Keywords maternal death - maternal mortality - physician density - pregnancy-related mortality
Introduction
The maternal mortality rate in the United States has been on the rise, increasing
from 17.4 per 100,000 live births in 2018 to 32.9 per 100,000 live births in 2021.[1 ] The major contributors to maternal deaths—accounting for nearly 75% of cases—are
severe maternal morbidity, which includes conditions such as hemorrhage, infections,
hypertensive disorders, and complications arising from unsafe abortions.[2 ] It is estimated that 20 to 40% of maternal morbidity and mortality are preventable
with timely and appropriate interventions.[3 ]
While obstetricians and gynecologists (OBGYNs) are vital for routine maternal care,
maternal–fetal medicine (MFM) physicians offer a more advanced level of care for complicated
pregnancies, providing clinical care that can be lifesaving. As specialists trained
in managing high-risk pregnancies, MFM physicians are uniquely positioned to identify
and treat complications that can arise during pregnancy, labor, and postpartum periods.
However, access to MFM physicians is unevenly distributed across the United States,
with the majority practicing in urban areas.[4 ] Although there are approximately 140 MFM fellowship positions available each year,[5 ] most of these training programs are concentrated in suburban settings. This geographic
imbalance raises concerns, particularly for rural and underserved areas, where lower
densities of MFM physicians may contribute to higher rates of severe maternal morbidity.
A study published two decades ago suggested that increasing the number of MFM physicians
by 5 per 10,000 live births could lead to a 27% reduction in maternal mortality risk.[6 ] As maternal mortality rates continue to rise, there is an urgent need to assess
the distribution and impact of MFM physicians on maternal health outcomes. This study
aims to explore the association between MFM physician availability and pregnancy outcomes,
focusing on how their presence—or lack thereof—affects maternal mortality.
Materials and Methods
This was a cross-sectional analysis of publicly available, state-level data from the
Centers for Disease Control and Prevention, Wide-Ranging Online Data for Epidemiologic
Research (CDC WONDER) database, covering the period from January 2018 to December
2021.[7 ] The CDC WONDER includes comprehensive databases, including Natality, Underlying
Cause-of-Death, and Fetal Death databases. Data on live births were sourced from the
Natality databases, providing counts of live births occurring in the United States
(50 states and the District of Columbia) to U.S. residents.[8 ] Mortality data were retrieved from the Underlying Cause-of-Death databases, based
on U.S. resident death certificates, each listing a single underlying cause of death,
up to 20 additional multiple causes, and demographic information.[9 ] Stillbirth data were drawn from the Fetal Death databases, which include fetal deaths
at 20 weeks of gestation or more within the United States.[10 ] The analysis was limited to individuals aged 15 to 44 years, as prior studies indicate
a high false-positive rate for maternal mortality in those aged 45 years and older.[11 ] Consistent inclusion and exclusion criteria were applied across datasets. Institutional
Review Board review was deemed unnecessary, as only publicly available de-identified
data were used. This study followed the Strengthening the Reporting of Observational
Studies in Epidemiology reporting guideline.
Exposures
We obtained de-identified data on practicing MFM physicians from the American Medical
Association Masterfile for 2018 to 2021. This dataset provided yearly counts of active
MFM physicians per state. To ensure accuracy, MFM fellows in training and retired
MFM physicians were excluded, focusing exclusively on actively practicing physicians.
To calculate MFM physician density, we divided the number of active MFM physicians
in each state by the total number of live births for the corresponding year and state,
scaling the result per 100,000 live births to standardize comparisons across states
and years. States were categorized annually into three groups based on MFM density
median across states of 31 MFM physicians per 100,000 live births: (1) low density
(<30 MFM physicians per 100,000 live births), (2) moderate density (30–59 MFM physicians
per 100,000 live births), and (3) high density (≥60 MFM physicians per 100,000 live
births). These thresholds were chosen to reflect meaningful differences in access
to MFM care across states, with < 30 as low access, 30 to 59 as moderate access consistent
with the national median, and ≥60 as high access, capturing states with comparatively
high MFM physician availability.
Outcomes
The primary outcome was maternal mortality. Maternal death is defined by the World
Health Organization as “the death of a woman during pregnancy and up to 42 days postdelivery,
regardless of the pregnancy's duration or location, from any cause related to or aggravated
by the pregnancy or its management, excluding accidental or incidental causes.”[12 ] Therefore, cases with an underlying cause of death assigned to the International
Statistical Classification of Diseases, 10th Revision (ICD-10) code numbers A34, O00–O95,
and O98–O99 were included.[13 ] Secondary outcomes included pregnancy-related mortality and stillbirths. The pregnancy-related
mortality rate, which includes late maternal deaths (within 1 year postdelivery, coded
as ICD-10 O96 and O97), was examined due to elevated risks within the first postpartum
year.[14 ] Stillbirth was defined as fetal death at 20 weeks of gestation or later. All outcomes
were assigned based on the state where delivery occurred.
