Last month, a body was being transported to a refrigerated truck at the Kinnsbruck Jewish Medical Center in Brooklyn. Stephanie Keith for The New York Times
Many supporters of Trump president argue that the number death new coronavirus is exaggerated, and Trump himself has pushed questions about the accuracy of some viral data.
That`s right. death numbers seem to be biased —— but not in his direction. We analyzed state data and found that between 100,000 and 110,000 people died from the pandemic, rather than 83,000 people who were listed as new coronavirus pneumonia.
this is my estimate with the help of harvard statistician rafael irissari (Rafael Irizarry), based on a comparison of death rates this spring compared to previous years. Some states are largely unaffected —— some even seem to have fallen death, possibly due to fewer driving and car accident —— but others have seen a sharp rise in death.
Generally speaking, in more than two months, the number of people killed by the new coronavirus in the United States exceeded the number death in the Korean War, the Vietnam War, the Persian Gulf War, the Afghanistan War and the Iraq War in the past 70 years.
These are our estimates; they are based on preliminary data and therefore not the most authoritative, and I welcome your discussion.
First, death reasons are often uncertain. A majority of death did not perform autopsy, and many had never tested for new coronavirus. In a sense, the exact number of new coronavirus deaths is unknown.
Nevertheless, a standard measure of this pandemic impact is to observe "excess death ",, that is, death rates above the average for a given period.
For example, in the United States, in the seven weeks ending April 25, the number of death was about 70,000 higher than normal (the death rate was seasonal, and the spring and summer usually declined). excess death figures for these 70,000 people do not include connecticut, north carolina, and pennsylvania, which are excluded because of missing or suspicious data.
In the meantime, the official death toll in other parts of the United States was 49,100. As of April 25, the number of death associated with the coronavirus was down by more than 20,000.
On the basis of today`s total of 83,000 death, combined with these 20,000 undercounting death, more than 100,000 pandemic-related death. may persist after 25 April, although the proportion has declined.
we do not have sufficient death data to assess excess death at the end of april and early may, during which time more than 30,000 americans were reported to have died of the new coronavirus. And the number of tests has increased dramatically, and doctors seem increasingly willing to list the new coronavirus as death cause.
"The underreporting may be decreasing ," said Robert N· Anderson (Robert N.Anderson), head of statistics at the National Centre for Health Statistics (National Center for Health Statistics) of the Centers for Disease Control and Prevention. Also in New York city, a study found severe underreporting in the first half of April and then gradually decreasing in early May.
During the period of our study, the undercounting situation was also decreasing. initially, the number of excess death was more than twice the number of reported coronavirus death, but by april 25, the number of excess death was only 40% more than the number of reported coronavirus death. If the undercounting rate since then is 10 per cent, it will increase the total number of death by several thousand, possibly bringing the total number of death to close to 110,000.
Although these numbers are uncertain, they mean that, in addition to death statistics, more than 25,000 Americans have died in the pandemic.
This analysis does not determine whether they died directly from the new coronavirus or indirectly. Some people may be afraid to go to the hospital, late call 911, or because ambulances are too scarce, eventually died of myocardial infarction or stroke. In other words, perhaps a small number of people died of the virus, though not infected.
An overwhelming majority of the excess death is believed to be directly caused by the virus, one reason being that in states such as Massachusetts, where the report appears to be more exhaustive, the number of excess death much different from the number of new crown deaths.
As irisari, head of the department of data science at the dana-farber cancer institute (Dana-Farber Cancer Institute) in boston, points out, our new estimates are based on several levels of uncertainty. usually takes two months to report death cases in a reasonably complete manner, so a key question is how to adjust the reported lag. CDC tries to estimate the total number of death based on incomplete reports, and our estimates rely on its algorithms.
death rates vary widely across states. up to now, texas and california seem to death more people than coronavirus in the 2018 flu epidemic. According to the Centers for Disease Control and Prevention, the number of death in Arkansas, Hawaii, Iowa, Kansas, North Dakota, Oregon, South Dakota and Rhode Island actually fell below normal for the seven-week period ending April 25. could be due to reduced driving and reduced unexpected death.
irisari estimates that about 70% of the excess death nationwide comes from five states: New York, new jersey, michigan, massachusetts and illinois.
The notion that official figures are undercounted is widely accepted. On Tuesday, Dr Anthony S· Fauci (Anthony S.Fauci) told the Senate Health Committee that he did not know whether the actual number of death was 50 per cent higher than the official figure, but that the actual number was "almost certainly higher ".
"Most first-line doctors will tell you that these numbers are grossly underestimated ," said Michael P· Jones, a first-aid physician who works at several hospitals in the Bronx. especially early, he said, many new crown death cases were simply listed as variants of "respiratory failure" or "multiple organ failure ".
"We signed so many death certificates, we couldn`t sign them, and then we took turns signing them ," said Dr. Alicia Scaringbas (Alicia Skarimbas), a New Jersey practitioner. "
For obvious cases, she would list the new coronavirus as a death cause, but her colleagues might conclude that death was due to "respiratory failure" unless the virus was tested positive. hence, whether new coronavirus is listed death the reason is usually random.
A fall in death numbers is a global problem, not just that of the United States. Washington University (University of Washington) Institute of Health Indicators and Assessment (Institute for Health Metrics and Evaluation), Dr. Christopher Murray (Christopher Murray) estimated that globally, the number of excess death is approximately twice the officially announced number due to new coronavirus.
Excess death is often used to measure the death rate of an event or epidemic. At the time of Hurricane Maria`s attack on Puerto Rico in 2017, the official death number was 64. But Irisari and other scholars use a variety of techniques to calculate that the number of excess death after the disaster more than 1000 people, may exceed a lot. According to statistics, the number of official death is now 2975.
Over the next few weeks, the disease will inevitably continue to cause people death. death in the next week or two may have been infected a few weeks ago. Dr. Tom Frieden (Tom Frieden), former director of the Centers for Disease Control and Prevention, said that even if there were no new infections, thousands of infected people would die.
Given the uncertainty and lag of the data, why should such death data estimation work be carried out? Because error data based on undercounting has become the core of discussion and decision-making basis, and Trump are already doing so.
"We have saved thousands of lives ," he said on May 1. "Our death population is expected to be below 100,000. In this context, it seems noteworthy that the 100,000 mark has been exceeded.
Nicholas Kristof became a Times columnist since 2001. He won the Pulitzer Prize twice for his coverage of China and Darfur. Welcome to pay attention to him in Twitter and fb.
