What causes biennial epidemics?

I read the Wikipedia article on acute flaccid myelitis which contains a graph showing confirmed US cases peaking biennially in 2014, 2016 and 2018.

That got me wondering because I found that pattern to seem unusual compared to an annual one tied to seasons or a constant or increasing one. So I did a quick Google search on "biennial patterns in rates of infection" and turned up this paper: Regular biennial cycles in epidemics caused by parametric resonance which discusses, mathematically, the biennial patterns sometimes found in the incidence of measles.

What I'd like to know, in layman's terms, is the causes of biennial patterns such as these in terms of the way these infections spread among victims and develop within them.

Acute Flaccid Myelitis: The Replacement Polio

Taking health officials off guard, a spike in polio-like disease was first described in a Morbidity and Mortality Weekly Report from the CDC in October of 2014. The report detailed clusters of children in California and Colorado suffering from acute flaccid paralysis (AFP), a neurological illness that results in the sudden onset of paralysis (Figure 1). AFP can have a wide range of causes including environmental toxins, genetic conditions, and infection by viruses. Using magnetic resonance imaging (MRI) to peer inside the body, investigators discovered that all of the patients exhibited inflammation of the spinal cord (referred to as myelitis) in a specific neuron-dense region known as grey matter. This specific form of myelitis is usually the calling card of poliovirus infection. Surprisingly, though, investigators found no evidence of poliovirus. There have been a few reports of polio-like myelitis also occurring during West Nile virus and adenovirus infections, but investigators could not find evidence of infection by these viruses either. With no suspect to blame for the sudden spike in polio-like paralysis, the term acute flaccid myelitis (AFM) was coined to encompass all AFP conditions that display polio-like myelitis.

Figure 1. AFP. Acute flaccid paralysis (AFP) is a broad classification of disorders defined by the rapid onset of limb weakness. AFP can be caused by complications from neurotoxic snake venom, tick bites, HIV infection, and genetic disorders such as Guillain-Barre syndrome. Acute flaccid myelitis (AFM) is a specific kind of AFP that is linked to infection by a number of different viruses such as poliovirus, EV-D68, West Nile virus, and adenoviruses. The paralysis in AFM is thought to result from inflammation in the grey matter (a thick bundle of motor nerves) of the spinal cord caused by viral infection. The classic example of AFM is poliomyelitis (“Polio”) that is caused by the poliovirus.

Emerging and Re-emerging Diseases

The distribution of a particular disease is dynamic. Therefore, changes in the environment, the pathogen, or the host population can dramatically impact the spread of a disease. According to the World Health Organization (WHO) an emerging disease (Figure (PageIndex<3>)) is one that has appeared in a population for the first time, or that may have existed previously but is rapidly increasing in incidence or geographic range. This definition also includes re-emerging diseases that were previously under control. Approximately 75 percent of recently emerging infectious diseases affecting humans are zoonotic diseases, zoonoses , diseases that primarily infect animals and are transmitted to humans some are of viral origin and some are of bacterial origin. Brucellosis is an example of a prokaryotic zoonosis that is re-emerging in some regions, and necrotizing fasciitis (commonly known as flesh-eating bacteria) has been increasing in virulence for the last 80 years for unknown reasons.

Figure (PageIndex<3>): The map shows regions where bacterial diseases are emerging or reemerging. (credit: modification of work by NIH)

Some of the present emerging diseases are not actually new, but are diseases that were catastrophic in the past (Figure (PageIndex<4>)). They devastated populations and became dormant for a while, just to come back, sometimes more virulent than before, as was the case with bubonic plague. Other diseases, like tuberculosis, were never eradicated but were under control in some regions of the world until coming back, mostly in urban centers with high concentrations of immunocompromised people. The WHO has identified certain diseases whose worldwide re-emergence should be monitored. Among these are two viral diseases (dengue fever and yellow fever), and three bacterial diseases (diphtheria, cholera, and bubonic plague). The war against infectious diseases has no foreseeable end.