Data Analysis
We plotted the number of live births, maternal mortality, pregnancy-related mortality,
stillbirths, and MFM by calendar year. The annual trend was statistically examined
using Poisson regression. We integrated data from 2018 to 2021 and applied a Bayesian
spatial autoregressive model using the Geostan package in R to account for spatial
dependencies. This Bayesian spatial analysis allowed us to estimate adjusted state-wise
averages of MFM density and maternal mortality rate and visualize the results on a
state-level map.[15 ]
[16 ]
We examined individual characteristics by MFM density. To address clustering within
states, we employed multivariable negative binomial mixed effects models with Huber–White
robust estimates, calculating adjusted incident rate ratios (aIRRs) with 95% confidence
intervals (CIs) for maternal mortality, pregnancy-associated mortality, and stillbirths,
using low MFM density as the reference. The negative binomial regression model was
chosen due to overdispersion. Models were adjusted for state-specific demographics
(proportion of non-White individuals, obesity, chronic hypertension, pregestational
diabetes, and education levels below high school) from the Natality databases, as
well as state-specific poverty rates from the United States Census Bureau.[17 ] In addition, the year variable was adjusted to account for annual variation and
the spike in maternal mortality during the coronavirus disease 2019 (COVID-19) pandemic.
Lastly, we also adjusted for the density of OBGYNs and midwives per 100,000 births.
These covariates were selected based on established associations with maternal mortality
in the literature.[18 ]
[19 ]
[20 ]
[21 ]
[22 ] Additionally, to improve the interpretability, adjusted rates and average marginal
effects (AME) were calculated using the marginal standardization form of predictive
margins; we estimated the probability in each group by fixing state-level characteristics
at the average level of the categories. The AME represents the difference in adjusted
rates across states, helping to mitigate potential misinterpretation, especially when
baseline risks are low.[23 ] Furthermore, we plotted the adjusted rates of maternal mortality per 100,000 live
births against the MFM density.
Sensitivity Analyses
To ensure robustness, several sensitivity analyses were performed. First, a leave-one-out
analysis was conducted, sequentially excluding each state to examine any disproportionate
impact on results. Second, to account for the potential influence of antenatal MFM
care, analyses were repeated using the state of residence rather than the delivery
location. Third, stability across classifications was evaluated using alternative
MFM density categories (<20, 20–39, 40–59, and ≥ 60 MFM physicians per 100,000 live
births). Fourth, we calculated the E-value to estimate the minimum strength of association
required from an unmeasured confounder to explain the observed relationship between
MFM density and MMR.[24 ]
Effects were considered significant if the 95% CI did not include the null or if the
p -value was less than 0.05. All statistical analyses were conducted using Stata version
18.5 (StataCorp, College Station, Texas, United States) and R version 4.4.1 (R Foundation
for Statistical Computing, Vienna, Austria).
Results
Overall, there were 14,792,743 live births, 3,440 maternal mortalities, 4,980 pregnancy-related
mortalities, and 90,848 stillbirths during our study period from January 2018 to December
2021. The median MFM density across states was 31.6 per 100,000 live births (interquartile
range: 21.9–42.5).
[Fig. 1 ] presents the trends in the number of births, maternal outcomes, MFM physicians,
and stillbirths from 2018 to 2021 in the United States. During this period, the number
of live births significantly decreased (p < 0.001), while the number of maternal mortality, pregnancy-related mortality, and
MFM physicians increased (all p < 0.001). There was no clear trend for stillbirth (p = 0.07).
Fig. 1 Trends in births, maternal outcomes, maternal–fetal medicine physicians, and stillbirths
(2018–2021). The blue line represents the trend in the number of births, the red line
represents the trend in the number of maternal morbidity and mortality, the green
line represents the trend in the number of pregnancy-related mortality, the black
line represents the trend in the number of MFM physicians, and the orange line represents
the trend in the number of stillbirths. Each dot on the lines represents the observed
yearly value for each metric. MFM, maternal–fetal medicine; MM, maternal mortality;
PRM, pregnancy-related mortality.
