance and mitigation measures," *EFSA Journal* 2024;22(4):e8755, doi 10.2903/j.efsa.2024.8755; plain-language summary read in full at accessed 2026-08-18. Guinat C, Fourtune L, Lambert S, Martin E, Gerbier G, Pellicer AJ, Guérin J-L, Vergne T, "Promising Effects of Duck Vaccination against Highly Pathogenic Avian Influenza, France, 2023–2024," *Emerging Infectious Diseases* 2025;31(7):1468–1471, (doi 10.3201/eid3107.241445), accessed 2026-08-18 go to source 1 ↗ go to source 2 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
The EFSA plain-language summary of the Part 2 opinion was retrieved and read verbatim on 2026-08-18 (the Wiley full text of doi 10.2903/j.efsa.2024.8755 returned HTTP 402 this session — flagged for deeper sourcing; every EFSA quotation here is verbatim from the plain-language summary page, including "Field studies are scarce…", the uncertainty statement, the passive-surveillance trigger wording, and the sampling options). The EID article was fetched and read for the French campaign figures (35 million ducks two doses, 1.5 million three doses as of 1 July 2024; 1,374 / 396 / 10 outbreaks; 487 predicted, 95% CI 314–756; two of the ten outbreaks in vaccinated flocks) and for the authors' caveat that "the extent to which vaccination contributed to this reduction remains unclear."
The highest-stakes inference in this brief is the masking mechanism — that vaccination suppresses the mortality signal on which passive surveillance depends. This is stated here as the rationale that EU law and EFSA act on (enhanced active surveillance is legally required in vaccinated areas; EFSA recommends passive surveillance in unvaccinated farms in vaccinated areas "to enhance the overall sensitivity of the surveillance"; passive surveillance in vaccinated flocks continues, with unusual mortality as a trigger for additional testing), not as a measured shedding-versus-mortality quantity — indeed, the absence of that measurement is the problem the brief describes.
Related collection briefs, all distinct: `agriculture-colistin-resistance-last-resort-failure` (agricultural drug use driving human AMR), `health-idsr-district-reporting-collapse-africa` (human disease surveillance reporting), `food-safety-vbnc-pathogen-false-negative-detection` (interventions pushing pathogens into a non-culturable state — a related "our control measure blinded our test" structure, but in food processing, not animal-health surveillance).
Source type: Self-articulated (EU risk-assessment agency naming the evidence gap in the scheme it designed), plus an independent field evaluation.
Verified at intake 2026-08-18: gate (net) + adversarial source check + contested-tag second coding. Verifier: EFSA plain-language summary and the EID article both fetched 2026-08-18; every quotation and figure confirmed verbatim (EID: >35 million two doses / 1.5 million three doses as of 1 July 2024; 1,374 / 396 / 10; 487, 95% CI 314–756; two of ten in vaccinated flocks). Added a hedge noting the EID authors' own conclusion that the reduction is likely attributable to vaccination.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against every source it names, with a second reader going over the repairs on the same day, fetching the same three texts independently. The EFSA plain-language summary was read in full at https://www.efsa.europa.eu/en/plain-language-summary/vaccination-poultry-against-highly-pathogenic-avian-influenza-part-2. The Emerging Infectious Diseases article was read in full at https://wwwnc.cdc.gov/eid/article/31/7/24-1445_article. The full EFSA opinion (doi 10.2903/j.efsa.2024.8755), which the intake note says could not be opened, is open access through Europe PMC as PMC11024799 (https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11024799/fullTextXML); its full text was searched by both readers and the passages on clinical signs, passive surveillance, environmental samples, sentinel birds, uncertainty and recommendations were read, but it was not read end to end. (1) The Source line called the author "EFSA Panel on Animal Health and Welfare." The Crossref and Europe PMC records give "EFSA Panel on Animal Health and Animal Welfare (AHAW), European Union Reference Laboratory for Avian Influenza" followed by named authors, EFSA Journal volume 22, issue 4, e8755; the Source line now follows the record. (2) The text called the French vaccination "mandatory." Neither source uses that word; the article says ducks in the production stage are vaccinated and that vaccinating breeder ducks remains optional. The word was removed. (3) The text said poultry surveillance "has always leaned on the cheapest sentinel in agriculture" and that vaccination "is designed to remove exactly that signal." That is not in the sources. The sentence now gives what the full opinion says: outbreaks in unvaccinated flocks are generally detected by passive surveillance, and the absence of mortality or overt clinical disease is the basic effect of vaccination, so vaccination can remove that signal (the opinion adds that degrees of clinical protection vary, so the second reader changed "removes" to "can remove"). (4) The text said the opinion "designed" the schemes and "stated the problem plainly" in the words "Field studies are scarce…". EFSA was asked to assess