Summary

Extreme pandemics could result in mass shortages of vital workers, severely limiting the ability of countries to provide food, water, and other basic needs to the public. We think someone should be working on developing contingency plans to prevent this.

We’ve been looking into this risk area and potential interventions at the Alliance to Feed the Earth in Disasters (ALLFED), and in 2024 we collaborated with Oxford Biosecurity Group for an initial research project on the topic — you can read our preprint here: Developing contingency plans to protect vital sectors in extreme pandemics. We’ve since started a follow-up project with the Australian National University, which should be completed later this year. We think further work is likely important, neglected, and possibly tractable^[1]. However, our capacity is limited. ALLFED is facing a serious funding gap and will have to cut half our budget in the coming months if our emergency appeal is unsuccessful. What’s most important to us in this area is that we have strong response plans in place for keeping vital services functioning in extreme pandemics, even if we do not have the funding to do the work ourselves. It would be equally great to see others launching projects to build resilience in this area.

In the rest of this post, I’ll give a condensed overview of what we’ve done so far, the research we have planned, and some of the policy opportunities we think are particularly promising. We are keen to hear your thoughts, so please feel free to comment below or reach out directly at james.m{at}allfed.info.

Our research so far

The problem

Future pandemics caused by a virus that is both extremely transmissible and extremely virulent could cause mass labor shortages due to illness, death, quarantine, caring for those who are ill, or fear of infection. The 2014–2016 Ebola outbreak in West Africa provided a glimpse of what this pathway might look like when rice yields dropped by 12% nationally in Liberia, and by up to 25% in the worst-affected districts, largely driven by workers being afraid to risk infection. This was a situation with a very high case-fatality rate, but low prevalence in comparison to outbreaks of COVID-19, influenza, or measles. As rapid developments in technology and artificial intelligence (AI) make pathogen engineering more and more accessible, the threat of food, water, energy, communications, and healthcare systems collapsing from worker shortages rises. This may be particularly relevant for those with short AI timelines to consider, as near-term artificial general intelligence (AGI) could amplify the risk of engineered pandemics and leave little time for building preparations^[2].

While some biosecurity projects in the EA community could improve the resilience of vital infrastructure to extreme pandemics given enough time, what if we are not prepared when a pandemic strikes? Masks and filtering facepiece respirators were heavily relied upon during COVID-19, but a pandemic virus as transmissible as measles would likely require more protective PPE (i.e., pandemic-proof PPE) which we do not have enough of. Germicidal ultraviolet (UV) light (both upper-room and Far-UVC systems) could reduce transmission between workers, but is currently close to nonexistent in the workplace. Developing response plans for extreme scenarios could present a low-hanging fruit for increasing pandemic resilience at relatively low cost.

In order to explore this idea in more detail, we set out to answer three research questions on vital workers, defined as the minimum workers who are necessary for the basic functioning of society and who likely cannot complete their work from home. This is a subset of essential workers as used during COVID-19, taking only those who are absolutely necessary to weather the initial stage of the pandemic before ramping other services up.

1. Is there enough pandemic-proof PPE (elastomeric respirators and powered air purifying respirators (PAPRs)) to cover all vital workers in the US during an extreme pandemic?
2. What are the main vulnerabilities in vital sectors with respect to labor shortages in extreme pandemics?
3. What response planning recommendations could help to protect vital workers in extreme pandemics?

Our methods are covered in detail in the preprint, but in short we created a Monte Carlo model of estimates for PPE availability and requirements, reviewed the academic and grey literature on vital sectors, and interviewed 8 key informants across the food, water, energy, telecommunications, and PPE sectors.

1. Vital industries are expected to face severe shortages of adequate PPE in an extreme pandemic

Firstly, we found that vital worker requirements for pandemic-proof PPE (33 million units; 90% CI: 16–65 million) far outweigh stockpiles of elastomerics and PAPRs in the US, taking into account national stockpiles, military stockpiles, and local hospital stockpiles (1.5 million units; 90% CI: 0.86–2.7 million). Adding all manufacturer, retailer, and private workplace pandemic-proof respirators to stockpiles (“all rapidly-accessible PPE”) would still fall short of requirements at at 7.1 million (90% CI: 4.2–12 million) units, suggesting that even a perfect response to round up almost every possible source of pandemic-proof PPE would not be enough. We suspect this to be the case for most other countries given that most respirators are produced in China and the US. Given these shortages, there seems to be a significant need for further interventions to protect vital workers.