[Fig. 2 ] illustrates the distribution of MFM physician densities and maternal mortality rates
across states. The Northeast exhibited the highest density of MFM physicians, with
states like Massachusetts and New York having over 50 MFM physicians per 100,000 live
births, correlating with lower maternal mortality in 100,000 live births in these
regions. In contrast, the South and Midwest showed a comparatively lower density of
MFM physicians and higher maternal mortality in 100,000 live births, highlighting
disparities in healthcare resources and outcomes across regions.
Fig. 2 Distribution of MFM physicians and maternal mortality per 100,000 live births across
states. (A ) Displays the density of MFM physicians per 100,000 live births across states. States
are color-coded on a gradient from purple (lower density) to yellow (higher density),
with yellow indicating areas with the highest density of MFM physicians. (B ) Displays maternal mortality rates (MMR) per 100,000 live births across states. States
are color-coded on a gradient from yellow (lower MMR) to green and purple (higher
MMR), with purple highlighting areas with the highest maternal mortality rates. MFM,
maternal–fetal medicine; MMR, maternal mortality rate.
Demographics are presented in [eTable 1 ] (available in the online version). Individuals in high MFM density states compared
with those in other states were more likely to be older, White, less likely to have
an education less than a high school degree, and more likely to have chronic hypertension.
Furthermore, high MFM density states were more likely to have a higher number of OBGYNs
and midwives.
Maternal mortality, pregnancy-related mortality, and stillbirth rates according to
MFM density are displayed in [Fig. 3 ]. States with high MFM density had a reduced risk of maternal mortality (aIRR: 0.70;
95% CI: 0.58–0.85) and pregnancy-related mortality (aIRR: 0.83; 95% CI: 0.71–0.98)
compared with states with low MFM density, corresponding to 7.29 (AME: 95% CI: 3.58–11.00)
and 5.57 (AME: 95% CI: 0.74–10.40) less mortality per 100,000 live births, respectively.
States with moderate MFM density had a similar risk of maternal mortality compared
with low MFM density states (aIRR: 1.02; 95% CI: 0.87–1.20). Unadjusted rates and
IRRs were consistent with the primary analysis ([eTable 2 ], available in the online version). In contrast, stillbirth rates did not differ
significantly across MFM density groups.
Fig. 3 Maternal mortality, pregnancy-related mortality, and stillbirth, stratified by MFM
physician density levels. Each dot represents the adjusted rate for a specific outcome
at a given density level. The horizontal line extending from each dot shows the 95%
confidence interval for that estimate. Models were adjusted for state-specific rates
of non-White individuals, chronic hypertension, pregestational diabetes, education
levels below high school, poverty rates, OBGYN/Midwives density, and year. AME, average
marginal effect; CI, confidence interval; IRR, incident rate ratio.
[Fig. 4 ] depicts the frequency distribution of MFM density and the association between varying
MFM density and maternal mortality. As MFM density increased, adjusted rates of maternal
mortality decreased.
Fig. 4 Frequency distribution of MFM density and its association with maternal mortality.
Gray bars represent the frequency of states by MFM physician density per 100,000 live
births. The black line represents the adjusted rate for maternal mortality across
MFM density levels, with the blue shaded area indicating the 95% confidence interval.
Models were adjusted for state-specific rates of non-White individuals, chronic hypertension,
pregestational diabetes, education levels below high school, poverty rates, OBGYN/Midwives
density, and year. MFM, maternal–fetal medicine.
Sensitivity analyses supported the primary findings. In the leave-one-out analysis,
the exclusion of any single state did not affect the observed findings ([eFig. 1 ], available in the online version). Analyses based on the state of residence also
produced results consistent with the primary analysis ([eFig. 2 ], available in the online version). Alternative threshold analyses showed similar
patterns, except for the loss of statistical significance in the association between
MFM density and pregnancy-related mortality ([eFig. 3 ], available in the online version). Lastly, the E-value for maternal mortality risk
in high MFM density states was 2.21, suggesting that any unmeasured confounder would
need to have a risk increase of over twofold to fully account for the observed association
between high MFM density and reduced maternal mortality.
Discussion
In this cross-sectional study using national U.S. databases, we found that states
with high MFM density had a significantly lower risk of maternal mortality compared
with states with low MFM density. In contrast, stillbirth rates showed no significant
association with MFM density. Our analysis also highlighted a geographically uneven
distribution of MFM physicians, corresponding with disparities in maternal mortality
across regions. We found that an MFM density of 60 or more per 100,000 live births
is associated with a decreased risk of maternal mortality compared with lower densities.
These results remained consistent across multiple sensitivity analyses, underscoring
the robustness of our findings.
