Models predict new COVID-19 surge could bring 25% more deaths

This story was written and produced by NJ Spotlight. It is being republished under a special NJ News Commons content-sharing agreement related to COVID-19 coverage. To read more, visit

By Lilo H. Stainton for NJ Spotlight

Statistical tools that closely mapped earlier cases, death tolls now show more hospitalizations, fatalities looming

Gov. Phil Murphy
New Jersey Gov. Phil Murphy speaks at a COVID-19 briefing. Photo by Rich Hundley III

Editor’s Note: This story has been updated to reflect changes to the COVID-19 model from the University of Washington, which were posted late Friday night. 

New Jersey’s hospitals could once again fill with COVID-19 patients and the state’s death toll could increase by as much as 25% by February, according to one respected national model, and that’s if people keep on wearing masks and social distancing.

If those now-common infection control mandates are eased, the number of confirmed and probable coronavirus fatalities could reach even higher — with an average of 68 people dying daily by then  — predictions from the Institute for Health Metrics and Evaluation at the University of Washington suggest.

If the public declines to wear masks and follow state public health mandates, the model predicts that come February more than 23,000 New Jerseyans may have lost their lives to COVID-19, under the worst-case scenario. If people comply with the restrictions, fatalities could be kept under 17,000, if all goes well, IHME found.

New Jersey officials have employed different modeling, using a tool called CHIME developed by the University of Pennsylvania, which state Department of Health Commissioner Judy Persichilli has said was highly accurate in its predictions related to hospital resources. The Washington university model was also relatively effective in anticipating the volume and timing of the state’s peak hospital surge, in mid-April.

The DOH did not provide specific prediction figures. But representative Dawn Thomas said, “The models are currently depicting increases in cases and hospitalizations from the virus for the late fall and winter months.”

Generally speaking, these predictive models rely on factors including coronavirus testing data, the rate of its spread, hospital capacity, historical predictors and the level of public compliance when it comes to public health restrictions, like maintaining distance and wearing masks — something that has been hard to measure, especially at the start. Persichilli told legislators last month that her team eventually settled on 50% compliance, which appeared to be accurate, in retrospect.

Historically, public health experts have used 31% compliance as a modeling benchmark, Persichilli said — but at that rate, patients would have overwhelmed the state’s hospital capacity. She said the DOH team first started with an input of zero compliance, which predicted a flood of 80,000 COVID-19 patients at the peak this spring. Instead, the surge peaked at closer to 8,000 patients.

“I can remember the Saturday that they showed me that predictive modeling in this conference room, and I looked at the team and I said, ‘We’re not going to make it,’” Persichilli recalled in September while testifying before the Assembly Budget Committee.

‘50% social distancing saved lives’

Modeling that used a 31% compliance rate resulted in lower predictions for patient volume, but still beyond what could be accommodated by New Jersey’s 71 acute care facilities, plus the field hospitals and other sites the DOH was setting up, Persichilli said. But 50% compliance produced a number within the range of what they could handle. “That’s what we hit, on the mark,” she said. “So 50% social distancing saved lives.”

Thomas, from DOH, said they continue to run the model with compliance rates of 31% and 50%, and she stressed that many factors are considered in the process.

As of Friday, nearly 3.9 million coronavirus tests had been performed in the state, and more than 212,000 New Jerseyans had tested positive for the virus, including at least 14,400 who have died.

Real-life experience has impacted the modeling in various ways, according to the DOH, although limited details were available last week.

“The state is using multiple modeling tools and analytics for cases, hospitalizations and hospital resources,” Thomas said. “As our knowledge of the virus has grown, our modeling has become more robust including multiple scenarios and factors. Using multiple scenarios allows for vigorous preparation and stockpiling in the event of a resurgence.”

While new daily cases and hospital volume have declined significantly since the initial surge, there have been occasional spikes, and state officials were clearly concerned about the rising trends last week. On Thursday, New Jersey logged an additional 1,300 COVID-19 cases — the highest daily total since late May — and added 880 more on Friday, with much of the increase in Monmouth and Ocean counties.

State officials suggested that while models predicted an increase, the recent spikes may be something the formulas didn’t anticipate.

