Biosafety risk

ISO 35001:2019, Biorisk management for laboratories and other related organizations (1st ed., 2019). The standard leaves the definition of likelihood levels to the organization. Unlike nuclear safety frameworks, where “very rare” is often taken as one event per million to ten million reactor-years (in France) this allows context-specific interpretation aligned with national regulations or internal risk matrices.

The WHO Laboratory Biosafety Manual requires a documented risk assessment and mitigations proportionate to the risk. It does not itself set a numerical frequency. Under the ALARP principle, a risk is tolerable if further reduction would be disproportionate in cost, and the numerical threshold is set by the organization or regulator. A threshold of one event per million laboratory-years is sometimes used as a reference.

In HAZOP and QRA studies of BSL-3 biocontainment facilities, rare events with catastrophic consequences (i.e., a pandemic) are typically assigned target frequencies between 10⁻⁴ and 10⁻⁶ per laboratory-year. The value depends on the organization, national regulations, the pathogens handled, and the date of construction.

Simplified estimate. Definitions:

N = number of laboratories: 3,000 today, doubling every 10 years
f = proportion of laboratories working on pandemic potential pathogens (PPP): f = 0.1
T = period considered: 10 years
k = average growth factor of N over T: k = 1.5 (linear approximation; the exponential average is about 1.44)
p = probability of a pandemic-causing event per laboratory per year (events per laboratory-year)
P = probability that at least one laboratory causes a pandemic during T

Assuming independent laboratories and a constant p:

P = 1 − (1 − p)^(f·N·T·k)

The exposure is f·N·T·k = 0.1 × 3,000 × 10 × 1.5 = 4,500 laboratory-years. For p = 10⁻⁴, 10⁻⁵ and 10⁻⁶ per laboratory-year, P = 1 − (1 − p)^4500 = 36.2%, 4.4% and 0.45% respectively.

Interpretation: if every PPP laboratory operated exactly at the tolerated frequency, the probability of at least one laboratory-leak pandemic over the next ten years would lie between about 0.45% and 36%. These figures do not measure the actual risk. They show the cumulative risk that would result if every pandemic-potential laboratory operated exactly at its tolerated frequency. The upper value (36%) assumes the least stringent criterion (10⁻⁴ per laboratory-year), and the lower value (0.45%) assumes the most stringent (10⁻⁶). Real laboratories may perform better or worse than their design targets.

References: Assessing and Managing the Risks of Potential Pandemic Pathogen Research Daniel J Rozell 2015 Jul 21;6(4):e01075-15. doi: 10.1128/mBio.01075-15

Biocontainment laboratory risk assessment: perspectives and considerations Amy Patterson, Kelly Fennington, Ryan Bayha, Diane Wax, Rona Hirschberg, Nancy Boyd, Michael Kurilla, July 2014,, doi.org/10.1111/2049-632X.12162

Mapping biosafety level 3 (BSL-3) and BSL-4 laboratories for public health threats reduction Atalay Goshu Muluneh, Aye Moa, Samsung Lim & Chandini Raina MacIntyre, 20 may 2025

CWA 15793, WHO Laboratory Biosafety Manual

Expertise ANSR France

Under of better global surveillance, we consider 90% of labs working on pandemic potential pathogens PPP with p1=10-6, 9% of labs with p2=10-5 and 1% of labs with p3=10-4

P = 1 − exp[ −f·N·k·T·(0.90·p₁ + 0.09·p₂ + 0.01·p₃) ] With f·N = 300, k = 1.5 and T = 10, the exponent is 4,500 × 2.8×10⁻⁶ = 0.0126, so P = 1.25%

P=1,25% probability of at least one lab leak pandemic in the next ten years under a global governance that is still to be put in place.

Macro approach using Poisson distribution:

If we don't do global laboratories' surveillance, knowing 1977 H1N1 & 2019 COVID pandemics have been lab caused, we consider one lab leak pandemic every 50 years. The number of labs doubling every ten years, and using poisson distribution, the probability to have at least one lab leak pandemic in the next ten years is therefore P=1−p(0;0,3) = 25%

P (at least one pandemic) = 1-e-λ ; λ=1/50 *T*1,5= 0,3 forT=10 years

Route of escapes

Routes of escape of pathogenes from a BSL-3/BSL-4 lab can be release of inadequately decontaminated exhaust air, liquid waste, or solid waste, leaks caused by defects in pneumatic doors, control system, electrical stability, essential systems & UPS, fire fighting system, autoclave, desinfectants etc.. ; contamination of shoes or clothes of lab staffers, contamination of lab workers (LAI: Laboratory Aquired Infection), escape of infected animals, inproper inactivation of infected animal carcass.

Picture: WIV P4 Chinese Academy of Sciences - Wuhan Institute of Virology, BSL-4, http://www.whiov.ac.cn/

Air Handling Units

Escape by release of inadequately decontaminated lab exhaust air is what happened in Lanzhou, China in 2019 with escape of Brucella from a vaccine facility, and in 1979 in Sverdlovsk, USSR, with escape of Bacillus anthracis from a BSL-3 facility.

Liquid effluents

Escape by release of inadequately decontaminated laboratory liquid waste is what happened in Pirbright, UK in 2007, with escape of foot-and-mouth-disease virus from a vaccine facility.

Escape from a freezer by lab accident or live vaccine experiment

The 1977–1978 H1N1 influenza outbreak was unusual because the virus was remarkably similar to H1N1 strains circulating in the 1950s. It disproportionately affected younger people, particularly those born after the older H1N1 strains had disappeared from circulation.

The central hypothesis is that the virus did not simply evolve naturally from the influenza strains circulating in 1977. Instead, an older H1N1 virus may have been reintroduced, potentially through an accidental release from a laboratory freezer.

What probably caused it?

Geographic emergence: The virus was identified in northeastern China in May 1977, before outbreaks were reported in the Soviet Union later that year.

Genetic evidence: The 1977 virus was almost identical to H1N1 strains from around 1950, despite roughly 27 years of normal viral evolution. This strongly suggests that an old strain had been preserved and reintroduced.

Likely mechanism: An accidental release from a laboratory, or an escape during experiments involving a live influenza vaccine, are the leading explanations. Some viral characteristics were consistent with laboratory manipulation, but the exact route remains unproven.

Illustration by AI

Lab aquired Infection LAI

Escape of SARS virus by infection of a lab worker (LAI) is what happened in 2003 and 2004 twice in Beijing IVCDC BSL-3 and once in Singapore NUS BSL-3, and once in Taiwan military BLS-4.

Picture: entrance of the Beijing IVCDC in 2015. Since, the CDC has moved to the north of Beijing

Page updated

Google Sites

Report abuse