Medical Examiner

Are We Calculating Maternal Mortality Correctly?

Our narrow focus ignores mental health, substance abuse, and other issues that shouldn’t necessarily be separated from pregnancy.

A hospital bed is seen in a patient room under a shining light. To its right side is a cabinet with a monitor on top.
Thomas Northcut/Getty Image Plus

I begin my shift on labor and delivery with two patients. Patient A has been on oxytocin to make her contractions stronger after her water broke without labor. She has been on this medication for almost 12 hours and has been making slow progress. Her heart rate and temperature are trending up; she is probably on her way to getting an infection. When I click to enter her chart, the electronic medical record throws up a window, a blinding orange and red. This patient has a high hemorrhage risk, it says. Would you like to place the following recommended orders for this patient? They are good suggestions; I click yes. The window turns a soothing yellow and disappears.

Patient B came in for an induction of labor. She was recently started on oxytocin, but she is already in active labor. No fibroids, no fever—she’ll probably have a baby in the next few hours. This patient’s chart looks medically uninteresting, until I click on the “social” tab. There I find that in her first trimester, this patient attempted suicide via a Tylenol overdose. She was found by her boyfriend before she had taken very much; she was brought to the emergency department. She was discharged from the psych ward after three days of observation. The pregnancy continued; she came to most of her appointments. She was seen by the clinic social worker during the pregnancy. She refused referral to a psychiatrist to discuss antidepressant therapy. She told the social worker that she has been feeling better; she has always denied further suicidal impulses.

When she was admitted to the hospital that day, Patient B denied signs or symptoms of depression. She is planning on moving out of state right after she delivers, and has declined postpartum follow-up with an obstetrician or a mental health provider. No window pops up here while I’m in her chart; everything stays a nice reassuring blue.

Which of these patients is in more danger? I thought I knew.

Maternal mortality is big news in this country. Women are dying in pregnancy, more than ever, and shamefully, undeniably, women of color more than anyone else. High-profile op-eds talk about this issue, Democratic candidates offer platforms to combat pregnancy-related death, and Congress was even moved to take action. The large leadership organizations for OB-GYNs are rolling out campaigns and action items to address this crisis.

When we talk about maternal mortality in a high-resource country like the United States, we generally talk about the same list of causes over and over again: hypertensive diseases of pregnancy (such as preeclampsia), embolic disease (that is, clots), infection, bleeding, heart disease. These are the causes that we focus on during prenatal care, and during labor and delivery hospital admissions. I want my patients to be safe. I pay very close attention to fevers, I have new tools for quantifying blood loss, and I review the list of medications to give for high blood pressure emergencies with my trainees until they can recite them in their sleep.

But to really understand maternal mortality, I think we have to realize that when public health institutions count pregnancy-related deaths, they first make some choices. These choices are aimed toward giving us the most helpful set of data. To do so, they define their terms to exclude deaths that are not directly related to the pregnancy. On first glance, that makes sense: They are, after all, in the business of pregnancy-related health care.

But when you look at the overall deaths for women in and around pregnancy, without regard to whether the cause is traditionally considered “pregnancy-related,” the number of dead women goes up a lot. And the list of causes of death becomes a very different one: motor vehicle collisions, homicide, suicide, substance abuse. This makes some sense—these are the things that young women die from, more generally, pregnant or not.

So perhaps this is obvious. Of course, women die from what women die from, and some of those women are pregnant. But see this chart, which is from research about women who died in and after pregnancy in Illinois:

A chart depicting mortality rate per 100,000 live births.
Data from an article by Abigail Koch. Chart by Slate.

Illinois is an extreme example, and these numbers vary, from state to state and from year to year. But to me this chart is shocking, because I used to spend almost all of my professional time thinking about those blue bars, and almost none thinking about those red bars.

When I did sit down to think about those red bars, I ended up with this: I don’t think we can be entirely sure that there is such a big difference between the red bars and the blue bars. How many of those not-pregnancy-related deaths are really, truly, absolutely not affected by pregnancy?

Let’s take, for example, the most obvious connection: pregnancy-related depression (or postpartum depression). Mood disorders during and after pregnancy are real and incredibly common. Why wouldn’t we think about suicide risk, which in the chart above kills 2 to 3 out of every 100,000 women during or after pregnancy, as being related to their pregnancy?

There are other connections to draw. Pregnant and postpartum women are also at higher risk for intimate partner violence. This probably contributes to that long red bar of homicide risk, but it’s hard to know by how much, because this is data that’s not routinely collected in every state, or studied.

And what about those motor vehicle collisions? Not entirely unrelated, either. Women who get in a serious collision during pregnancy are more than twice as likely to die, to some extent because being pregnant means we have more places to be injured, but also (and more frustratingly) because “the performance standards for vehicle safety designs are based on the stature and anatomy of average male drivers,” according to the American Journal of Lifestyle Medicine. (That’s a major problem for the nonpregnant women amongst us as well, but I digress.)

I know that we have to chop up death when we study it. I know that’s how epidemiology works—it’s how we quantify and analyze and figure out where death is coming from. It is how, ultimately, we increase our understanding and make progress on beating it back. Chopping up the data provides us with a way to attempt to sort out which deaths of pregnant women are tragic and deserve resources and which ones we will just live with.

The way we chop them up is also changing: The CDC has added indicators for homicide, suicide, and mental health issues to its maternal mortality review data system that the agency has rolled out nationwide. It’s on track to being used in close to 40 jurisdictions by the end of this year.

But for me, a doctor in the field, my job isn’t just to stop pregnancy-related disease and suffering and death. It is to stop disease and suffering and death for my patients, period. If I’m going to do that job, I will need to take a more comprehensive view of the dangers my patients face. And I need to understand how the assumptions we’re making when we analyze the data might end up sorting some deaths that are really in the first category into the second category.

Once I found this larger way of looking at maternal mortality, it changed the way I practice and the way I teach. For example, at one of my clinics, we had a policy of universal perinatal depression screening. But our screening rate was low—depression screening was always the first task to get lost when the clinic got busy or when we were running late. And that seemed to make sense: We had to check blood pressure, and weight, and fetal heart tones, because we thought that those were more important. We’d definitely catch that depression screening next time, or the time after that, we’d say.

This isn’t the way I feel about perinatal depression screening now. And so, one day during a staff meeting, I pulled out this graph. I showed the clinic staff—the doctors, but also the nurses and the secretaries—that the women we served were more likely to die of suicide during and after pregnancy than of all elevated blood pressures we would catch. This didn’t mean we would stop checking blood pressure, but it sure as hell meant we needed to check for depression. And that information put new urgency and consistency into our depression screening. Our screening rate went up dramatically the following month.

Maternal mortality is a crisis. But it’s a more complex one than we currently understand. When we don’t widen our understanding of what these women truly face, we narrow, too, our ability to solve it.

Patient A went on to deliver; she did bleed, but our team was ready with medications and other techniques. We stopped her hemorrhage as quickly as we could. She was shaky and pale after delivery and did end up needing a transfusion. She stayed an extra day in the hospital to receive it and recover. She felt much better on the day of her hospital discharge. We had solved her problem. She was no longer in danger.

Patient B had a completely uncomplicated delivery. She declined to talk with our social worker on the postpartum unit. She was discharged home with her baby 48 hours later. Patient B never came to her postpartum checkup. I don’t know where she is today. I just know that we did not come anywhere close to solving her problem—and I know that she left my care in more danger than she should have.