Death Data

About This Episode

Apologies for the unscheduled break in the podcast schedule, there was a death in the family and there’s a whole lotta shit to deal with around times of passing. I don’t mean the grieving process, I mean the dirty logistics of how death is administered — the credit bureaus, Social Security, wills and probate, utility bills, car titles, even what happens to a person’s Facebook account. Where does it all go?

There is so much data orbiting each of us. And it all carries on a life of its own, propelled by some kind of digital momentum. Eventually, the reality of a person’s death catches up to the data and something happens to it, but a lot of that process is still undefined. I mean, what happens to your old Hotmail account if nobody tells Hotmail that you’re gone? For a very long time, absolutely nothing. Your shadow remains.

Now, one of the most important documents at moments like this is the death certificate. The owners of all that shadow data are all very sorry for your loss, of course, but are powerless to do anything without some verifiable data to prove that a person has died. And the very last piece of data that you will generate, which goes on that document, is your cause of death. Cause of death is actually really important data for all sorts of reasons: it’s what governments use to research and understand death statistics; it lets us identify long-term trends in societies, such as the American epidemic of diabetes or the share of deaths in police and fire departments caused by COVID; it helps us set priorities for medical research. It’s really useful data.

This data is so useful, in fact, that the World Health Organization publishes the International Statistical Classification of Diseases and Related Health Problems, or ICD for short. They’re currently on the 10th revision, or ICD-10, and ICD-11 goes live next January. It’s a huge amount of work to develop this classification system. The ICD-11, for example, has been in development for over a decade with hundreds of specialists and nearly three dozen different working groups. It contains over 80,000 different diseases, symptoms, and conditions.

ICD codes are used everywhere in our healthcare system, from medical records to insurance and Medicare to, yes, death certificates.

They are also impossibly inaccurate, at least on death certificates. Researchers have conducted studies on death certificate ICD code accuracy for decades now, and regularly find error rates close to 50% — that’s half of death certificates having some sort of error on them. It could be a minor error, such as abbreviating something in a way that makes the underlying cause of death more ambiguous; oftentimes, it’s related to swapping between a symptom and the actual cause.

For example, take a cancer patient with a weakened immune system. They develop pneumonia. They die. Did they die from cancer, or from pneumonia? It’s a bit of a coin flip as to which of those two conditions get recorded as the primary cause of death, though usually both will at least be indicated somewhere in the records.

Here’s a few more data points for you:

• A 1991 study reviewed 433 death certificates from autopsied hospital patients. Significant discrepancies were found in half of them, comparing the autopsy reports and the post-mortem medical examination documents. In 25% of them, the cause of death was wrong. In 9%, the death certificate was signed before the autopsy was performed.
• Another study of death certificates issued in Broward County, Florida in 2007 found that 48% of them contained errors.
• A 2017 study of death certificates in Vermont also found half to contain errors.

In other words, it’s pretty much always been like this.

Now, by coincidence, former soldier and Secretary of State Colin Powell also passed away in the past couple of weeks in exactly this situation. He had a blood cancer, which compromises the immune system. He also had Parkinson’s disease. He then caught COVID, despite being vaccinated, in part because we know the vaccine immune response for leukemia patients is very weak. Did he die from leukemia? Or from COVID? It didn’t take long for vaccine critics to shout about the futility of vaccines, but that’s only because Powell’s family had said he died from COVID-related complications.

Apparently, when it comes to death, the only statistic that we generally get right is how many people have died.

So why does this happen? Researchers generally cite two reasons: First, there are different types of people that fill out death certificates and there is little to no training required. This is why you’ll see different error rates between death certificates completed at hospitals versus coroner’s offices or funeral homes.

The second reason is the inconsistent paper trail. In our case, for example, the funeral home asked our family about the cause of death, we answered, and they typed it into a computer. There isn’t a conversation that happens between the funeral home and a medical care team. So sometimes, the person filling out the form has access to medical history and sometimes they don’t.

The real root cause, though, that explains both of these reasons, is that most of the time nobody really cares about this data. There are no direct consequences for inaccurate cause of death, even though systemic errors can lead to some pretty bad societal side effects. In New York City, for example, the disparity in premature heart disease between white and black people was underestimated because hospitals were incorrectly overreporting premature heart disease, and those hospitals served larger proportions of white people.

We’re in a weird moment in this pandemic, though, where some people care deeply about whether or not COVID is listed as the cause of death on a death certificate. I’ve had conversations with a couple of people who were suing their state to try to remove COVID from a family member’s death certificate, because “COVID is a fake virus” and it offended them to have it listed as cause of death. Early in the pandemic, some states pressured to remove COVID from death certificates because it would become official evidence that the virus was present in their state.

Maybe, if people continue to see the accuracy of death statistics as important, we’ll see some gradual improvement in their accuracy. Personally, I see little reason to expect this to change. So the next time you see some statistics about cause of death, take it with a grain of salt and think about it more holistically, in terms of trends.

And if you care about what’s on your death certificate when the time comes, put that in your will.