Figure (PageIndex<4>): Lyme disease often, but not always, results in (a) a characteristic bullseye rash. The disease is caused by a (b) Gram-negative spirochete bacterium of the genus Borrelia. The bacteria (c) infect ticks, which in turns infect mice. Deer are the preferred secondary host, but the ticks also may feed on humans. Untreated, the disease causes chronic disorders in the nervous system, eyes, joints, and heart. The disease is named after Lyme, Connecticut, where an outbreak occurred in 1995 and has subsequently spread. The disease is not new, however. Genetic evidence suggests that Ötzi the Iceman, a 5,300-year-old mummy found in the Alps, was infected with Borrelia. (credit a: James Gathany, CDC credit b: CDC scale-bar data from Matt Russell)


Our model unites predictions for the climate drivers of RSV across both temperate and tropical locations, finding that the effect of low humidity trades off against rainfall drivers, depending on location. The striking similarity in the role of specific humidity for both influenza and RSV suggests that parallel mechanisms are at play for the two viruses (Fig. 4) however, the lack of an animal model for RSV means experimental work cannot currently be used to support our findings. This correspondence appears to hold even though the population dynamics of influenza and RSV are palpably different. Influenza disappears at high latitudes in the summer whereas RSV is much more endemic. Disentangling the roots of these differences is an important area for future work. Precipitation has also been hypothesized to be important for influenza transmission, although this has not been tested in laboratory settings, likely due to the relative complexity of experimental set-up 35 . It is unknown whether physical processes such as aerosol deposition, or behavioral processes such as grouping indoors, explain the mechanism of precipitation drivers of transmission for both RSV and influenza.

Comparison with influenza results. Removing precipitation from our regression model and including a quadratic humidity term reveals a very similar response (a) to earlier work on influenza (b adapted from Tamerius et al. 9 , Fig 3a) suggesting potential similar mechanisms underlie the climate effect on the two diseases. Predicted minimum transmission for RSV occurs at 11.16 g/kg, with influenza found to be similarly 11–12 g/kg 9 . Tamerius et al. suggest precipitation may drive the right hand side of the humidity-influenza curve, as we find for RSV.

There are important caveats to our results. We made the decision to remove dummy variables in future projections so that annual variation in forced climate could flexibly determine transmission. However, this means our simulations do not account for non-climatic factors that may also structure transmission. We crudely tested this by using a common seasonal fixed effect which gave qualitatively similar results, reinforcing the robustness of our assumption. Developing a richer understanding of the mechanisms connecting climate to transmission would help validate our results further. Our model also does not take into account genetic sub-type of RSV 30 due to data limitations in the US and Mexico. We also do not consider age structure or potential secondary infections: secondary RSV infections are less severe 27 . Nonetheless, we are able to capture the limit cycle structure of RSV dynamics well and elucidate multiple streams of evidence for a fundamental climate effect.

The public health implications of changes to future RSV epidemic dynamics will depend on location-specific climatic changes. Precipitation-driven increases to epidemic intensity, caused by climate change or other large-scale climatic events such as the El-Nino-Southern Oscillation, will require increased surge capacity in locations that typically observe more uniformly distributed cases 31 . Humidity-driven reductions to the seasonal change in transmission in other locations will result in persistent epidemics, meaning cases will occur outside of the typical “RSV season” requiring changes to the temporal allocation of resources. Changing dynamics can also alter the age of infection 36,37 . This consideration is important given that early RSV infections have been implicated in the later life development of asthma 17 .

Understanding the impact of climate change on the transmission dynamics of infectious diseases is critical for predicting and preparing for future outbreaks and the optimal deployment of future vaccines. Our results highlight the nonlinear impact of climate on RSV dynamics, across a wide range of climatic conditions. Shifts in the timing and magnitude of RSV outbreaks will be location-specific, depending on the realized climatic change. In regions where transmission is dominated by humidity, we can be relatively confident in our projections. However, in regions where precipitation dominates transmission, the future dynamics of RSV are more uncertain. Efforts to better resolve precipitation projections across climate models will improve the precision of these results. These findings also have potential implications for other airborne pathogens and future research efforts should be directed towards understanding the various processes through which climate affects airborne transmission. More broadly, our preliminary comparison between RSV and influenza reveals the potential for comparative studies to elucidate the mechanisms of climatic forcing for major respiratory infections.

Эпидемии - динамика инфекционных заболеваний

Not so long ago, it was almost guaranteed that you would die of an infectious disease. In fact, had you been born just 150 years ago, your chances of dying of an infectious disease before you've reached the tender age of 5 would have been extremely high.