Nearly two decades ago, a study reported a maternal mortality rate of 7.5 per 100,000
live births, estimating that an additional 50 MFM physicians per 100,000 live births
could reduce maternal mortality by 27%.[6 ] By comparison, the current maternal mortality rate has risen sharply to 32.9 per
100,000 live births in 2021—over four times higher than previously reported.[18 ] While this increase could partly be explained by changes in maternal mortality surveillance
methodologies, such as the introduction of the “pregnancy checkbox,”[13 ] other factors, including rising maternal age, higher obesity rates, and a growing
prevalence of preexisting medical conditions, likely play a role.[25 ]
[26 ] Notably, our study did not use the “pregnancy checkbox” for mortality classification
to reduce potential misclassification bias.
This study highlights the critical role of MFM density in improving maternal health
outcomes. Our findings underscore the importance of ensuring adequate distribution
of MFM physicians, particularly in regions with elevated maternal mortality rates.
Addressing these disparities by allocating MFM resources more evenly across states
could improve maternal outcomes, especially in underserved regions with high maternal
mortality rates, such as the South and Midwest. Additionally, increased support for
MFM training programs in underserved regions, alongside incentives to retain physicians
in these areas, could serve as practical steps toward improving access to specialized
maternal care. This focus on equitable access to MFM resources aligns with broader
public health goals of reducing maternal mortality and achieving health equity across
the United States.
MFM physicians can play a critical role in reducing MMR by providing prevention and
acute treatment. MFM physicians specialize in the early identification and management
of conditions such as hypertension, diabetes, and maternal cardiac diseases, which
are closely linked to maternal mortality.[27 ] By recognizing these risks early, MFMs could implement preventive measures that
reduce the likelihood of severe complications.
In addition, MFMs facilitate multidisciplinary management, working alongside specialists
such as cardiologists, anesthesiologists, and neonatologists to ensure comprehensive
care. MFMs also play a pivotal role in obstetric emergencies such as hemorrhage, severe
hypertension, and thromboembolism.[28 ] Furthermore, MFM physicians help develop clinical care protocols for high-risk patients
as well as participate in quality improvement processes, which ultimately improve
maternal care standards and outcomes. In underserved areas, they may provide training
and consultative support to general obstetricians, indirectly raising the level of
care even where MFM density is lower.
Our findings suggest significant research implications. While we observed an association
between high MFM density and reduced maternal mortality rate, the specific mechanisms
driving this relationship remain unclear. Future studies should explore whether the
observed benefits stem from increased access to specialized prenatal care, enhanced
management of high-risk pregnancies, or improved interdisciplinary collaboration.
Recent trends indicate that one-third of OBGYNs and MFMs relocated within a decade,
often to more urban, affluent areas.[4 ]
[29 ] Addressing MFM maldistribution may, therefore, require policy changes or economic
incentives, warranting further investigation into strategies for increasing MFM densities
in underserved areas.
Our study has several strengths. First, it utilized large, national datasets, offering
a broad perspective on MFM distribution and maternal outcomes across diverse U.S.
regions, enhancing the generalizability of our results. Second, by examining geographic
disparities, the study reveals regional patterns in MFM density and MMR, highlighting
areas of healthcare inequity. Third, multiple sensitivity analyses were conducted
to ensure consistency across various thresholds and exclusions, reinforcing the robustness
of our findings. Lastly, the use of government-provided datasets enhanced data accuracy,
reliability, and replicability, enabling transparent analysis on a representative
scale.
Our study is not without limitations. As a cross-sectional analysis, it limits causal
inferences regarding the relationship between MFM density and maternal mortality.
Although robust, our analysis may still be influenced by unmeasured confounders, such
as healthcare infrastructure and socioeconomic factors. However, given that the E-value
exceeded 2.0 for maternal mortality, it is unlikely that unmeasured confounders alone
would account for the observed association. Finally, despite efforts to minimize misclassification
by restricting maternal age to 15 to 45 years and excluding the “pregnancy checkbox,”
some misclassification risk persists.[11 ]
[13 ] Finally, we were not able to determine the type of individual MFM practices (e.g.,
solely consultative vs. assuming primary patient care). This lack of detail limits
our ability to differentiate MFM roles, which is essential for understanding the direct
impact of MFM density on maternal outcomes.
In conclusion, we found a significant association between high MFM density and decreased
maternal mortality, highlighting disparities in MFM distribution across the United
States. These findings underscore the potential benefits of optimizing MFM density
to improve maternal health outcomes, particularly in underserved regions. This study
contributes valuable insights into how access to specialized maternal care impacts
health disparities.
