Risk Screening in a Medicaid-Managed Pregnancy Medical Home: The Need to Center Maternal Health Outcomes in Public Health Programming

Imagine the healthcare system as a giant, high-tech lighthouse. Its job is to spot ships in trouble before they crash into the rocks. For decades, this lighthouse has been very good at spotting one specific kind of danger: the "baby ship." If a baby is going to be born too early or too small, the lighthouse flashes a bright warning, and a team of helpers rushes out to guide the ship to safety. This system is called "risk screening," and it's a standard tool in many places, including North Carolina's Medicaid program.

But here's the twist: what about the "mother ship"? The person carrying the baby is also a vessel, and sometimes they face their own storms—like severe bleeding, infections, or needing to stay in the hospital longer than expected. These are called "adverse maternal outcomes." For a long time, the lighthouse was built to look at the baby's path, assuming that if the baby was safe, the mother must be okay too. But scientists have started to wonder: Is the lighthouse actually looking at the right storms? Does the same warning system that spots a baby in trouble also spot the mother in trouble? This paper dives into that exact question, using a massive amount of data to see if the current "safety net" catches both passengers or just one.

The Big Test: Does the Baby Radar Catch the Mom?

North Carolina has a long-running program called the "Pregnancy Medical Home." Think of it as a super-organized care team for pregnant people on Medicaid. When a person joins this program, they fill out a "risk screen"—basically a long checklist of questions about their health, their history, and their life. Does the mom have diabetes? Is she stressed about housing? Has she had a baby before? Is she smoking?

The goal of this checklist was originally to find pregnancies at high risk of having a baby born too early (preterm) or too small (low birth weight). If the checklist flagged someone as "high risk," they got extra help, like a care manager to make sure they kept their doctor appointments and got the support they needed. And guess what? It worked! The program successfully helped reduce the number of early and small babies.

But the researchers behind this paper asked a curious question: Does this same checklist also spot the moms who are at risk of getting seriously sick themselves?

To find out, the team acted like digital detectives. They linked together three huge piles of data from 2014 to 2019:

1. The risk screen checklists (the questions the moms answered).
2. Hospital records (what actually happened when the baby was born).
3. Medicaid claims (the paperwork showing who got paid for what).

They looked at 205,916 births. That's a lot of ships! They wanted to see if the people who checked "yes" to risky things on their form were the same people who ended up having a "bad day" at the hospital. A "bad day" in this study meant a serious event like:

• Severe maternal morbidity (SMM): A near-miss event where the mom almost died (like a heart attack or severe bleeding).
• ICU admission: The mom had to go to the Intensive Care Unit.
• Staying in the hospital way too long or having to come back for readmission.

The Results: A Foggy Forecast

Here is where the story gets interesting. The researchers ran the numbers through a fancy computer model (called a "random forest," which is like a team of decision-making trees) to see if the checklist could predict these scary events.

The findings were a bit of a letdown for the system's designers. While the checklist was great at spotting risks for the baby, it was poor at predicting risks for the mom.

• The Score: The model's ability to tell the difference between a "safe mom" and a "sick mom" was measured by a score called the "Area Under the Curve" (AUC). A perfect score is 1.0, and a coin flip is 0.5. This model scored 0.63. That's barely better than guessing.
• The Miss Rate: When the model tried to be as accurate as possible, it only caught 56% of the moms who actually had a bad outcome. That means it missed almost half of them! It also incorrectly flagged many healthy moms as "at risk" (specificity was only 63%).

What Did the Checklist Get Right (and Wrong)?

The study did find that certain things were linked to moms getting sick. If a mom was older than 25, Black, overweight or obese, smoked, or had chronic diseases like diabetes or high blood pressure, she was indeed at higher risk. The checklist also picked up on things like a history of preterm birth or high blood pressure in a previous pregnancy.

However, when the researchers put all these clues together into a single prediction tool, the tool just didn't work well enough to be useful for moms. It's like having a weather app that is amazing at predicting rain for your garden but terrible at predicting if your car will break down.

Why Did the Lighthouse Miss the Mom?

The authors suggest a few reasons why the "baby radar" failed to catch the "mom storms":

1. Wrong Questions: The checklist was designed to find baby problems. It might be missing the specific clues that signal a mom is about to get sick. For example, it might not ask the right questions about mental health or specific types of pain that lead to ICU stays.
2. Timing Issues: Some problems, like severe high blood pressure (preeclampsia), might not show up until after the checklist is filled out. If you fill out the form at your first visit, you can't predict a problem that starts three months later.
3. The "Fix" Might Have Helped: The moms who were flagged as "high risk" got extra care. It's possible that this extra help actually prevented some bad outcomes, making it look like the checklist wasn't working because the people it flagged didn't get sick as often as they otherwise would have.

The Bottom Line

The paper concludes that while North Carolina's program is a fantastic model for keeping babies safe, it is not currently a good tool for spotting moms who are in danger. The system was built to optimize neonatal (baby) outcomes, and it did that job well. But to protect the mothers, we need a new kind of map.

The authors aren't saying the program is broken; they are saying it needs to be rebalanced. They argue that public health programs need to stop looking at the baby and the mom as separate ships. Instead, they need to design risk screens that look at both passengers equally. If we want to save lives, we need to make sure our lighthouses are bright enough to see the storms coming for the whole crew, not just the cargo.