Geographic variation in pregnancy associated overdose and substance use disorder mortality, 2016 to 2022

This national study reveals that pregnancy-associated overdose and substance use disorder mortality follow a distinct geographic pattern from all-cause maternal mortality, characterized by a convergence of overdose rates across rural and urban areas during the pandemic, a dominance of substance use disorder deaths over acute overdoses, and a concentration of fatalities in the late postpartum period, highlighting the need for integrated addiction care and extended postpartum support.

The Gist

The Big Picture: Two Different Crises Colliding

Imagine the United States is facing two massive storms at the same time.

1. Storm A (The Drug Crisis): A wave of dangerous drugs (like fentanyl) is sweeping across the country.
2. Storm B (The Maternal Health Crisis): Pregnant women and new mothers are dying at alarming rates, often because they can't get good medical care.

Usually, these two storms hit different places. Storm B has always been worst in rural areas (small towns and farms) because those places have fewer hospitals and doctors. Storm A used to be worst in cities, but then it started hitting small towns hard too.

This study asks a simple question: When these two storms hit a pregnant woman at the same time, does she die in the same places where she usually dies from other pregnancy complications?

The answer is a surprising "No." The geography of drug-related deaths is playing by a completely different set of rules.

---

Key Finding #1: The "Rural Disadvantage" Flipped Over

For decades, if you lived in a small town, your risk of dying from pregnancy complications was much higher than if you lived in a big city. It was like a "rural penalty."

But for drug overdoses, that penalty flipped during the pandemic (2020–2022).

• The City: Cities actually did better than expected. They had more resources (like naloxone kits and addiction clinics) to stop overdoses.
• The Countryside: Small towns saw a massive spike. It's as if the "drug storm" suddenly decided to target the places that were least prepared to handle it.

The Analogy: Imagine a flood. Usually, the low-lying rural valleys get the worst flooding because they lack levees. But in this specific case, the "flood" of drugs suddenly surged into the rural valleys, while the cities built temporary dams and held back the water. The gap between city and country closed, but not because the cities got worse—it's because the rural areas got hit much harder than anyone predicted.

Key Finding #2: The "Slow Burn" vs. The "Flash Fire"

The study looked at two types of drug-related deaths:

1. Overdose (OD): A "Flash Fire." This is an immediate, acute poisoning event.
2. Substance Use Disorder (SUD): A "Slow Burn." This includes deaths from long-term health damage caused by addiction (like heart failure, infections, or organ damage) that happen over months or years.

The Shocking Stat: For every one pregnant woman who died from an immediate overdose, more than two died from the long-term effects of addiction.

The Analogy: Think of a house fire.

• Overdose is like the house exploding instantly. It's loud, scary, and happens right away.
• SUD Mortality is like the house slowly burning down from the inside over a year. The smoke detectors (doctors) aren't there to warn you because the fire is smoldering in the walls.

The study found that most of these "slow burn" deaths happened long after the baby was born (between 43 days and one year postpartum). This is the "danger zone" where the mother is no longer in the hospital, insurance might have lapsed, and no one is checking on her.

Key Finding #3: The "Suburban Sweet Spot"

You might think drug deaths are highest in the deep countryside or the big cities. But the data showed something weird: The highest rates of long-term addiction deaths were in "Metro-Adjacent" counties.

The Analogy: Think of a ripple in a pond.

• The City is the center.
• The Deep Countryside is the far edge.
• The Metro-Adjacent areas are the "beach houses" right next to the city.

The study found that these "beach house" areas had the highest rates of addiction deaths. It suggests that drug markets are spreading out from the cities into the suburbs and exurbs, creating a "hotspot" that isn't quite rural and isn't quite urban.

Why This Matters (The "So What?")

The authors argue that we are trying to fix this problem with the wrong tools.

• The Old Way: We treat pregnancy deaths as a "hospital problem" (we need more OB-GYNs).
• The New Reality: Drug-related deaths are a "supply and safety problem." They are driven by where the drugs are, where the harm reduction (like naloxone) is, and where the insurance coverage stops.

The Solution:

1. Keep the Lights On: We need to keep health insurance for new moms for a full year, not just 60 days, because that's when the "slow burn" deaths happen.
2. Bring Help to the Countryside: We can't just wait for rural women to come to the city for help. We need mobile clinics and community programs to bring addiction treatment and safety tools to small towns.
3. Stop Punishing, Start Helping: In many places, pregnant women are afraid to seek help because they fear getting arrested or losing their kids. The study suggests this fear is driving them underground, making them more likely to die.

The Bottom Line

Drug-related deaths among pregnant women aren't following the same map as other pregnancy deaths. They are shaped by drug supply lines and where help is available, not just where the hospitals are. To save lives, we need to stop looking at this as just a "medical" issue and start treating it as a complex mix of addiction, geography, and policy.

Technical Summary

1. Problem Statement

The study addresses a critical gap in understanding the intersection of two public health crises in the United States: the escalating drug overdose epidemic and the maternal mortality crisis. While all-cause maternal mortality is known to be persistently higher in rural areas due to obstetric care deserts and provider shortages, it remains unclear whether drug-related pregnancy-associated mortality follows the same "rural disadvantage" pattern or if it is restructured by the geographic dynamics of drug markets (e.g., fentanyl diffusion, harm reduction infrastructure).

Furthermore, existing surveillance often relies on narrow "pregnancy-related" definitions (deaths causally linked to pregnancy) or acute overdose codes, potentially missing the broader burden of Substance Use Disorder (SUD) mortality and deaths occurring in the late postpartum period (43 days to 1 year post-delivery).

