Science Friday - Errors On Death Certificates May Be Skewing Mortality Data

Episode Date: August 28, 2024

According to the Centers for Disease Control and Prevention (CDC), the maternal mortality rate in the United States is very high compared to other wealthy countries: About 22.3 maternal deaths per 100,000 live births. This is on par with China and Iran, based on UNICEF data.So why is the US maternal mortality rate so high? It may have to do with how we fill out death certificates.A study from earlier this year found that misfiling of information in death certificates may be inflating the numbers. The study authors concluded that the US maternal mortality rate was actually half of the CDC-reported rate—about 10.4 per 100,000 live births—which is in line with countries like Canada and the United Kingdom.But if death certificates can skew maternal mortality statistics by such a huge margin, what else could they be influencing? And how does our system for filling out death certificates work?To answer these questions, guest host Maggie Koerth is joined by Dr. Robert Anderson, chief of the statistical analysis and surveillance branch at National Center for Health Statistics.Transcripts for each segment will be available after the show airs on sciencefriday.com. Subscribe to this podcast. Plus, to stay updated on all things science, sign up for Science Friday's newsletters.

The world of death certificates is filled with complexities. Determining the cause of death is not always straightforward. But how can this impact the way we view health statistics in the United States? It's Wednesday, August 28th, and this is Science Friday. I'm SciFri Radio Fellow Valeria Diaz. According to the Center for Disease Control and Prevention, the maternal mortality rate in the United States is very high compared to other wealthy countries. About 22.3 maternal deaths per 100,000.

live births. This is on par with China and Iran, based on UNICEF data. So why is the U.S. maternal mortality rate so high? It may have to do with how we fill out death certificates. A study from earlier this year found that misfiling of information in debt certificates may be inflating the numbers. Here's SciFri guest host Maggie Kerth with more. Earlier this summer, researchers uncovered evidence that suggesting that high death rate is just a trick of record keeping. You know, people who fill out death certificates can check a box showing whether the person who died was pregnant, and all those deaths are counted as maternal mortalities. But the study found that the actual cause of death in those cases isn't always pregnancy-related.

With only pregnancy-related statistics counted, the study found that U.S. maternal mortality falls by half down to 10.4 per 100,000 people, which brings us back into line with countries like Canada and the United Kingdom. And this brings us to an interesting and complicated problem. Data collected on death certificates is a crucial part of how we track public health, but that data does not always cleanly represent reality. It's got some flaws that have shown up in several stories I've covered as a journalist, from maternal mortality to gun violence to even early COVID deaths. If you want to know how Americans really die, you have to know how death certificates work, who fills them out, and how this data from all over the country gets compiled into

national statistics. Joining me to talk about that today is Dr. Robert Anderson, chief of the statistical analysis and surveillance branch of the National Center for Health Statistics. Welcome to Science Friday. Thank you. So let's start from a really basic place. When somebody dies in the United States, how do we record information about that death? Well, every state in the United States requires that a death be registered with the state. Well, it's actually a little bit more than just the state.

In some cases, New York City, for example, is separate from the rest of New York State for vital registration purposes. but generally we're talking about states. At the National Center for Health Statistics here, we have an arrangement with 57 jurisdictions, 50 states, District of Columbia, New York City, and five territories. And those jurisdictions are responsible for registering the deaths. So when someone dies,

typically a funeral director takes charge of the body and is responsible for disposition of that body. and they're also usually responsible for starting the death record, the death certificate. So the funeral director provides basic personal and demographic information about the decedent, and then a physician, medical examiner or a coroner, depending on the circumstances, is then responsible for providing the cause of death information. So that information is then compiled. It's registered by the state, and then the state.

sends that information to us for the national data file. And that process of determining the cause of death, who is that? Is it always somebody with a medical background? Does that ever change as you go from local record keeping to the national compilation of all of that data? Yeah, it's not always a somebody with medical training. Usually it's a physician in the case of natural causes. In the case of injury-related deaths, or deaths that occur suddenly and unexpectedly or that may be suspicious, it's a medical examiner or a coroner. And the medical examiner is always a physician, usually a forensic pathologist with extensive training. But the coroner may be an elected person, may be a former law

enforcement officer or a lawyer in Texas. The Justice of the Peace often acts as the coroner. and they may not have medical training, although they typically contract with physicians to provide information about the cause of death to do autopsies and things like that. So how do you standardize it? You know, these death reports come in from all over the country and you have different people that have different ideas of how things should be recorded. How does that all get, you know, kind of mush together at the end? Well, vital statistics in the United States is highly standard.