Since then, science has come a long way in understanding infectious diseases - what they are, how they spread, and how they can be prevented. But diseases like HIV/AIDS, Malaria, Tuberculosis, or the flu are still major killers worldwide, and novel emerging diseases are a constant threat to public health. In addition, the bugs are evolving. Antibiotics, our most potent weapon against bacterial infections, are losing their power because the bacteria are becoming resistant. In this course, we'll explore the major themes of infectious diseases dynamics. After we’ve covered the basics, we'll be looking at the dynamics of the flu, and why we're worried about flu pandemics. We'll be looking at the dynamics of childhood diseases such as measles and whooping cough, which were once considered almost eradicated, but are now making a comeback. We'll explore Malaria, and use it as a case study of the evolution of drug resistance. We'll even be looking at social networks - how diseases can spread from you to your friends to your friends' friends, and so on. And of course we’ll be talking about vaccination too. We’ll also be talking about how mobile phones, social media and crowdsourcing are revolutionizing disease surveillance, giving rise to a new field of digital epidemiology. And yes, we will be talking about Zombies - not human zombies, but zombie ants whose brains are hijacked by an infectious fungus. We're looking forward to having you join us for an exciting course!


Description of spatiotemporal patterns of RSV

Hospitalizations for RSV were strongly seasonal, with annual epidemics occurring during the winter months in most states ( Fig. 1A , S1 Fig.). Some states (e.g. Colorado, Iowa, California in the 1990s) exhibited biennial patterns of alternating �rly-big” epidemics in/around January of even-numbered years and “late-small” epidemics in/around February of odd-numbered years. The peak in RSV hospitalizations was notably earlier in Florida (occurring in November/December) compared to the other states, and hospitalizations occurred throughout the year ( Fig. 1A ). The vast majority (㺗%) of RSV-coded hospitalizations occurred among children υ years of age, and �% occurred among children ρ year of age. The age distribution of cases varied slightly by state ( Fig. 1B ).

(A) Time series of weekly RSV hospitalizations in select states. Raw hospitalization data is shown in blue, while the rescaled data accounting for the addition of an RSV-specific ICD-9 code in September 1996 is shown in green. (B) Age distribution of RSV hospitalizations across ten states. (C) Center of gravity of RSV activity in states with at least ten consecutive years of laboratory reports. (D) Strength of biennial cycle in RSV activity, as indicated by the ratio of the biennial to annual Fourier amplitude for laboratory report data.

Laboratory reports of RSV-positive specimens exhibited a distinct spatial pattern, with mean timing of RSV activity (as indicated by center of gravity, a measure of mean epidemic week (S1 Text) [36]) occurring earliest in Florida and latest in Montana ( Fig. 1C ). Again, some states exhibited a biennial pattern of RSV epidemics these states were highly concentrated in the upper Midwest and West regions ( Fig. 1D ). The laboratory and hospitalization data were highly correlated for those states with both types of data available (rϠ.71, pπ.0001, S1 Table).

Linking environmental drivers and timing of RSV activity

We explored trends between a variety of climatic and non-climatic variables and timing of RSV activity across US states, as measured by both center of gravity and phase difference with Florida ( Table 1 and S2 Table, S2 Fig.). Negative associations were found with annual mean vapor pressure, temperature, precipitation, and potential evapotranspiration (PET), and were generally stronger when considering the mean value for the fall months (September-November) for each climatic factor. Population size and latitude were also associated with RSV timing ( Table 1 ). Fall vapor pressure had the highest explanatory power (R 2  =��%), and was also the only significant factor in an exploratory multivariate analysis (pπ.0001) (S3 Table). Note that while these analyses may be indicative of statistical trends, they do not account for the intrinsic nonlinear epidemic dynamics of RSV.