2. Specific vulnerabilities in vital industries provide targets for future interventions

After confirming that vital workers are likely at high risk of infection due to PPE shortages, we collected information on vulnerabilities in each sector included in our analysis. Developing a deep understanding of vulnerabilities within specific sub-sectors and roles can identify likely chokepoints for infrastructure failure, helping focus efforts for future interventions.

Across vital infrastructure, operational roles such as operators, technicians, and drivers were consistently identified as a risk to service continuity. Organizations in the food, water, energy, and telecommunications sectors all cited these roles as specific vulnerabilities, with some differences in risk-mechanisms. For example, food processing is high-risk due to workplace transmission rates. High worker-density, difficulties in maintaining social distancing, and shouting over loud conditions all facilitate transmission. For operators and technicians in powerplant control rooms, close working quarters are still a factor but this is combined with increased difficulty in replacing specialized workers. A single infection requires the whole shift team to quarantine, which could reportedly cause energy organizations to run out of workers in a matter of weeks without adequate planning. However, workplace transmission is only one piece of the puzzle.

Protecting vital workers includes keeping them safe when not working, whether they stay at work or off-site. Farm workers were found to face disproportionately higher rates of infections from crowded on-site housing, sharing modes of travel to reduce costs, and lack of labor protections such as paid sick leave disincentivizing isolation when infected. This is particularly applicable to migrant workers, undocumented workers, and racial and ethnic minorities. Working in a low-contact environment cannot keep vital workers safe if they get infected outside of work. While stay-at-home orders would help reduce transmission, they alone would not be sufficient and improved response planning will have to account for this. 

Additionally, interdependencies in vital industries mean that reduced access to one sector can have dangerous ripple effects. Beyond more obvious examples such as the dependency of all vital sectors on a functioning electricity grid, some studies suggest a US national moratorium on utility shutoffs could have reportedly prevented COVID-19 infections by as much as 9% and deaths by as much as 15%. At a larger scale, disconnections caused by utility failure in an extreme pandemic could trigger a cascading spiral of infrastructure collapse: losing access to water or electricity leads to increased infections, which leads to shocks to the vital labor supply, which leads to higher risks of more utility failures and subsequent labor shocks.

3. Contingency-planning interventions for extreme pandemics

Given that vital industries remain vulnerable to extreme scenarios, response plans can serve as an additional layer of protection if preparedness interventions fail to prevent an outbreak from spreading. Based on our findings, we put together a shortlist of recommendations for developing response strategies.

1. Mobilize & allocate PPE. In the event of severe shortages, figuring out how to efficiently locate and distribute pandemic-proof PPE from non-traditional stockpile sources (e.g., private sector businesses, manufacturer and retail stock) may help plug the gap. Moreover, it is important that risk assessments stratify transmission risk across different roles to ensure that, at a minimum, vital workers in high-contact roles receive pandemic-proof PPE.
2. Adapt workplaces. Engineering controls that can quickly be ramped up to reduce transmission should be implemented — in-room filtration, upper-room germicidal UV, ventilation, and glycol vapors seem like promising candidates in particular. General workplace adaptations such as maximizing teleworking, implementing reduced shift schedules, and cross-training workers on mission-critical tasks should be combined with this approach. Though this work is focused on wildfire pandemics, cross-training and rapid training of new staff may be particularly important for a stealth pandemic that has already spread widely.
3. Provide high-quality on-site worker housing. Learning from the small-scale trials of workers isolating on-site during COVID-19, contingency planning should prepare for rolling this out at scale. These plans will need to encourage participation through generous compensation, put protections in place to prevent exploitation, and set stringent quality standards to avoid infections spreading in employee housing.
4. Address sociodemographic vulnerabilities. The essential workforce is largely made up of lower-income and minority groups, who are at higher risk of infection. Response plans should account for this heightened risk by actions such as banning utility disconnections, providing safe transportation options, and providing safe on-site agricultural worker housing that does not pose an infection risk from overcrowding and poor sanitation.
5. Develop post-collapse plans. Response plans should include guidance on meeting basic needs in the event of local or widespread infrastructure collapse. As collapse is likely to occur in only the most severe scenarios, even localized loss of basic services poses a significant threat as neighboring regions or countries would likely be too preoccupied in protecting their own populations to provide any humanitarian aid or assistance.