And we've been doing this for a long time. You know, the vital statistics system in the United States goes back more than 100 years. And, you know, as a result, we've come up with standardized format for reporting. And so all certifiers use the same standardized format to report the cause of death. So that doesn't mean that we don't have variation from certifier to certifier. You know, we do. But at least the format in which they report is the same. same state to state. So we have this example this year of a study that found this discrepancy

in how maternal mortality statistics are getting reported. And it's resulting in us thinking that there are more of these deaths than might actually be there. And I'm curious if there are other examples of situations where you've seen something like a large scale misreporting or even just situations where there's confusion and debate over what we should be taking away from these death certificates? Well, we certainly saw some confusion in the early parts of the COVID pandemic, and you reported on that. The main reason was because a lot of the certifiers, a lot of the cause of death certifiers, didn't really understand what it was that they were looking at early in the pandemic. And so we saw increases in pneumonia deaths, Alzheimer's, Alzheimer's, Alzheimer's,

disease, cardiovascular deaths, diabetes deaths, those in particular, we saw pretty huge spikes early in the pandemic due to those causes. And, you know, in retrospect, looking back, it's almost certain that those spikes were actually COVID deaths, affecting people who had comorbidities that were risk factors for severe disease. What similarities do you see between these different types of? of misreported deaths and maternal mortality? You know, like, are there common threads when you're looking at something like what happened with COVID

and what's happening with maternal mortality? Are there things that sort of connect them and point to some systemic issue with how we treat mortality data? Well, maternal mortality is kind of a tricky issue. You know, we have studied this issue extensively at the CDC. And if you go back to pre-2003 when the death certificate was modified to add this checkbox, the studies were showing that we underestimated maternal mortality by 40, 50%.

So we were underestimating substantially. And it was thought that if we added this checkbox, that that would resolve the issue. So that checkbox was added in the 2003 standard certificate. The problem that we had was not all states implemented at the same time. So it took 15 years for all states to implement that standard certificate, the pregnancy checkbox item in particular. As we were getting to the point where we felt like we had sufficient information and all states were reporting in the same way, we also discovered that there were errors in

the checkbox. That checkbox item was prone to error and that we were seen in some cases, decedents that were checked as pregnant who weren't actually pregnant. Most of these were in the older age range. So, you know, when we discover things like this, we mean, we take it very seriously. And so we did make some modifications to our coding procedures to mitigate those errors. And then we've also instituted some followback certifiers to confirm information as well. So we are working to address these issues. We know that we probably are counting some deaths that shouldn't be counted, but we're also still missing some deaths because if no pregnancy information is reported, then we can't count it

as a maternal death. And there are maternal deaths out there. where pregnancy may exacerbate an existing condition that should be counted as maternal deaths, but it wouldn't be if we didn't know the circumstances of a pregnancy. So with COVID and with this, there's this kind of situation happening where you're simultaneously knowing that there's undercounts and there's overcounts and they're happening at the same time. And I'm curious if there's ever a way to solve that kind of problem. Or do you just kind of have to chip away at these overcounts and these undercounts little by little and sort of hope to get closer to the truth?

You know, is there any way to actually get truth? Well, there's no way to completely solve these issues. I mean, the fact is that determining the cause of death is not always straightforward. It can be very difficult, very complicated to determine what somebody died from, particularly if they die from an undiagnosed condition. Diagnosing that condition after the fact can be very difficult, even with a complete autopsy. So, you know, I do want to acknowledge that, you know, certifying the cause of death, determining the cause of death is hard, often, very hard.

There are going to be times when the cause of death just isn't a problem. parent and the certifier has to make their best judgment. And in some cases, that judgment may be wrong. It's just the nature of medicine. So there's no way to completely make it go away. That said, I mean, there are some things we can do. And with maternal mortality, I mean, we did make some changes in coding to mitigate errors. We have working with the states to do linkages between the death record and birth records and fetal death records to try to get at the outcome of the pregnancy to determine whether there was a pregnancy and we're trying to get certifiers to confirm the information that they're reporting on death certificates to try to catch errors of the source.