Table 1

Model explaining phase timingModel explaining gravity timing
Explanatory variableParameter estimate (SE)R 2 Parameter estimate (SE)R 2 Parameter estimate (SE)R 2 Parameter estimate (SE)R 2
Climate variables:Annual averageFall average a Annual averageFall average
Vapor pressure 𢄠.065*** (0.006) 67% 𢄠.064*** (0.006) 72% 𢄠.58*** (0.05) 72% 𢄠.57*** (0.04) 76%
Temperature b 𢄠.049*** (0.005) 4663% 𢄠.05*** (0.005) 4564% 𢄠.45*** (0.04) 5372% 𢄠.46*** (0.04) 5573%
Precipitation 𢄠.006*** (0.001) 35% 𢄠.005*** (0.001) 34% 𢄠.05*** (0.009) 43% 𢄠. 05*** (0.009) 39%
Potential evapotranspiration 𢄠.12 * (0.06) 6% 𢄠.14 * (0.06) 8% 𢄡.15 * (0.05) 8% 𢄡.39** (0.52) 11%
Wet days𢄠.013 (0.015)0%𢄠.004 (0.015)0%𢄠. 13 (0.13)0%𢄠.03 (0. 13)0%
Cloud cover0.006 (0.006)0%0.006 (0.005)0%0.05 (0.05)0%0.07 (0.04)3%
Diurnal temperature range0.04 (0.02)5%0.04 (0.02)5%0.34 (0.19)4%0.29 (0.18)3%
Non-climate variables:
Pop size 𢄡.42 E-8 * (6.4 E-9) 7% 𢄡 E-7 (6 E-8)4%
Pop density𢄠.00004 (0.00003)2% 𢄠.00036 (0.00024)2%
Latitude 0.03*** (0.005) 44% 0.30*** (0.04) 53%
Longitude𢄠.003 (0.002)1% 𢄠.030 (0.018)1%
Sampling (# RSV tests)𢄦 E-7 (4E-7)0% 𢄥 E-6 (4E-6)1%

Dynamic modeling analyses

Mathematical modeling of the transmission dynamics of RSV allows us to explore the mechanistic relationship between the potentially important environmental variables and seasonal variation in the transmission rate, via which the environmental variables would likely act to affect the incidence of RSV [30]. We developed an age-stratified SIRS (Susceptible-Infectious-Recovered-Susceptible) model for the transmission dynamics of RSV, accounting for repeat infections and using natural history parameters derived from RSV cohort studies ( Table 2 ). The model was able to reproduce the age distribution (χ 2 π.17, pπ.005) and seasonal pattern of RSV hospitalizations in ten states (correlation between observed and predicted annual center of gravity: r =𠂠.87, pπ.005) ( Fig. 2 , S1 Fig.). Notably, the model was able to reproduce the biennial pattern of epidemics evident in some states even though we assume that the transmission rate of RSV follows the same seasonal pattern every year. Furthermore, the model was able to replicate the transition from biennial epidemics during the 1990s to annual epidemics during the 2000s that occurred in California, possibly due to changes in the birth rate.

(A) Compartmental diagram illustrating the structure of the model. White boxes represent infection states in the model, while grey boxes represent diseased/observed states (severe lower respiratory disease, D, and observed cases, H). (B) Model fit to weekly RSV hospitalization data for California and Florida. The ICD9-CM coded hospitalization data is shown in blue, the rescaled data is shown in green, and the fitted models are shown in red. (C) Age distribution of RSV hospitalizations in California and Florida for hospitalization data and fitted models.

Table 2

Parameter descriptionSymbolParameter valueSource
Duration of maternal immunity1/ω16 weeks [65]
Duration of infectiousness
𠀿irst infection1/γ 110 days [71], [72]
 Second infection1/γ 27 days
 Subsequent infection1/γ 35 days
Relative risk of infection following
𠀿irst infection σ 10.76 [37], [66]–[68]
 Second infection σ 20.6
 Third infection σ 30.4
Proportion of infections leading to lower respiratory tract infection
𠀿irst infection, φ months old dp ,00.5 [37], [66], [69]
𠀶11 months old dp ,0.50.3
𠀱2 years old dp ,10.2
 𢙒 years old dp ,20.1
 Second infection ds ,a =𠂠.75*dp,a
Relative infectiousness
 Second infections ρ 10.75 [37], [66], [69]
 Subsequent infections ρ 20.51Estimated

From the best-fitting model to the aggregate data from the nine states with complete age-stratified hospitalization time series from 1989�, we estimated the relative infectiousness of third and subsequent infections compared to first two infections to be 0.51 ( Table 2 ). The mean value of R 0 was estimated to be 8.9, but we observed state-specific variation in R 0 (with estimated values between 8.9 and 9.2), which was significantly correlated with population density (r =𠂠.77, pπ.01) (S3 Fig.). The estimated hospitalized fraction (h) also varied among states (from 3.2% in California to 6.9% in Colorado), but was not significantly correlated with population size or density, nor were estimates of R 0 and h significantly correlated with one another (S4 Table). Our estimates of the hospitalized fraction are similar albeit slightly lower than the 7𠄸% of infants with lower respiratory tract infections who were hospitalized during cohort studies conducted in the US and Kenya [37], [38] this is not surprising given one US-based study noted that only 45% of RSV-positive inpatients received an RSV-associated diagnosis [39].