“You know, this is somewhat within the modeling … of what we expected,” Murphy said at his media briefing Thursday, noting that officials were watching the numbers closely.

“With the exceptions of the hyper flare-ups that we’ve seen in places like Lakewood or at Monmouth University or Rowan (University), the modeling suggested at a certain point we’re going to start to see numbers like this,” he added.

Murphy has largely attributed the rise to indoor gatherings, not the recent reopening of restaurants or schools, and said he would take a cautious, local approach when considering reversing these steps to reopen society.

“I lean more scalpel than blunt instrument, but they both have to be on the table,” he said.

At Thursday’s briefing, Persichilli said that while case numbers are expected to go up, public compliance with the ongoing mandates can help reduce the impact on health care facilities going forward. “We are anticipating a second wave and we are preparing based on lessons learned from our prior experiences. If individuals do not adhere to social distancing, masking guidelines, washing your hands, staying home if you’re sick, this wave has the potential to become a surge,” she said.

“We know that the virus did not take a break. It is presenting itself in younger individuals who are experiencing mild and moderate symptoms,” Persichilli continued. “As the cold weather sets in and people move indoors, the threat of spread is even greater. We also know that we will be fighting this enemy for a longer period of time.”

The limits of predictive modeling

Dr. Ed Lifshitz, who heads the DOH’s communicable disease service, also said Thursday he was not surprised by the increase in new coronavirus cases, based on the modeling the department has done. But he also said the numbers from some areas were “higher than would be expected” and a cause for concern. “Modeling takes you so far; modeling is not 100% predictive as far as things go, and certainly we’re keeping a very close eye on this,” he said.

“What we’re seeing overall is spread in many different places, some associated with colleges and other places. But I do strongly want to encourage people — I know that they’re tired, I know that this is hard, I know that they’ve been doing this for a long time with all that is happening — but wear the mask, social distance,” Lifshitz said. “You need to do this continually, regularly, all the time. You always need to be thinking about it.”


  1. Ivor Cummins has an excellent video review of recent data and evidence. Search youtube for Ivor Cummins September 8 Viral Update — or visit

    The IHME (U Washington) modeling available at has been comincally / tragically inaccurate. The predictions for NJ hospitals being overrun were off base. At recent check (10/13/2020 5 PM), past hospital resource usage still showed NJ hospitals overrun. This of course did not happen, not even close. Yes, some facilities were stressed and many healthcare workers performed heroically.

    Gov. Murphy presented in briefings throughout June showing peak hospital usage in mid-April. There was ample capacity and ample ability to stand up surge capacity if needed.

    Meanwhile, the costs of mitigations continue to mount. The costs are not merely lost economic output — although that is devastating in itself. “Economic output” means jobs. Having a job matters. Job security matters. More than six months on, every job is essential, because each person’s hopes and dreams for the future are essential.

    Failure to keep up with routine wellness activities, cancer screenings, vaccinations. Reticence to deal with other urgent or emergent medical conditions. Physical and psychological impacts have health consequences that echo for years, whether it is gaining fat, loss of mobility from too much Zoom-ing, or despair setting in from loneliness.

  2. Thanks PrincetonRez. I appreciate your taking the time to respond.

    Cummins is an engineer by training and not even a doctor.

    His argument is based on records of population experience.

    Criticize Cummins’ content. Cummins gives citations within the video and links to many scientific sources. I would love to see further evidence. A specific criticism of Cummins’ content, with a time stamp to the video, and evidence (not just a model) supporting an alternative view would be great.

    Of particular note in Cummins’ video is the discussion of rising positive test results without (for now) corresponding rise in bad disease outcomes. This part begins at 24:15 into the video to end (total length of the video is 37:35).

    The NY Times article cites experts saying 85% to 90% of US population is susceptible to COVID-19. These experts base this assessment on prevalence of specific antibodies.

    As mentioned in the article, there are other immunity mechanisms besides specific antibodies. For example, T-cells.
    Reference here, almost certainly not the last word on the subject:
    Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals

    Cummins’ video is worth watching in its entirety. Set the speed at 1.5x.

Comments are closed.