So we are trying to do things like that. With COVID, we tried to educate certifiers to make them aware of what the issues were, how to determine whether decedent had COVID, and then how to certify the cause of death properly. Yeah, I remember when I was reporting on gun violence in the past, we ran into some really interesting problems that kind of boiled down to, you know, like, who is doing the certification and their subjective opinion of a situation. And that came into play in particular when you're trying to determine what is an accidental gun violence death versus what is a homicide versus what is a suicide. And that seems like a particularly difficult place because there's not necessarily a single right answer. It's not necessarily something you can just educate your way

out of. And I'm curious how you guys deal with those kind of problems. Yeah. So sometimes you have a situation where homicide might be defined as death at the hands of another, regardless of intent. And a lot of medical examiners and coroners take that approach. They define, so you might have an accidental shooting, but if one person shot another person, then it would be considered a homicide. That said, if the certifier provides sufficient information about the circumstances, we can code to an accident if it's clear that the circumstances were accidental. So we take the narrative that the certifier reports as the sort of pre, we give priority to that information rather than the manner of death item because we know that sometimes there are

these discrepancies. So if even with a homicide, if it's clear that the circumstances say that it was accidental, then we would code the death as unintentional rather than as a homicide. So we can mitigate it to some extent if we have sufficient information to do that. The National Center for Injury Prevention and Control also has a system called the National Violent Death Reporting System, which collects additional information to try to deal with those issues to get at those homicides that may actually be accidental. How does our way of tracking death affect how we look at public health across the nation? You know, are there things coming out of these death tracking systems that affect how we legislate around things, how we understand things?

Yeah, I think so. I mean, I'm not really involved in the policymaking part of things. The National Center for Health Statistics is a statistical agency. We provide data and we focus on the data. specifically and on the statistical part of things. But we do know that the information that we provide, and that is part of the intent, providing the information is for policymaking

so that people can take that information and allocate resources and develop programs and policies that will improve health generally. And so, you know, it is important that our data be as accurate as possible so that it can inform programs and policies in an efficient way. Are there any other causes of death that you're watching right now or that you've been digging into just to kind of make them more accurate? You know, other places where you see problems popping up.

Another thing that we're looking at carefully or heat-related deaths, that's another one that can be very tricky. So, you know, if somebody dies from heat stroke, that's usually fairly apparent. but the heat can exacerbate pre-existing conditions as well. And so while there might not be substantial heat damage to the body directly, the heat can cause somebody with cardiovascular disease to have a heart attack or can cause other sorts of problems in people with multiple comorbidities. So one of the things we're trying to do is to educate certifiers to be aware of this

and to make it make sure that if they think heat was involved, that they report it on the death certificate so that we can count it. I'm sort of curious that, you know, when you have something like a heat death issue, you end up having these kinds of downstream political implications, because if you're counting heat death, you're often talking about climate change, and that can have implications for how people approach climate funding or how they approach research.

And I'm sort of curious about, you know, like where you are coming at this from a perspective of science and a perspective of record keeping. Do you ever think about those political implications? Not directly. I mean, obviously we're thinking about how our data are used. And we're thinking about how we can provide the best data possible. but ultimately, you know, the policymaking and the programmatic stuff is out of our hands. So we don't think about that stuff directly.

I mean, we have enough to worry about just to get the data and make sure that it's of sufficient quality and to get it out in a timely fashion. So that's really where our focus is. We don't really think too much about the other stuff other than that we know that the data have to be good. so that folks can use it for that purpose. That's great. Thank you so much. This has been really interesting, and I really appreciate you taking the time to talk to us. You're very welcome. It was fun.

That was Dr. Robert Anderson, chief of the statistical analysis and surveillance branch at the National Center for Health Statistics. And that's all the time we have for now. A lot of folks helped make this show happen this week, including Kathleen Davis, Diana Plasker, Beth Rami, Danielle Johnson.

Santiago Flores And many more On tomorrow's episode The History of Teeth From ancient fish to our pearly whites But for now I'm Sci-Fi Radio Fellow Valeria Diaz

Thanks for listening