Current and planned follow-on work

Research opportunities

We would be excited to see more work done to protect vital sectors in extreme pandemics, whether this work develops our current set of interventions or takes a new approach. The projects listed in this section are primarily focused on response planning — what we can do post-catastrophe. We think there are also promising pre-catastrophe scale-up interventions worth developing for vital sectors (e.g., stockpiling of elastomeric respirators), but we have chosen to draw attention to contingency plans as we think they are underexplored and lower cost pre-catastrophe. Given our current funding situation, we have limited capacity to take on these projects. However, we have partnered with Australia National University (ANU) to reduce costs by running one of these projects as a final year undergraduate team project^[3]. 

Additionally, we have provided a list of other projects that we think have the potential to be impactful pieces of research. We plan on completing this list if we can get funding to resource it, but this appears highly uncertain so we also encourage others to consider this work. If you feel confident that you could make a solid attempt at any of the projects, please feel free to reach out and I would be happy to provide any information that could be useful.

1. [Ongoing at ALLFED + ANU] Rapid scale-up of in-room filtration.

In the absence of robust, widespread pandemic-proof indoor air quality systems (e.g., Far-UVC), emergency ramp-up of existing systems has the potential to decrease transmission in order to protect vital sectors. During the COVID-19 pandemic, makeshift solutions such as Corsi-Rosenthal boxes highlighted that emergency filtration systems can provide equivalent or greater improvements in air quality to conventional systems at a tenth of the cost. This project will estimate how different systems — either conventional or makeshift solutions — could be applied to vital infrastructure at scale and at equivalent clean airflow rates order(s) of magnitude higher than current air quality standards. Tied to this, project members will quantify the number of vital workers in different sectors across different country contexts globally in order to understand filtration requirements for different settings. For example, a country with a high proportion of its workforce in (smallholder) agriculture such as Ethiopia is likely to require a different approach than a country with a highly industrialized agriculture system such as Canada.

2. [Open] Rapid scale-up of other indoor air quality interventions.

In addition to in-room filtration, there are other technologies that could potentially be used to reduce transmission and add to the toolkit of extreme pandemic contingency plans. Each of these projects should consider the likelihood that the technology reduces transmission; the contexts in which they could be used (e.g., which countries and which industries); the scale-up potential over time of existing materials, repurposing of construction facilities, and new construction facilities; and bottlenecks.

• 2a. [Open] Germicidal UV. Various germicidal UV technologies — Far-UVC, louvered upper-room UV, open upper-room UV, eggcrate upper-room UV, UV LEDs — could possibly be scaled up in a crisis. Upper-room systems appear promising for this due to 254nm bulbs being used in current-day industries (e..g, hospitals, water treatment), but it is worth exploring all options in detail.
• 2b. [Open] Ventilation. High ventilation rates may be able to reduce infections even in the absence of filtration. Developing plans to carefully modify buildings to massively increase ventilation rates while providing adequate thermal comfort may be particularly useful for reducing transmission in low-resource settings.
• 2c. [Open] Glycol vapors. Glycols were brought to our attention recently by another org as they can potentially reduce transmission by inactivating viruses when aerosolised. Glycols are widely used across many industries, such as in fog machine fluid, e-cigarettes, food manufacturing, and antifreeze, meaning they could provide a significant cache of materials to quickly scale up air disinfection.

3. [Open] Rapid scale-up of on-site worker housing.

