If the patient is in hospital. Those who die in nursing homes or in their own homes are less reliably classified - and there are an awful lot of them.
Jeremy
You're absolutely right. Getting precise data is difficult and messy and requires best estimations by experts as a result. There is an article from Scientific American that addresses the issue fairly well. Non-experts that have a desire to claim either over or under reporting will find plenty to glom onto in the article. I do not and would not make any claims regarding the estimates from epidemiologists as they're educated and trained in the field and I am not.
https://www.scientificamerican.com/article/how-covid-19-deaths-are-counted1/Regarding Deaths in the Hospital...
How hard it is to accurately determine whether COVID-19 was truly that last straw depends on the situation. Most COVID-19 deaths seen at Mount Sinai Health System in New York are in people who have comorbid (or co-occurring) conditions such as coronary artery disease or kidney disease, said Dr. Mary Fowkes, the chief of autopsy services at Mount Sinai. But it’s not typically difficult to tell what killed them.
“Most of the cases are pretty straightforward,” Fowkes told Live Science. “The lungs are usually so severely involved with pathology, so they are two to three times or more the normal weight of a normal lung.” (The excess weight is due to fluid and cell detritus from damaged lung tissues.)
Regarding deaths at home...
In some cases, particularly those where someone dies at home or quickly perishes after entering an emergency room, the determination can be a little more fuzzy, said Dr. Jeremy Faust, an emergency physician at Brigham and Women’s Hospital in Massachusetts. “The challenge is knowing who died of COVID-19 versus who died with the virus that causes COVID-19,” Faust told Live Science.
All of the inconsistencies of cause-of-death reporting precede the COVID-19 pandemic, says Jay Wolfson, a professor of public health at the University of South Florida (USF). But getting good data about deaths is now extremely pressing, he told Live Science. Death certificates are often used by epidemiologists and public health officials to detect strange clusters of deaths or to link certain risk factors to certain causes of death. But because different states and localities have different rules about recording and reporting causes of death, the cumulative data is always messy.
“I think some states are reluctant to open their databases up, knowing they have validity problems or knowing the data might be misused,” Wolfson said. But public health officials need access, he said, and they need to figure out ways to dig into the data and standardize them. Wolfson and other researchers at USF are already working with state officials to see what kind of data the state can legally release, he said.
Both undercounts and overcounts of COVID-19 deaths are possible, Wolfson said, but it’s not yet clear which is more likely, or whether they might simply balance each other out. Fowkes said that based on her experience, it’s more likely that COVID-19 deaths are being missed than overcounted. That’s because New York is among several cities that show spikes in deaths at home, and these anomalous spikes could be due to untested, untreated COVID-19.