Copyright © 2020 Albuquerque Journal
SANTA FE – The numbers are baffling.
Thousands of people could die, one statistical model suggests, as the coronavirus tears through New Mexico in the next year. Another model, by contrast, estimates 155 deaths by this summer.
Each projection, meanwhile, produces new figures almost daily.
But even amid the uncertainty, scientists and state health officials say, the modeling is valuable. Like a hurricane forecast, it can help guide policymakers who have to decide now, not later, how to prepare.
State health officials reported eight more deaths Thursday – the most in one day. The state says COVID-19, the disease caused by the coronavirus, has contributed to 44 deaths this year.
Disease models “are not intended to be a perfect prediction of the future,” said Kathryn Hanley, a virologist and biology professor at New Mexico State University. “No one can give you that.”
New Mexico health officials are tapping the expertise of scientists at Sandia and Los Alamos national laboratories as they try to forecast the demand for hospital beds, ventilators and other medical resources amid the pandemic.
“There’s a saying: All models are wrong, but some are useful,” said Bradley Dickerson, senior manager of a Sandia group that counters global chemical and biological security threats.
Two main models have grabbed attention in New Mexico – one by the University of Washington and another used by New Mexico health officials.
In late March, researchers at the University of Washington projected about 510 coronavirus deaths in New Mexico through late June or early July – the first wave of the pandemic. But they’ve repeatedly revised the estimates downward in recent weeks.
On Thursday, the model was projecting just 155 deaths in New Mexico through late June.
New Mexico health officials often deliver a far more dire forecast. The latest estimates they released – from last week – projected 2,984 deaths in New Mexico over the next 12 months, along with severe shortages of intensive care beds and ventilators.
New Mexico Gov. Michelle Lujan Grisham and Human Services Secretary David Scrase have repeatedly pushed back on the Washington model, describing it as far too optimistic.
“We still think it’s too low,” Lujan Grisham said Thursday of the Washington projection. “I wish it wasn’t.”
The University of Washington model is largely driven by the number of deaths in a community, Scrase said, with infection rates and other factors inferred from how many people have died.
But New Mexico has had relatively few deaths – just 44, including eight announced on Thursday – and is still early in its outbreak, making the Washington approach, Scrase said, less reliable.
Instead, New Mexico health officials are using what they call an SIR Model, which examines the rates of people who are susceptible to the disease, infected and recovered. It projects the number of deaths based on how many people are infected and similar factors.
And the local model is projecting far more deaths than the Washington one.
Scrase and Lujan Grisham say COVID-19 may be particularly dangerous in New Mexico because of the state’s high rates of chronic liver disease and other conditions – all factors they said the Washington model doesn’t consider.
Lujan Grisham said cultural or genetic factors may also influence the disease. The state’s highest rates of coronavirus infections are in McKinley, Sandoval and San Juan counties, where outbreaks have hit two pueblos and the Navajo Nation.
Lujan Grisham said states throughout the country are generally using their own models based on local data, not the Washington projections.
New Mexico’s modeling team meets by phone every Tuesday afternoon. Experts from Sandia and Los Alamos national laboratories, Scrase said, are on the call.
They go over several models that project how the disease might move through New Mexico. The team considers the Washington projections, New Mexico’s own model and other forecasts.
But state health officials, Scrase said, need the flexibility of having their own New Mexico-specific model. The local projections can factor in the latest information on how many people have tested positive each day and produce updates based on the latest data.
Models from other states don’t offer that flexibility.
Modeling “is sort of like buying a car,” Scrase said in a public briefing this week. “If you went to buy a car and they wouldn’t let you test-drive it, you probably wouldn’t want to buy the car. We need to be able to put our own New Mexico data in.”
Scrase said the local model produces new projections as more information becomes available.
New Mexicans’ willingness to stay home and engage in social distancing, he said, has helped flatten the projected growth in cases and pushed back the projected peak, which could come in late April or May.
Deadliest day yet
The number of deaths in New Mexico surged this week.
The record eight deaths announced Thursday include people ranging in age from their 30s to their 80s – including five adults in San Juan and McKinley counties, which include the Navajo Nation.
All but two of the deaths involved people with underlying health conditions.
Altogether, state health officials reported 116 new cases of COVID-19, for a total of 1,597 confirmed cases.
Ninety patients are now hospitalized with the disease, and 353 people are classified as having recovered.
Predicting how high the death toll will climb is incredibly difficult, scientists say.
Changes in assumptions lead to an almost breathtaking difference in the potential outcomes.
“There’s so much uncertainty because this is a new virus and so there’s a lot of information we don’t know,” said Sara Del Valle, a mathematical epidemiologist at Los Alamos National Laboratory.
The unknowns include the number of people who are infected but don’t have symptoms, how contagious those people are, and whether people who have survived the disease are immune from getting it again.
“In addition,” Del Valle said, “human behavior is a critical factor that contributes to disease spread, and we don’t have real-time information about what people do or will do.”
Dickerson, at Sandia, has been working with the state on projections for medical resources. Even with uncertainty, he said, the models are useful – as long as the decision-maker is comfortable with the model and its limitations.
“I think any of these models are better than weather predictions,” Dickerson said.
Hanley, a regents professor of biology at NMSU, said one of the challenges that people find difficult to grasp is how their community’s response to a disease will affect its spread. Throughout history, she said, communities have nearly eradicated one disease or another only to stop – because the effort is so successful people begin to doubt the disease is actually a threat.
“That is just human nature,” Hanley said, “but it’s led to a number of failed health actions because they stop too soon.”
Note: Biology Professor Kathryn Hanley of New Mexico State University isn’t part of the modeling team producing projections for New Mexico health officials.