Epidemic Forecasting Models Overestimated Potential Infection Numbers Amid COVID-19 Lockdown Debate

The Imperial College London’s model predicted 40 million deaths in the first year of COVID-19 while the real count is 6.9 million deaths altogether.

A worker sits next to a fence close to a residential area during the COVID-19 lockdown in the Huangpu district of Shanghai on June 10, 2022. Hector Retamal /AFP via Getty Images

A new study has shown that flawed epidemic modeling techniques can lead to overestimating the number of people who could get infected during a pandemic—resulting in unnecessary measures such as lockdowns and mass vaccination campaigns.

The peer-reviewed study, published in the Journal of Physics Complexity on Jan. 9, associated existing models of forecasting epidemics with the structure of social networks among people. The most widely used forecasting method is a “compartmental model,” which usually makes an assumption of “random mixing,” meaning that any individual can infect any other person. However, this is a flawed assumption that can lead to “greatly overestimating the number of infections,” the study pointed out.

“People are in fact connected according to a social network” that tends to be heterogeneous as some people have a lot more contacts than others.

“COVID-19 is driven largely by ’super-spreading events’ (SSEs), with one estimate suggesting that fewer than 10 percent of infectious individuals accounted for 80 percent of infections,” the study said. “Some SSEs saw over 100 people apparently infected by a single individual within a few hours.”

Because social networks are heterogeneous and some people have more contacts than others, epidemic waves are smaller than they are predicted to be by existing standard models.

To test this proposition, the study looked at two epidemic models—one with random mixing and another with a heterogeneous network. An analysis found that the former predicted that almost 90 percent of individuals would become infected and the heterogeneous model predicted only a 20 percent infection rate.

“Not everyone has the same numbers of friends, family, and colleagues, or goes out to places where large groups of people may be present,” said Samuel Johnson, an associate professor in applied mathematics at the University of Birmingham who conducted the study, according to a Jan. 9 press release.

“And the fact that superspreader events play such a significant role in the early stages of an epidemic supports the hypothesis that the real network of contacts is, like many other social networks, highly heterogeneous.”

He said that taking social networks into account should be a “fundamental part” of epidemic modeling even if details about such networks are not known.

Underestimating Infection Waves

The study highlighted another major flaw in current epidemic modeling—underestimating the number of waves in a pandemic.

Existing models propose that recovered people can never again become infected. “In reality, we know that diseases such as COVID-19 can re-infect, either because of waning immunity or new variants,” the study reads.

The COVID-19 pandemic involved multiple waves of infections, which some experts attributed to factors such as more infectious viral variants, waning vaccine immunity, and changing non-pharmaceutical interventions (NPIs). However, these waves could be explained by modeling flaws, the study said.

“Once an epidemic has petered out naturally, it is often assumed that herd immunity must have been achieved, and the population is no longer vulnerable unless immunity wanes or transmissibility increases significantly,” the authors wrote.

But even after the first wave of infections has died down, there still remains a “large pool of susceptible individuals.” And even if herd immunity has been achieved, these individuals can infect other people and start new waves of infections, the research stated.

The study thus made three conclusions:

• Each “wave” of a disease such as COVID-19 may infect fewer people than assumed, even in the absence of NPIs due to network heterogeneity.
• If networks are heterogeneous and change in time, this can lead to multiple waves of infection that would not be predicted by random-mixing models.
• NPIs focused on avoiding super-spreading events are likely to be particularly effective at controlling the epidemic.

Flawed Modeling Consequences

In March 2020, the Imperial College London’s COVID-19 Response Team modeling projected 7 billion infections and 40 million deaths in the first year of the epidemic if lockdowns were not imposed.

In an interview with The Epoch Times in November, Dr. Ari Joffe, a clinical professor of pediatrics at the University of Alberta in Canada, said that the Imperial College modeling of COVID-19 infections only ended up generating widespread fear.

As a consequence, he fully supported government-imposed lockdowns at the time as he believed such measures “would reduce viral transmission and deaths, as famously, inaccurately, and tautologically modeled at Imperial College.”

However, Dr. Joffe said that as the pandemic unfolded, he began to reconsider his position. “In the first few months of lockdown, I realized that my (and similarly trained medical colleagues’) expertise was poorly suited to give advice during a pandemic,” he said.

The professor admitted not noticing that the high-risk groups in the Imperial College’s modeling were people aged 70 years and above and those aged 60–69 years who had severe comorbidities.

“The modeling was flawed, and in general, modeling (forecasting) failed during the pandemic. This was because the models were based on flawed assumptions and non-transparent methods,” Dr. Joffe said.

“If you put in inaccurate assumptions (e.g., the infection fatality rate was way too high; the population was modeled as homogeneous, when in reality, it is highly heterogeneous in terms of risk and exposure; the outbreak was modeled as never-ending exponential increase, unlike any epidemic in history; the herd immunity threshold was assumed to be far too high; and more), the model will show what you want it to show.”

Doubts are being raised about the lockdowns and vaccination campaigns imposed on people by various governments that implemented the policies by citing the epidemic modeling.

A Jan. 6 study published in the Journal of Clinical Medicine said that it could not find “clear or consistent evidence” that NPIs or vaccinations “reduced the progression of the pandemic.”

Speaking on “Real Time With Bill Maher” last year, Scott Galloway, a marketing professor at New York University’s Stern School of Business, admitted that he was wrong in pushing for harsh pandemic lockdown policies.

“I was on the board of my kids’ school during COVID. I wanted a harsher lockdown policy, and in retrospect, I was wrong. ... The damage to kids from keeping them out of school longer was greater than the risk,” he said.

According to data from Worldometer, more than 702 million people had been infected by Jan. 20, 2024. Total reported global deaths were 6.9 million.

Katie Spence contributed to the report.