Michael Coston
Editor, Senior Moderator
PLoS GPH: Quantifying H5N1 Outbreak Potential and Control Effectiveness in High-risk Agricultural Populations
R[SUB]0[/SUB] (pronounced R-nought) or
Basic Reproduction Number
#19,004
In the spring of 2023, in UK Novel Flu Surveillance: Quantifying TTD, we looked at a technical briefingfrom the UKHSA which attempted to quantify (via statistical modelling) the UK's ability to detect community transmission of a novel flu virus, such as HPAI H5N1.
At that time, the UKHSA estimated it would likely take between 3 and 10 weeks before community spread would become apparent to authorities, after anywhere between a few dozen to a few thousand community infections.
This is their `best case' R[SUB]0[/SUB] 1.2 scenario
Soberingly, this was for the UK; where public health and testing capabilities are far stronger than in a lot of countries currently dealing with HPAI. Even so, detecting community spread early in an outbreak would require a good deal of luck.
With HPAI H5 now primarily a disease of poultry (and in some countries, other livestock), the ideal place to detect, and halt, human transmission would be on the farm.
Today we've a new study from researchers from the UKHSA and the University of Bristol which attempts to model the ability for basic public health tools to control the early spread of HPAI (up to an R[SUB]0[/SUB] of 1.1) among agricultural workers.
The authors first surveyed bird‑exposed farm workers about how many people they met in a day, and then generated computer models of outbreaks starting from a single human infection at various R[SUB]0[/SUB] rates (0 up to 1.1 in steps of 0.1).
Note: The 2023 UKHSA Community Transmission study assumed an R[SUB]0 [/SUB]starting at 1.2, up to 2.0.
What they found was - using somewhat idealized public health tools (contact tracing of symptomatic cases’ contacts & self‑isolation of symptomatic traced contacts) - that when the transmission R[SUB]0 [/SUB]was < 1, the number of cases would remain low; often under 10 cases.
But, as the R[SUB]0 [/SUB] approaches 1.0, or if the number of asymptomatic cases go up, cluster sizes increase.
Due to its length and technical nature, I've only presented a brief overview. Below you'll find the abstract and a few excerpts, but you'll want to follow the link to read the report in its entirety.
I'll have a bit more after the break.
Quantifying H5N1 outbreak potential and control effectiveness in high-risk agricultural populations
Izel Avkan ,Suzanne Gokool,Louise E. Smith,Genevieve Clapp,Rachel Cox,Amy C. Thomas,Ellen Brooks-Pollock
PLOS x Published: December 29, 2025
https://doi.org/10.1371/journal.pgph.0005463
Abstract
Avian influenza is a global public health threat. Since 2021, the ongoing H5N1 panzootic has brought a major shift in H5Nx epidemiology, including unprecedented spread, wide host range and lack of seasonality. Infections in marine mammals, wildlife and livestock have heightened concern for human-to-human transmission and pandemic potential. Contact tracing and self-isolation are used as public health measures in the UK to manage contacts of confirmed human cases of avian influenza.
In this study, we aimed to estimate potential outbreak sizes and evaluate the effectiveness of contact tracing and self-isolation in managing community outbreaks of H5N1 following spillover from birds to people. We characterised contact patterns from an underrepresented agricultural population at high risk of avian influenza exposure through contact with birds (Avian Contact Study).
Informed by these realistic social contact data, we modelled outbreak sizes using a stochastic branching process model. Most simulations resulted in small-scale outbreaks, ranging from 0 to 10 cases.
(SNIP)
Principal findings
Our model provides insights into the potential outbreak size of avian influenza among humans under different levels of transmission and a full range of proportions of asymptomatic infections to account for uncertainties around these values. Most simulations resulted in small-scale outbreaks, with outbreaks exceeding 100 cases when the basic reproduction number was above 0.8, consistent with previous estimates [34].
Contact tracing and self-isolation reduced outbreak size, but their potential effectiveness in preventing cases decreased as the proportion of asymptomatic infections increased, since they rely on symptomatic cases.
Additionally, larger values of R0 would increase outbreak size and the number of contacts, which may slow down contact tracing efforts and reduce the effectiveness of current interventions. Previous modelling studies similarly concluded that contact tracing and self-isolation alone were highly unlikely to prevent large outbreaks, particularly when asymptomatic infections were present [28,35]. Our findings support and extend these conclusions to a different pathogen.
(Continue . . . )
The late George E. P. Box (18 October 1919 – 28 March 2013) - Professor Emeritus of Statistics at the University of Wisconsin - is often credited with coining the familiar adage:
“All models are wrong, but some models are useful.”
It is impossible to account for all of the variables in a real-life situation, and so certain parameters must be set, such as capping the R[SUB]0 [/SUB]at 1.1.
Some assumptions in this model that may depart from reality are that (1) all symptomatic farm workers would be immediately tested, (2) that contact tracing would be timely and 100% effective, and (3) that all contacts would fully self-isolate.
Even if this remarkable hat-trick could be achieved in the UK, the reality is that surveillance, testing, contact tracing, and reporting from many other countries is - to put it kindly - suboptimal.
None of this is to suggest we shouldn't try, of course.
If the missing ingredient to human-adaptation of HPAI H5 is a long-chain of human infections (plausible, but not certain), then anything we can do to prevent or limit that type of uncontrolled serial passage experiment is very much worth the effort.
But whether we have the political and societal will do to so, is far less certain.
https://afludiary.blogspot.com/2025/12/plos-gph-quantifying-h5n1-outbreak.html
R[SUB]0[/SUB] (pronounced R-nought) or
Basic Reproduction Number
#19,004
In the spring of 2023, in UK Novel Flu Surveillance: Quantifying TTD, we looked at a technical briefingfrom the UKHSA which attempted to quantify (via statistical modelling) the UK's ability to detect community transmission of a novel flu virus, such as HPAI H5N1.