Some small-scale examples of workers being kept on-site during crises were rolled out successfully during COVID-19 in industries such as the energy sector and polypropylene manufacturing. However, protests and controversy regarding the lockdown of a 200,000-employee iPhone factory in China highlight that more work is needed to improve the logistical and ethical standards of ramping this up at massive scales. To address this, this project would quantify the scale-up potential and speed of large-scale repurposing and construction of emergency accommodation for vital sectors, analyse the different requirements across various sectors and organisation sizes, and investigate how to set-up facilities to minimize infections in an ethical manner (e.g., incentivising employees by providing high-quality housing, high pay, attractive benefits, and agreeable terms rather than forcing people to work and live on-site).

4. [Open] Rapid increase of PPE for vital workers.

In the event of (pandemic-proof) PPE shortages during an extreme pandemic, creative interventions to increase availability for vital workers could be highly impactful. Our analysis suggests there are significant volumes of pandemic-proof PPE outside of traditional strategic stockpiles that could be rapidly mobilised, such as elastomerics in non-vital industrial workplaces. Additionally, identifying and solving bottlenecks in PPE manufacturing could enable faster ramp-up of production in a crisis. Figuring out how to increase PPE availability through mechanisms such as these seems like a worthwhile time investment.

Policy opportunities

In order to facilitate this research being translated into effective pandemic contingency planning, we plan to engage governments, multilateral organizations, and industry groups. We have been exploring various policy frameworks in which this research could be implemented — the WHO Pandemic Treaty, the EU Directive on the Resilience of Critical Entities, the US National Security Memorandum on Critical Infrastructure Security and Resilience, the Africa CDC-WHO Joint Emergency Action Plan, Information Sharing and Analysis Centers (ISACs), the Water/Wastewater Agency Response Network (WARN) — but we are currently focusing on the EU Directive. 

The EU has recently strengthened its approach to protecting critical infrastructure through the Directive on the Resilience of Critical Entities, providing an opportunity for improved extreme pandemic preparedness. Key obligations include conducting risk assessments every four years and developing resilience plans to address identified risks. Member states will need to adopt a national strategy to meet these obligations, and entities deemed critical will need to carry out their own risk assessments and take measures to enhance their resilience. This provides multiple avenues to provide input on risk assessment and response planning where tail-risk scenarios would otherwise be unlikely to be considered. While it is currently unclear whether there is scope for extreme pandemics in these considerations, we will be seeking to explore opportunities to feed in relevant work on this topic.

Acknowledgements

I'd like to give a huge thanks to the Oxford Biosecurity Group team that we collaborated with for the research project kicking off this work — Lin Bowker-Lonnecker, Okan Ozbek, Biak Tial, Niklas Keßeler, Natalie Kiilu, Nadia Montazeri, and Shreeman Misurya.

Feedback

We think this area of work is promising and worth exploring further. We would be keen to hear your perspective on whether you agree or disagree, and why. Please feel free to leave your comments below or reach out directly at james.m{at}allfed.info

1. ^{^}

   While we may look into publishing a more rigorous cost-effectiveness analysis in the future, some initial simple models (internal estimates and Squiggle-AI-informed estimates) suggest contingency planning may be competitive with or more cost-effective than alternative interventions such as stockpiling.

2. ^{^}

   Currently on Metaculus, the 25th percentile forecast for artificial general intelligence (AGI) is 13 August 2028 (and 6 August 2025 for weak AGI). The advent of AGI could significantly lower the barriers to a non-state actor, state actor, or misaligned AI creating and releasing an engineered pandemic. If this occurs, three years is a very short amount of time to convince governments to spend billions on pandemic-proof PPE stockpiles, reach widespread implementation of Far-UVC, instate globally-coordinated pathogen-agnostic early detection systems, significantly accelerate our ability to produce novel vaccines and medical countermeasures, or otherwise dedicate massive resources to prepare for an extreme pandemic in advance.

3. ^{^}

   We have had good success running undergraduate team research projects in the past for low costs. We have also had success with post-graduate projects. One advantage of the undergraduate format is you can have a whole team working simultaneously on the project.