The amplitude and timing of sinusoidal seasonal variation in the transmission rate estimated by fitting the model to the hospitalization data were both negatively correlated with mean vapor pressure and mean precipitation (pπ.01), and positively correlated with the amplitude and timing of PET in each state (pπ.01) ( Table 3 ). The seasonal offset parameter (illustrating timing of peak transmissibility) was also significantly correlated with mean minimum temperature (pπ.01). These parameter estimates were also positively correlated (pπ.001) (S4 Table).

Table 3

Hospitalization dataLaboratory data
Climatic variable Amplitude of seasonality (b) Seasonal offset (φ) Amplitude of seasonality (b) Seasonal offset (φ)
Vapor pressure
 Mean𢄠.832 * 𢄠.942***𢄠.788***𢄠.862***
𠀺mplitude𢄠.511𢄠.600𢄠.307𢄠.437 *
Minimum temperature
 Mean𢄠.575𢄠.801 * 𢄠.760***𢄠.782***
𠀺mplitude0.2530.2010.469 * 0.404
 Mean𢄠.844 * 𢄠.774 * 𢄠.760***𢄠.733***
Potential evapotranspiration
𠀺mplitude0.810 * 0.6990.689***0.671***
 Offset0.802 * 0.930**0.611***0.787***
Wet days
Mean𢄠.537𢄠.361𢄠.487 * 𢄠.256
Cloud cover
𠀺mplitude0.3850.4790.470 * 0.588**
 Offset𢄠.824 * 𢄠.812 * 𢄠.554**𢄠.755***
Diurnal temperature range
𠀺mplitude0.0840.2840.574**0.493 *

Fitting our model to the laboratory surveillance data for RSV allowed for a more extensive analysis of the relationship between state-specific climate indicators and the amplitude and timing of seasonal variability in the transmission rate across a large number of states with different climates. Since the laboratory data did not contain the age of cases, we estimated R 0 for each state based on the observed relationship between R 0 and population density prior to fitting the model.

Again, we found significant negative correlations between the amplitude and peak timing of RSV seasonal forcing (i.e. seasonality in the transmission rate) and the mean vapor pressure, minimum temperature, and precipitation across the 38 states (pπ.0001) ( Table 3 , Fig. 3A-C ), i.e. warmer, wetter states tended to exhibit less seasonal variation and an earlier peak in the transmission rate of RSV than cooler, drier states. We also found a weaker but still significant positive correlation between the amplitude of seasonal forcing and the amplitude of variation in minimum temperature (pπ.005). Estimates for peak RSV transmissibility, however, were not correlated with the timing of the seasonal trough in minimum temperature or vapor pressure ( Table 3 ). A strong and significant positive correlation was observed between both the amplitude and peak timing of RSV seasonal forcing and the seasonal variation in PET (pπ.0001) ( Table 3 , Fig. 3D ). However, the state-to-state variability in the timing of peak RSV transmissibility was more than twice the observed variability in the timing of PET troughs ( Fig. 3D ).

The estimated amplitude of seasonal forcing in RSV transmission (top) and the estimated seasonal offset parameter (bottom: φ =𠂠 represents January 1 and φ = 𢄠.2 represents October 19) is plotted against (A) annual mean vapor pressure (hecta-Pascals), (B) annual mean minimum temperature (ଌ), (C) annual mean precipitation (mm/month), and (D) amplitude (relative to the annual mean) and timing of trough in potential evapotranspiration (PET 0 = January 1, 0.1 =�ruary 6). The colorbar on the right indicates the ratio of the biennial to annual Fourier amplitude for the observed data (outer circle) and fitted model (inner diamond). Select states are labeled: Arizona (AZ), Florida (FL), Georgia (GA), Hawaii (HI), Louisiana (LA), Montana (MT), New York (NY), South Dakota (SD), Texas (TX), Wyoming (WY).

