IHME Model Nearly Doubles Coronavirus Death Projections to 134,475

New York
John Nacion/STAR MAX/IPx/AP, File

The Institute for Health Metrics and Evaluation (IHME) released an updated epidemiological model on Monday that nearly doubled projections of coronavirus deaths in the United States by August 4 from 72,433 in its April 29 release to 134,475 in Monday’s May 4 release.

“These projections are considerably higher than previous estimates, representing the combined effects of death model updates and formally incorporating the effect of changes in mobility and social distancing policies into transmission dynamics,” the IHME acknowledged in a statement accompanying the release of the updated model.

As of May 5, there are 72,106 coronavirus deaths in the United States, according to Worldometers.

IHME coronavirus death projections have bounced around significantly from the initial projection of 81,114 when the first version of the model was released on March 26. On April 1, the projection jumped to 93,765, then dropped back to 81,776 on April 6. On April 8, the IHME coronavirus death projections dropped to 60,405, then jumped to 68,841 on April 13, before dropping back to 60,308 on April 17.

White House Coronavirus Task Force coordinator Dr. Deborah Birx expressed concerns about the methodology used in Monday’s update of the IHME model on Tuesday, pointing out, “that particular model is very sensitive to actual any changes in the mortality and very much utilizes that data,” as Breitbart News reported:

She also said that the model likely failed to note the continued mitigation parameters in both phases of the reopening plan.

“I’m not sure all of the mitigation parameters were taken into account,” she said. . .

Birx also said that the IHME model likely had not taken into account the task force guidelines for asking seniors and more vulnerable communities to continue sheltering in place in both phases of the reopening America plan.

Since April 17, the IHME coronavirus death projections have moved upward, first to 65, 796 on April 21, then up slightly to 67,641 on April 22, and up again five days later to 74,473 on April 27. Following a slight drop to 72,433 on April 29, the IHME coronavirus death projections have nearly doubled to 134,475 in the May 4 update.

When the IHME model was first introduced on March 26, it projected that many states would experience coronavirus patient hospitalization requirements well in excess of their existing capacities. Those projections were vastly overstated, and fears that coronavirus patients requiring regular hospital beds, ICU hospital beds, or ventilators would not have them available were never realized.

The May 4 update of the model does not address a major flaw previously identified in that it does not include actual hospitalization data at the state or national level, even though most states have collected that data on a daily basis from early March until the present, with May 4 being the most recent date for which actual hospitalization data is available. In contrast, actual data for coronavirus deaths is included at the state and national levels every day from February 4 to April 30.

In April, Breitbart News reported on this problem with the IHME model’s far off-the-mark hospitalization projections in New York and eight other states:

Breitbart News compared projections of the number of regular hospital beds needed to treat coronavirus patients made in the Friday, April 3 iteration of the IHME model for nine states to the actual number of cumulative hospitalizations (a number that includes current regular bed hospitalizations, current ICU hospitalizations, previously hospitalized and released patients, and previously hospitalized and deceased patients) reported by the respective health departments in those states as of Thursday, April 2.

The results showed that the actual number of cumulative hospitalizations as a percentage of regular hospital beds needed projected by the IHME model ranged from nine percent in Tennessee to 47 percent in Florida. Even in New York State, the epicenter of the coronavirus pandemic, the actual number of cumulative hospitalizations as a percentage of regular hospital beds needed projected by the IHME model was only 37 percent.

These numbers, however, actually understate the degree to which the IHME model overestimates the number of hospital beds currently needed for coronavirus patients because the numerator–cumulative hospitalizations–includes data well in addition to the actual number of current regular hospital beds in use for coronavirus patients, since it also includes current ICU hospital beds in use for coronavirus patients as well as previously hospitalized coronavirus patients either released or deceased.

Though IHME did not include actual hospitalization data in the May 4 update of its model, it did make a number of other additional changes.

“To date, our focus has been on capturing the relationships between social distancing policy implementation and COVID-19 trends: first, based on the timing of policy enactment and, since April 17, also using changes in human mobility patterns to estimate the relative importance of different social mandates,” the IHME statement that accompanied the May 4 update noted.

With Monday’s updated release, the IHME said, “we are directly modeling disease transmission as a function” of four key drivers.

Those key drivers are:

  1. Daily temperature
  2. Percentage of population living in highly dense areas
  3. COVID-19 testing per capita
  4. Changes in human mobility and its relationship to social distancing policies

With regards to daily temperature, the IHME said, “[t]o date, the relationship between temperature and estimated changes in transmission appears to modest for our currently included locations.”

As for the percentage of populations living in highly dense areas, the IHME said, “potentially due to its time-invariant nature, this indicator is not as strong of a predictor of changes in COVID-19 epidemic trends as our other currently included drivers.”

The IHME appears to believe that COVID-19 testing per capita is an important driver because “Changes in testing per capita predictions are related with changes in predicted beta (effective reproductive number, or Rt), such that increases in testing correspond with declines in the transmission parameter. With all else being equal, rising rates of COVID-19 testing contribute to downward trends in epidemic trajectories.”

The IHME also appears to believe changes in human mobility and its relationship to social distancing policies are perhaps the most important driver in disease transmission. Although it goes into great detail in describing the data sources used in modeling human mobility — which include Google’s COVID-19 Community Mobility Reports and Facebook’s Data for Good — it never clearly explains why it believes more mobility of the general population is a driver in COVID-19 transmission.

The May 4 updated model included dramatic increases in coronavirus death projections among five of the hardest-hit states when compared to the April 29 model updates: (48,000)

  • New York increased by 7,817, from 24,314 to 32,132
  • New Jersey increased by 8,798, from 7,246 to 16,044
  • Pennsylvania increased by 6,207, from 2,400 to 8,607
  • Massachusetts increased by 2,063, from 5,634 to 7,697
  • Michigan increased by 3,160, from 3,920 to 7,080

In addition, seven states that are not currently in those hardest hit saw dramatic increases in projected coronavirus deaths in the May 4 updated model when compared to the April 29 update of the model. Some of these states saw projected coronavirus death increases due to what IHME said is “the easing of social distancing policies has begun and mobility patterns are on the rise (or even started increasing before easing actions occurred).”

  • Indiana increased by 5,207, from 1,041 to 6,248
  • Illinois increased by 4,016, from 2,337 to 6,353
  • Georgia increased by 2,904, from 2,009 to 4,913
  • California increased by 2,563, from 2,104 to 4,666
  • Texas increased by 2,344, from 1,288 to 3,632
  • Florida increased by 2,073, from 1,898 to 3,971
  • Alabama increased by 2,015, from 294 to 2,308

About 79 percent of the 62,042 projected increases in coronavirus deaths between the May 4 update and the April 29 update, or about 49,167, were in these 12 states.


Please let us know if you're having issues with commenting.