2. Methodology

Study Design & Data Source:

• Type: National population-based surveillance study.
• Data: Individual-level mortality and natality records from the National Vital Statistics System (NVSS) for 2016–2022.
• Population: 25,007,723 live births.
• Definition of Pregnancy-Associated (PA) Death: Adopted the CDC's broader definition (death during pregnancy or within one year of pregnancy end) using the NVSS 2018 algorithm. This includes deaths where the pregnancy checkbox was marked or specific ICD-10 codes (O00–O99) were present, capturing accidental/external causes like overdoses that narrower definitions exclude.

Exposures & Classifications:

• Rural-Urban Continuum: Counties were stratified into three groups using USDA Rural-Urban Continuum Codes (RUCC):
  1. Metropolitan (RUCC 1–3).
  2. Metro-Adjacent (RUCC 4, 6, 8).
  3. Non-Metro-Adjacent (RUCC 5, 7, 9; most isolated).
• Geographic Regions: Categorized by US Census regions (Midwest, Northeast, South, West).
• Time Periods: Pre-COVID (2016–2019) vs. Post-COVID (2020–2022).

Outcomes:
Two non-mutually exclusive outcomes were identified via ICD-10 coding:

1. Overdose (OD) Mortality: Identified by drug poisoning T-codes (T40.0–T40.9, etc.), representing acute poisoning events.
2. Substance Use Disorder (SUD) Mortality: Identified by mental/behavioral disorder codes (F11–F16, F18–F19) and obstetric code O99.32, representing underlying chronic disorder.

Statistical Analysis:

• Rates: Calculated per 100,000 live births with 95% confidence intervals (Poisson assumption).
• Excess Mortality Modeling: A Bayesian hierarchical Poisson model (using MCMC in R/NIMBLE) was employed to estimate excess OD mortality.
  • Counterfactual: A first-order random walk (RW1) temporal process was fitted to pre-COVID data (adjusting for race/age) to project expected trends into the post-COVID era.
  • Excess: Defined as observed counts minus posterior-mean expected counts. Credible intervals excluding zero indicated statistically significant deviations from pre-pandemic trajectories.

3. Key Contributions

• Distinct Geographic Logic: The study demonstrates that drug-related maternal mortality does not follow the traditional rural disadvantage pattern seen in all-cause maternal mortality. Instead, it follows a logic shaped by drug supply dynamics and harm reduction geography.
• SUD vs. OD Distinction: It highlights that SUD mortality (chronic disorder) significantly outweighs acute OD mortality (2:1 ratio), suggesting that surveillance relying solely on poisoning codes severely underestimates the burden.
• Temporal Convergence: It identifies a unique "convergence" phenomenon during the COVID era where rural overdose rates accelerated while urban rates decelerated, narrowing the rural-urban gap.
• Late Postpartum Vulnerability: It pinpoints the "late postpartum" window (43 days to 1 year) as the period of highest mortality concentration, revealing a surveillance gap in current obstetric care systems which typically end at 6 weeks.

4. Key Results

Overall Burden:

• National Rates: 2.06 OD deaths and 4.32 SUD deaths per 100,000 live births.
• Ratio: SUD deaths exceeded OD deaths by more than 2-to-1 across all strata.
• Timing: Both outcomes were heavily concentrated in the late postpartum period (43 days to 1 year).

Geographic Variation (Rural-Urban Gradient):

• Overdose (OD):
  • Pre-COVID: Rural areas had near-zero rates; urban areas had moderate, rising rates.
  • Post-COVID (2020–2022): A convergence occurred. Rural rates rose substantially, while urban rates fell below pre-pandemic projections.
  • Excess Mortality: Credible excess OD mortality was found only in non-metropolitan counties in the South and Northeast. The Midwest and West non-metro areas showed no credible excess.
• Substance Use Disorder (SUD):
  • Rates were non-monotonic. Metro-adjacent counties had the highest rates (7.08/100k), followed by non-metro-adjacent (4.59/100k) and metropolitan (4.01/100k).
  • This suggests suburban/exurban drug market dynamics play a critical role, inconsistent with a simple "access to care" gradient.

Regional Heterogeneity:

• Northeast: Showed the most extreme single-year excess in non-metro areas (though based on small numbers).
• South: Showed a clear pattern of convergence driven by fentanyl market expansion into areas lacking harm reduction infrastructure.
• Urban Areas: Across all regions, metropolitan areas performed better than their pre-pandemic trajectories, likely due to concentrated harm reduction infrastructure (naloxone, MOUD, syringe services).

5. Significance and Implications

Clinical & Policy Implications:

• Surveillance Reform: The study argues for shifting from "pregnancy-related" to "pregnancy-associated" definitions to capture the true burden of drug-related maternal death. It calls for standardized SUD ascertainment and mandatory autopsies for pregnancy-associated deaths.
• Extended Coverage: The concentration of deaths in the late postpartum year supports the policy of extending Medicaid coverage to 12 months postpartum (as authorized by the American Rescue Plan Act) to ensure continuity of SUD treatment.
• Integrated Care: The findings necessitate the integration of obstetric care and addiction treatment (e.g., co-locating MOUD prescribing and naloxone distribution in prenatal settings) and extending harm reduction services (fentanyl test strips, mobile units) to rural and metro-adjacent areas.
• Criminalization: The authors note that punitive legal environments (criminal prosecution for drug use in pregnancy) likely drive users underground, exacerbating mortality risks, particularly in rural areas with fewer alternative providers.

Conclusion:
Drug-related pregnancy-associated mortality is a distinct crisis driven by drug supply dynamics and fragmented care systems rather than just obstetric infrastructure. The convergence of rural and urban overdose rates, the dominance of chronic SUD over acute overdose, and the late postpartum timing of deaths reveal a systemic failure to protect pregnant and postpartum people, requiring a unified approach to maternity care and substance use treatment.