The model was again able to capture the biennial pattern of RSV epidemics apparent in some states. The correlation between the observed and predicted ratio of the biennial to annual periodicities, as estimated by Fourier analysis, was 0.89 (pπ.0005). States with biennial RSV dynamics tended to have strong seasonal forcing (bϠ.25), which was associated with a large amplitude of variation in PET and low minimum temperature, vapor pressure, and precipitation ( Fig. 3 ). In general, the ratio of the biennial to annual Fourier amplitude was slightly greater in the data than predicted by the models this is likely due to random year-to-year variability in the size of RSV epidemics, which is not accounted for in our deterministic models.

It was not possible to explain unusually high or low RSV activity within a given state, apart from the regular biennial patterns, based on any of the climatic variables. Deviations from model-predicted patterns (observed minus predicted monthly RSV lab reports) were not significantly correlated with temperature, vapor pressure, precipitation, or PET (pϠ.05) (S4 Fig.). Furthermore, we were not able to explain year-to-year variation in epidemic timing and size by directly parameterizing variation in the transmission rate based on weekly variations in PET (S1 Text). Such a model also provided a poor fit to the data, as indicated by the log-likelihood (S5 Table).

Pandemic Preparation

A pandemic causes economic and social problems because so many people are ill or can’t work.

Here are a few things you can do to help your family and your community before and during a pandemic:

  • Make an emergency contact list.
  • Find local aid organizations in case you need information, support, or health services.
  • Find out whether you can work from home.
  • Plan home learning activities in case school is closed.
  • Store extra water, food, medicine, and supplies.
  • Stay as healthy as you can by getting rest, managing stress, eating right, and exercising.
  • Help seniors and neighbors by sharing information and resources.

For more information on what to do in a pandemic, call the CDC Hotline at 800-CDC-INFO (800-232-4636) or go to

About the Guest Editors:

Konstantin Blyuss

Guest Editor, PLOS ONE, PLOS Biology, and PLOS Computational Biology

Konstantin Blyuss is a Reader in the Department of Mathematics at the University of Sussex, UK. He obtained his PhD in applied mathematics at the University of Surrey, which was followed by PostDocs at Universities of Exeter and Oxford. Before coming to Sussex in 2010, he was a Lecturer in Complexity at the University of Bristol. His main research interests are in the area of dynamical systems applied to biology, with particular interest in modelling various aspects of epidemiology, dynamics of immune responses and autoimmunity, as well as understanding mechanisms of interactions between plants and their pathogens

Sara del Valle

Guest Editor, PLOS ONE, PLOS Biology, and PLOS Computational Biology

Dr. Sara Del Valle is a scientist and deputy group leader in the Information Systems and Modeling Group at Los Alamos National Laboratory. She earned her Ph.D. in Applied Mathematics and Computational Science in 2005 from the University of Iowa. She works on developing, integrating, and analyzing mathematical, computational, and statistical models for the spread of infectious diseases such as smallpox, anthrax, HIV, influenza, malaria, Zika, Chikungunya, dengue, and Ebola. Most recently, she has been investigating the role of heterogeneous data streams such as satellite imagery, Internet data, and climate on detecting, monitoring, and forecasting diseases around the globe. Her research has generated new insights on the impact of behavioral changes on diseases spread as well as the role of non-traditional data streams on disease forecasting.

Jennifer Flegg

Guest Editor, PLOS ONE, PLOS Biology, and PLOS Computational Biology

Jennifer Flegg is a Senior Lecturer and DECRA fellow in the School of Mathematics and Statistics at the University of Melbourne. Her research focuses on mathematical biology in areas such as wound healing, tumour growth and epidemiology. She was awarded a PhD in 2009 from Queensland University of Technology on mathematical modelling of tissue repair. From 2010 &ndash 2013, she was at the University of Oxford developing statistical models for the spread of resistance to antimalarial drugs. From 2014 &ndash April 2017 she was a Lecturer in the School of Mathematical Sciences at Monash University. In May 2017 she joined the School of Mathematics and Statistics at the University of Melbourne as a Senior Lecturer in Applied Mathematics.