surveillance strategies, and that sentence is in the plain-language summary, not in the opinion itself; the text now says so. (5) The text said vaccination is "the first intervention in two decades that has visibly bent the HPAI curve in a major poultry producer" and that every other country is deciding whether to follow. Neither claim is in the sources. The sentence now gives what the article says: most countries rely on poultry depopulation and movement restrictions, the authors of the field evaluation judge the fall in outbreaks likely attributable to vaccination, and they conclude that the potential of vaccination to reduce HPAI incidence warrants consideration. (6) The text described the testing programme as "running indefinitely across every vaccinated flock." The sources do not say indefinitely, and the summary's options for preventive vaccination cover 100%, 50% or 25% of vaccinated flocks; the phrase is now "in vaccinated flocks." (7) The text said the cost "may exceed what most producers or ministries will carry, and vaccination stalls at one country," and called vaccination "the most promising HPAI control tool in a generation." Neither is in the sources. The first now gives the opinion's own point that the effort of individual animal sampling is considerable and may itself be an obstacle to vaccination (the opinion speaks of effort, not cost, so the second reader took "cost" out of that clause); the second now says "a promising HPAI control tool." (8) The text said passive surveillance in vaccinated flocks "is reduced to a trigger," and the notes above said it "is reduced to an unusual-mortality trigger." The opinion says passive surveillance is by default applied to vaccinated flocks too, with unusual mortality between sampling events prompting additional testing; the text now says it still applies as a trigger, and the notes say it continues with unusual mortality as a trigger for additional testing. (9) The text glossed the recommendation on unvaccinated farms as "deliberately retaining unprotected sentinels," and the notes above said EFSA "retains unvaccinated sentinels." The opinion uses "sentinels" for unvaccinated birds placed in vaccinated flocks and says their use in vaccination plans is not recommended. The gloss was removed, and the notes now say what the summary says: passive surveillance in unvaccinated farms in vaccinated areas is recommended. (10) The sentence that European poultry "ranges from closed indoor layer houses to free-range duck flocks" is not in the sources; it now gives the opinion's statement that its transmission and clinical parameters may not be applicable to certain chicken layer, duck and turkey production systems in Europe. (11) The text said the article's authors "open by conceding" that the contribution of vaccination is unclear; the sentence is in the third paragraph, so the text now says they concede it. (12) The text said the schemes are "now protecting a €-billion sector" and rest on parameters "nobody has yet measured in a commercial vaccinated flock." No source gives the sector's value, says which schemes are in use, or says nobody has measured. The sentence now says the schemes rest on parameters drawn from previously published studies and that field studies of how surveillance schemes perform after vaccination are, in the summary's words, scarce. (13) The text said wastewater surveillance "solved" the same problem for human respiratory viruses and listed exhaust-air filters, dust, litter swabs, boot swabs and egg-belt residues as sampling substrates. None of this is in the sources. The comparison is kept as a comparison, and the list now gives the opinion's own examples (swabs from surfaces, drinking water, air samples) with its judgement that environmental samples could be a valid, time- and labour-saving alternative to animal samples but lack validated protocols; the second reader added the opinion's note that Delegated Regulation (EU) No 2023/361 requires clinical sample material in vaccination surveillance programmes, since it bears directly on that unlock. Confirmed in the sources and left as written: vaccination began in October 2023; more than 35 million ducks with two doses and 1.5 million with three as of 1 July 2024; 1,374, 396 and 10 outbreaks; 487 predicted (95% CI 314–756); the 96–99% reduction and the words "likely attributable to vaccination"; two of the three duck-farm outbreaks in vaccinated flocks; the four data sources behind the modelling; the surveillance options, including the 100%, 50% and 25% proportions; the EU legal requirement for enhanced surveillance; that serological testing depends on the vaccine used; that the opinion mentions escape mutants; the recommendations for future studies; and every quotation, each of which is word for word in the summary or the article. The Emerging Infectious Diseases citation matches the Europe PMC record. Could not be checked, because no source named here covers them: the statement in the notes above that the Wiley page returned HTTP 402 in August, and the descriptions of the three related collection briefs. The line at the top saying the full opinion was not read is left as written for the editor to decide; the opinion has now been read in part, as described here. No title changed. Genome tags untouched.