Hospitalization rates remain the most accurate scalable data. But even flawed data can help social scientists estimate the scope of the virus and provide better guidance for stopping the spread. JL
Jo Craven McGinty reports in the Wall Street Journal:
For every documented case of Covid-19, there are at least two undetected infections, and the unusually large number of deaths that occurred last year suggest the virus killed more people than the data record. By the end of 2020, 450,000 excess deaths occurred. Standardization is an issue, missing people who are asymptomatic or who were mildly sick and never tested. The way states calculate positivity rates is a grab bag. The best publicly available data are hospitalizations. In real time, testing, case counts, hospitalizations and death tallies help health-care workers and policy makers track the disease, craft strategies to curb its spread and allocate sometimes scarce resources.When the daily demands of pandemic management subside, researchers will want to survey the damage wreaked by Covid-19 and assess how health-care systems and policy makers responded to the scourge.
The exercise will help prepare for the next viral assault. But, one year after the first U.S. patient was hospitalized with Covid-19, the data needed to answer basic questions about the virus’s spread are—in the words of some experts—a mess.
Testing has been uneven across states and communities. Case counts, missing asymptomatic and mild infections, are too low. And death tallies are believed to be incomplete.
Now Trending
Covid-19 data collected by states in real time is useful for tracking the virus but suffers from a lack of standardization and in some cases is incomplete.
U.S. daily confirmed Covid-19 cases
300,000
200,000
100,000
0
March 2020
ʼ21
U.S. daily confirmed Covid-19 deaths
3,000
2,000
1,000
0
March 2020
ʼ21
U.S. hospitalized Covid-19 patients
150,000
100,000
50,000
0
March 2020
ʼ21
Total U.S. test results
300
million
200
100
0
Jan 2020
ʼ21
Note: Seven-day rolling average;
as of Jan. 20. (cases, deaths)Sources: Johns Hopkins CSSE (cases, deaths); Covid Tracking Project (hospitalizations and tests)
According to some estimates, for every documented case of Covid-19, there are at least two undetected infections, and the unusually large number of deaths that occurred last year suggest the virus might have killed more people than the data record.
By the end of 2020, nearly 346,000 deaths in the U.S had been attributed to Covid-19, according to Johns Hopkins University, but the Centers for Disease Control and Prevention estimates that 450,000 excess deaths occurred that year.
“That’s 450,000 more than we would have in a normal year,” said Robert N. Anderson, chief of the mortality-statistics branch of the CDC’s National Center for Health Statistics, which uses death certificates to track how many people die annually. “We already know we had well over 300,000 deaths due to Covid-19. Of the other group, probably 100,000 can be attributed to the pandemic but not necessarily to the virus.”
Currently, the best publicly available data are hospitalizations.
“We see it as one of the main bright spots in the whole Covid response,” said Alexis Madrigal, co-founder of the Covid Tracking Project, a volunteer organization that collects and publishes pandemic data. “It’s the best, most granular data set we have.”
Initially, the Health and Human Services Department released only state-level information, but since December, the data have provided hospital capacity and bed use for each facility, exposing which ones are overloaded.
“In the beginning, if that had been available, it might have stopped disinformation by showing how widespread it was and backing up the incredibly moving testimony of health-care workers about what was happening in hospitals,” Mr. Madrigal said.
In real time, testing, case counts, hospitalizations and death tallies help health-care workers and policy makers track the disease, craft strategies to curb its spread and allocate sometimes scarce resources.
In retrospect, analyzing the data sets—even if they are imperfect—can help evaluate the effectiveness of the decisions that were made and improve future responses.
“A lot of the reason we have struggled in the U.S. is because we have such a diffuse approach to health care,” said Beth Blauer, executive director of the Johns Hopkins University Centers for Civic Impact. “It has led to states implementing things pretty differently, and it has made it very difficult to create an apples-to-apples comparison.”
The first hurdle was testing.
Initially, tests used to identify Covid-19 were severely limited. When they became more accessible, they were administered inconsistently and reported differently across jurisdictions.
‘A lot of the reason we have struggled in the U.S. is because we have such a diffuse approach to health care.’
Some places counted only antibody tests performed by laboratories, while others also counted less-sensitive antigen tests performed outside of labs.
States then used different methods to calculate positivity rates—and because some people were tested multiple times, it raised questions about whether the rates should be based on the number of tests administered or the number of individuals tested.
“We thought it was really important to understand how many people were tested and what the yields were,” Dr. Blauer said. “It was not that straightforward.”
The way states calculate positivity rates is still a grab bag, she said, but the testing data might still be valuable for long-term research.
“When we compare numbers between states and counties, we can ask why was one place spared and another wasn’t?” said Jennifer Nuzzo, associate professor of epidemiology at the Johns Hopkins Bloomberg School of Public Health. “Was it really spared or did it just not get tested?”
Standardization is also an issue with case counts.
According to the Council of State and Territorial Epidemiologists, standardized surveillance is necessary to provide consistent case identification and classification, measure the potential burden of illness and inform the public-health response.
But some states report only confirmed cases, while others also report probable cases—a definition that has shifted over time.
Even with perfect reporting, case counts don’t capture the total number of infections. Missing are people who are asymptomatic or who were only mildly sick and never tested.
Still, the counts—perhaps augmented by additional research—will help epidemiologists estimate the number of infections. And causes of death listed on death certificates will help them deduce how many people actually died of the disease.
“We would figure out if this is a cause of death that is associated with Covid and if we can safely say these are probably Covid deaths,” Dr. Anderson said. “For some, it will be straightforward. If we get excess pneumonia deaths in the Covid pandemic, we know Covid commonly causes pneumonia.”
It’s more complicated to look at, say, heart disease.
“We do know Covid can cause a heart attack or a stroke,” Dr. Anderson said. “But it’s also possible that some are just heart-disease or stroke deaths. That’s where it’s going to be really tricky to sort. We would probably do models according to different assumptions.”
The CDC won’t alter the underlying records, but states have the opportunity to amend the death certificates, and a handful have been changed to reflect fatalities caused by Covid-19 that occurred before the first U.S. case was diagnosed on Jan. 20, 2020.
Two were on Jan. 4, 2020, and three were on Jan. 18, 2020.
“We’re never going to have a perfectly accurate count, but we can get a sense of how accurate it’s likely to be,” Dr. Anderson said.
Data collected by groups including the CDC, HHS, Johns Hopkins and the Covid Tracking Project are publicly available, and now President Biden has pledged to create a nationwide pandemic dashboard with ZIP-Code-level data.
All the information will be used to prepare for the next pandemic, Dr. Anderson said, which, God willing, won’t be for another 100 years.
0 comments:
Post a Comment