Does the California internal coronavirus model show a peak from mid to late May?

See full version for details Click here for Coronavirus section.

At a press conference last week, California governor Gavin Newsham and Secretary of Health and Human Services Dr. Mark Gurley said internal modeling suggests that California will peak in the curve of the coronavirus pandemic in California. May reported to residents of the country despite extensive social isolation measures.

“I know that bending and flattening a curve has two implications.” Girly said. “It means our peak goes down, but it also means it goes further. We move it lower and farther. Down, it moves it Yes. “

This prediction is inconsistent with other models in California. A very influential IHME model Used by federal-level policy makers.

The IHME model—a model that also considers state social distance initiatives—estimates that California hospitals will peak in resource use on April 13 and state deaths will peak on Wednesday .

In addition, the top line figures for Kaiser Permanente internal models, reviewed by SFGATE, indicate that the managed care consortium is expecting peak resource usage sometime in late April. The timeline is one to two weeks behind IHME’s forecast, but almost a month ahead of California’s internal forecast.

So why not?

SFGATE contacted the California Emergency Services Department for information on the science behind the state’s internal model, but received no response.

Dr. George Rutherford, an epidemiologist at UCSF, stated that the state model used two other models. University of Pennsylvania CHIME model And Stanford University Model — Make your own predictions. Rutherford is not completely convinced of how the state weights each model, but believes that relying on the CHIME model may overestimate the spread of outbreaks.

“The CHIME model is very sensitive to case doubling over time, which is purely a test feature,” Rutherford said of the interactive model. “I put some numbers in the case of twice 3 days instead of 4 days, and there was a huge difference. The doubling time is a function of the test, I don’t want to wear a hat. “

California testing conditions are beginning to be delayed Large backlog of unreturned tests State lawsuits have distorted doubling the number of hours.

In addition to highlighting the time to double the case, some question the number of infections used by the internal state model. John Ioanidis, a professor of medicine, epidemiology and collective health at Stanford University, I told POLITICO last week He believes that state models use infection rates as high as 80%. This is the worst “unrealistic” scenario “based on early observations elsewhere in the past”.

It is unclear whether the California model estimates based on the occurrence of New York, but Ioanidis said this was inappropriate because “ most of the world is not so dense and messed up. ” I think it will be great advice.

Newsom has recently stated that the state model shows a flattened curve, while other models show the same flattening without boosting the peak of the curve in May. For example, the IHME model reduced daily mortality multiple times, predicting the number of California hospital beds needed and peaking in mid-April on the same day in the past four weeks.

“All models are based on which data the data is based on,” said Dr. Alexei Wagner, assistant director of adult emergency medicine at Stanford University School of Medicine. “So if it looks like last week compared to this week [in the IHME model], The California figures are much flatter and the peaks are lower. And I think there is a sense that the work we have done to keep the social distance apart, social isolation, is surely beginning to flatten the curve. “

So what is the best data to feed the model that most accurately predicts both the height of the California peak and the expected timeline?

Rutherford believes that the answer is in-hospital data and states that the UCSF model is based on these figures because they are “hard numbers.”

of Bay Area hospitalization data has leveled off during the month of April. Rutherford believes that the Bay Area will peak earlier this week, a bit ahead of statewide IHME model predictions.

“The IHME model is a mixed model, reflecting what is happening in the Bay Area, Los Angeles, and Central Valley, and is not fragmented,” he said. [this] Weeks peak, followed by Los Angeles a week later. Los Angeles cases are much higher, such as 10 times higher than San Francisco … A week after statewide shelters were established, they grew very rapidly in Southern California, mainly driven by LA That’s because the Bay Area entered the shelter three days ago. “

If Rutherford, IHME Modeler, and Kaiser Permanente predict that California will peak and “go down the curve” sooner than the state’s internal model predicts, the resumption timeline may move up if correct There is.

inside Appear on CNN’s Anderson Cooper 360 on Friday night Newsom stated that he would follow the real trends the state is witnessing in real time over the next few weeks when making his decision.

“As we see this mild growth begin to turn in the other direction in the coming weeks, I will build on the expertise of our healthcare professionals, and based on where this virus is, Extensive community surveillance, or testing, is based on more and more decisions. “

The governor also warned against “running a 90-yard dash against this virus” and “being ahead of me” regarding a discussion about resumption, a shared sentiment among health professionals .

“If the media says” Oh, not too bad “and something changes or people go out, I’m worried that another spike or climb might be seen later. “Wagner said. “Or is this coming back in the fall? And that’s why everyone was a bit surprised. But we don’t have the big waves we planned earlier, We are fully aware and vigilant that the number of patients is likely to increase as more patients work, but we have the ability to care for them. “

Rutherford agrees, he says, and understands why California’s internal model was aimed at potential worst-case scenarios.

“If you want to miss, you want to miss it,” he said. “If you miss the lows, the results are incredible. There is no increase in numbers. Pay close attention. I saw what happened in Italy and New York. It’s difficult to measure, but I don’t need to procure a bunch of hospital beds. “

More Coronavir Coverage:

Sign up for “The Daily” newsletter for updates on coronavirus here.

Eric Ting is a SFGATE digital reporter. Email: [email protected] | Twitter:@_ericting