At that time, the UKHSA estimated it would likely take between 3 and 10 weeks before community spread would become apparent to authorities, after anywhere between a few dozen to a few thousand community infections.
This is their `best case' R[SUB]0[/SUB] 1.2 scenario
Soberingly, this was for the UK; where public health and testing capabilities are far stronger than in a lot of countries currently dealing with HPAI. Even so, detecting community spread early in an outbreak would require a good deal of luck.
With HPAI H5 now primarily a disease of poultry (and in some countries, other livestock), the ideal place to detect, and halt, human transmission would be on the farm.
The authors first surveyed bird‑exposed farm workers about how many people they met in a day, and then generated computer models of outbreaks starting from a single human infection at various R[SUB]0[/SUB] rates (0 up to 1.1 in steps of 0.1).
Note: The 2023 UKHSA Community Transmission study assumed an R[SUB]0 [/SUB]starting at 1.2, up to 2.0.
What they found was - using somewhat idealized public health tools (contact tracing of symptomatic cases’ contacts & self‑isolation of symptomatic traced contacts) - that when the transmission R[SUB]0 [/SUB]was < 1, the number of cases would remain low; often under 10 cases.
But, as the R[SUB]0 [/SUB] approaches 1.0, or if the number of asymptomatic cases go up, cluster sizes increase.
Due to its length and technical nature, I've only presented a brief overview. Below you'll find the abstract and a few excerpts, but you'll want to follow the link to read the report in its entirety.
I'll have a bit more after the break.
Quantifying H5N1 outbreak potential and control effectiveness in high-risk agricultural populations
Izel Avkan ,Suzanne Gokool,Louise E. Smith,Genevieve Clapp,Rachel Cox,Amy C. Thomas,Ellen Brooks-Pollock
PLOS x Published: December 29, 2025
https://doi.org/10.1371/journal.pgph.0005463
Abstract
Avian influenza is a global public health threat. Since 2021, the ongoing H5N1 panzootic has brought a major shift in H5Nx epidemiology, including unprecedented spread, wide host range and lack of seasonality. Infections in marine mammals, wildlife and livestock have heightened concern for human-to-human transmission and pandemic potential. Contact tracing and self-isolation are used as public health measures in the UK to manage contacts of confirmed human cases of avian influenza.
In this study, we aimed to estimate potential outbreak sizes and evaluate the effectiveness of contact tracing and self-isolation in managing community outbreaks of H5N1 following spillover from birds to people. We characterised contact patterns from an underrepresented agricultural population at high risk of avian influenza exposure through contact with birds (Avian Contact Study).
Informed by these realistic social contact data, we modelled outbreak sizes using a stochastic branching process model. Most simulations resulted in small-scale outbreaks, ranging from 0 to 10 cases.
When the basic reproduction number was 1.1, contact tracing and self-isolation reduced the average outbreak size from 41 cases (95% Confidence Interval (CI): 37–46 cases) to 7 cases (95% CI: 6–8 cases), preventing, on average, 8 out of every 10 infections.
However, controls became less effective in reducing the outbreak size when a higher proportion of cases were asymptomatic. Overall, our findings suggest that contact tracing and self-isolation can be effective at preventing zoonotic infections.
However, controls became less effective in reducing the outbreak size when a higher proportion of cases were asymptomatic. Overall, our findings suggest that contact tracing and self-isolation can be effective at preventing zoonotic infections.
Increasing awareness, encouraging self-isolation, and detecting asymptomatic cases through routine surveillance are important components of zoonotic infection containment strategies.
(SNIP)
Principal findings
Our model provides insights into the potential outbreak size of avian influenza among humans under different levels of transmission and a full range of proportions of asymptomatic infections to account for uncertainties around these values. Most simulations resulted in small-scale outbreaks, with outbreaks exceeding 100 cases when the basic reproduction number was above 0.8, consistent with previous estimates [34].
Contact tracing and self-isolation reduced outbreak size, but their potential effectiveness in preventing cases decreased as the proportion of asymptomatic infections increased, since they rely on symptomatic cases.
Additionally, larger values of R0 would increase outbreak size and the number of contacts, which may slow down contact tracing efforts and reduce the effectiveness of current interventions. Previous modelling studies similarly concluded that contact tracing and self-isolation alone were highly unlikely to prevent large outbreaks, particularly when asymptomatic infections were present [28,35]. Our findings support and extend these conclusions to a different pathogen.
(Continue . . . )
The late George E. P. Box (18 October 1919 – 28 March 2013) - Professor Emeritus of Statistics at the University of Wisconsin - is often credited with coining the familiar adage:
“All models are wrong, but some models are useful.”
It is impossible to account for all of the variables in a real-life situation, and so certain parameters must be set, such as capping the R[SUB]0 [/SUB]at 1.1.
Some assumptions in this model that may depart from reality are that (1) all symptomatic farm workers would be immediately tested, (2) that contact tracing would be timely and 100% effective, and (3) that all contacts would fully self-isolate.
Even if this remarkable hat-trick could be achieved in the UK, the reality is that surveillance, testing, contact tracing, and reporting from many other countries is - to put it kindly - suboptimal.
None of this is to suggest we shouldn't try, of course.
If the missing ingredient to human-adaptation of HPAI H5 is a long-chain of human infections (plausible, but not certain), then anything we can do to prevent or limit that type of uncontrolled serial passage experiment is very much worth the effort.
But whether we have the political and societal will do to so, is far less certain.
https://afludiary.blogspot.com/2025/12/plos-gph-quantifying-h5n1-outbreak.html