Louise Matthews

Guest Editor, PLOS ONE, PLOS Biology, and PLOS Computational Biology

Louise Matthews is Professor of Mathematical Biology and Infectious Disease Ecology at the Institute of Biodiversity, Animal Health and Comparative Medicine (BAHCM) at the University of Glasgow. She holds a degree and PhD in mathematics and has over 20 years research experience as an epidemiologist, with a particular focus on diseases of veterinary and zoonotic importance. Her current interests include a focus on drug resistance antibiotic resistance in livestock the community and the healthcare setting anthelminthic resistance in livestock and drug resistance in African Animal Trypanosomiasis. She is also interested in the integration of economic and epidemiological approaches such as game theory to understand farmer behaviour and micro-costing approaches to promote adoption of measures to reduce antibiotic resistance.

Jane Heffernan

Guest Editor, PLOS ONE, PLOS Biology, and PLOS Computational Biology

Jane Heffernan is a Professor in the Department of Mathematics and Statistics at York University, and York Research Chair (Tier II). She is also the Director of the Centre for Disease Modelling (CDM), and serves on the Board of Directors of the Canadian Applied and Industrial Mathematics Society (CAIMS). She is also very active in the Society for Mathematical Biology (SMB). Dr. Heffernan&rsquos research program centers on understanding the spread and persistence of infectious diseases. Her Modelling Infection and Immunity Lab focuses on the development of new biologically motivated models of infectious diseases (deterministic and stochastic) that describe pathogen dynamics in-host (mathematical immunology) and in a population of hosts (mathematical epidemiology), as well as models in immuno-epidemiology, which integrate the in-host dynamics with population level models. More recently, Heffernan is focusing on applying mathematics and modelling to studying pollinator health and disease biology.

What to know about pandemics

A pandemic is an outbreak of global proportions. It happens when infection due to a bacterium or virus becomes capable of spreading widely and rapidly.

The disease behind a pandemic can cause severe illness and spread easily from one person to the next.

As of March 2020, the world is currently dealing with a global outbreak of COVID-19. On March 11, the World Health Organization (WHO) advised that this disease has the characteristics of a pandemic.

Many governments have now restricted free movement and placed populations under lockdown to limit the spread of the pandemic.

In this article, we discuss the difference between epidemics and pandemics, how pandemics start, and future concerns.

Share on Pinterest During a pandemic, governments may restrict free movement and put populations under lockdown.

According to the WHO, a pandemic involves the worldwide spread of a new disease . While an epidemic remains limited to one city, region, or country, a pandemic spreads beyond national borders and possibly worldwide.

Authorities consider a disease to be an epidemic when the number of people with the infection is higher than the forecast number within a specific region.

If an infection becomes widespread in several countries at the same time, it may turn into a pandemic.

A new virus strain or subtype that easily transmits between humans can cause a pandemic. Bacteria that become resistant to antibiotic treatment may also be behind the rapid spread.

Sometimes, pandemics occur when new diseases develop the ability to spread rapidly, such as the Black Death, or bubonic plague.

Humans may have little or no immunity against a new virus. Often, a new virus cannot spread between animals and people. However, if the disease changes or mutates, it may start to spread easily, and a pandemic may result.

Seasonal influenza (flu) epidemics generally occur as a result of subtypes of a virus that is already circulating among people. Novel subtypes, on the other hand, generally cause pandemics. These subtypes will not previously have circulated among humans.

A pandemic affects a higher number of people and can be more deadly than an epidemic. It can also lead to more social disruption, economic loss, and general hardship on a wider scale.

Writing in March 2020, the current pandemic has had an unprecedented impact across the globe.

COVID-19 is a disease that develops due to infection with a type of coronavirus. The virus started causing infections in Wuhan, China, before spreading internationally.

On the recommendation of the WHO, more than one-third of the world’s population is on lockdown. Several countries — including the United States, United Kingdom, India, and China — have closed their borders, affecting global travel and industry.

People in many countries have also lost employment as a result of “nonessential” businesses closing to restrict the spread of the virus. Restaurants, gyms, religious buildings, parks, and offices have closed in many places.

A pandemic can also increase the pressure on healthcare systems by raising the demand for certain treatments.

People with severe COVID-19 symptoms use more ventilators and beds in intensive care. As a result, resources may be in short supply for others who need this equipment.

However, countries have put in place measures to counter this. For example, the U.S. government has requested that companies, including Ford and General Motors, start making respirators, ventilators, and face shields to meet increased demand.

Authorities hope that these emergency manufacturing measures and the restrictions of movement — which have a worldwide economic and social impact — will slow the spread of the disease.

Countries are collaborating on sourcing medical equipment and developing a vaccine, even though it may not be available for months or even years.