Defining Prenatal Care Surveillance Metrics Using Electronic Health Record Data

Sarah Conderino; Renata E Howland; Lorna E Thorpe; Justin S Brandt; Chuan Hong; Andrew Fair; Erinn M Hade

doi:10.1001/jamahealthforum.2026.1295

. 2026 Jun 5;7(6):e261295. doi: 10.1001/jamahealthforum.2026.1295

Defining Prenatal Care Surveillance Metrics Using Electronic Health Record Data

Sarah Conderino ^1,^✉, Renata E Howland ¹, Lorna E Thorpe ¹, Justin S Brandt ^2,³, Chuan Hong ¹, Andrew Fair ¹, Erinn M Hade ¹

PMCID: PMC13241944 PMID: 42247225

This cohort study assesses whether electronic health record data can be used to provide population-level surveillance metrics of early and adequate prenatal care.

Key Points

Question

Can electronic health record (EHR) data be used to monitor population-level metrics of early and adequate prenatal care use?

Findings

In this cohort study of 1 963 496 US patients with a live birth and prenatal care noted in the EHR, after adjusting for nonrepresentativeness of the patient population, early care initiation was underestimated in EHR network vs US natality data (68.0% vs 76.1%), but an equivalent estimate of adequacy of prenatal care was found (76.0% vs 75.2%).

Meaning

These findings suggest that near real-time availability of EHR network data has the potential to improve the timeliness of population-level pregnancy surveillance.

Abstract

Importance

Current pregnancy surveillance efforts in the US face substantial challenges in providing timely and accurate data on prenatal care use. Electronic health record (EHR) networks have the potential to enhance existing surveillance systems by providing near real-time, clinically documented data.

Objective

To assess whether EHR network data could be used to define valid and reliable surveillance metrics of prenatal care use.

Design, Setting, and Participants

This longitudinal cohort study included US adults (age ≥18 years) who received prenatal care and delivered a live birth from January 1, 2023, to December 31, 2024, at a facility that used the Epic Cosmos EHR network.

Exposure

Live birth at a facility that used the selected EHR network.

Main Outcomes and Measures

Prenatal care use was calculated as the proportions of patients who initiated care by the 13th week of pregnancy (early care) and who received adequate or better prenatal care (adequate care). Raking weights were applied to adjust the EHR sample to match the marginal distributions for US residents with live births by age, race and ethnicity, insurance, pregnancy risk factors, and geographic region. Electronic health records–based metrics were externally validated against published natality data estimates from National Center for Health Statistics (NCHS) using the two 1-sided test of equivalence. Patterns by demographics, state, and year were examined.

Results

In total, 1 963 496 patients (mean [SD] age, 29.5 [5.7] years; 100% women) had a live birth and evidence of prenatal care at a facility using the selected EHR network during the study period. Compared with all US birthing people (n = 7 224 951), patients who gave birth at a facility using the selected EHR network had lower Medicaid coverage (40.5% vs 21.1%) and a higher prevalence of pregnancy risk factors (eg, prior preterm birth: 4.0% vs 8.8%). After weighting to the national population, EHR-based estimates of early care were consistently lower than those from NCHS data (68.0% [95% CI, 67.9%-68.2%] vs 76.1% [95% CI, 76.1%-76.1%]). However, adequacy estimates were equivalent to NCHS-based estimates (76.0% [95% CI, 75.9%-76.2%] vs 75.2% [95% CI, 75.1%-75.2%]; P < .001 at 0.01 equivalence bound), aligned with expected demographic patterns, and were stable across place and time.

Conclusions and Relevance

In this cohort study, EHR network data reliably informed surveillance of prenatal care adequacy after adjusting for nonrepresentativeness of the patient population. These findings suggest that near real-time availability of EHR data has the potential to improve the timeliness of population-level pregnancy surveillance to better inform policy, public health, and clinical efforts aimed at enhancing prenatal care access and use among individuals receiving inadequate care.

Introduction

Inadequate, low-quality prenatal care is associated with increased risks of adverse maternal and infant outcomes, including maternal hospitalization and preterm birth.^1,2,3 Despite significant public health efforts, the proportion of birthing persons receiving early and adequate care remains below Healthy People 2030 national targets.^4,5 Moreover, disparities in adequacy of care by demographic factors have persisted over time.⁶ Robust pregnancy surveillance systems are critical to monitor and address these gaps in prenatal care.^5,7

However, current pregnancy surveillance efforts in the US face substantial challenges in evaluating timely prenatal care programs, policies, and clinical guidelines, such as the 2025 American College of Obstetricians and Gynecologists guideline to provide individually tailored prenatal care services.⁸ For example, birth and fetal death data from the National Center for Health Statistics (NCHS), the primary source of US pregnancy surveillance metrics, are released with a 2-year delay and contain limited information on prenatal care use, much of which is poor quality.^9,10,11,12 The Pregnancy Risk Assessment Monitoring System supplements these birth data with detailed survey data on a sample of birthing people; however, this system was suspended by the US federal government in 2025.^13,14 These limitations threaten the capacity of national surveillance systems and underscore the need for novel data sources to supplement or enhance US pregnancy monitoring efforts.

Electronic health record (EHR) data provide a promising opportunity to improve pregnancy-related surveillance.¹⁵ Electronic health records capture detailed clinical and administrative information in near real time across various aspects of patient care throughout pregnancy. Despite these advantages, EHR data have biases that may restrict their utility for surveillance,¹⁶ including fragmentation of care across institutions and nonrepresentativeness of EHR samples compared with broader target populations.^17,18,19 Emerging large-scale EHR networks that aggregate data from multiple institutions have the potential to alleviate some of these limitations. However, few studies have leveraged such EHR network data to characterize prenatal care use at a population level, and it remains unclear whether these data can reliably track use over time.^20,21,22

The goal of this study was to assess whether EHR network data could produce reliable and valid pregnancy surveillance metrics. We replicated 2 established metrics on early initiation and prenatal care adequacy using data from one of the largest US EHR databases that has been used to characterize severe maternal morbidity rates in the US.^23,24 We then externally validated these estimates compared with published estimates from NCHS natality data.^4,25 Finally, we summarized practical recommendations for developing prenatal care surveillance metrics using large-scale EHR network databases.

Methods

EHR Data and Study Population

Electronic health records data used in this cohort study came from Cosmos (Epic Systems Corporation), a dataset from a community of health systems using Epic that represented more than 300 million patients from more than 1800 hospitals and 41 000 clinics as of October 2025.²³ The community represents patients from all 50 states and the District of Columbia. The EHR data contain patient demographics, clinical information documented during pregnancies and deliveries (eg, gestational age at birth), and encounter-level data, including visits, diagnoses, laboratory tests, and procedures. This study was approved by the NYU Langone Health institutional review board. Informed consent was waived based on 45 CFR §46. The study followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

The study cohort included US patients who had a documented encounter involving a live birth at a facility using the selected EHR network from January 1, 2023, to December 31, 2024, to ensure consistency with published NCHS-based pregnancy surveillance metrics, which report on all live births. Documentation of self-reported history of pregnancies and pregnancies with deliveries outside facilities that used the study EHR database were excluded (eFigure 1 in Supplement 1). Births with missing gestational age or gestational age of less than 20 weeks or more than 43 weeks were also excluded to mitigate potential inaccuracies in vital status and ensure accurate calculation of prenatal care adequacy based on gestational age.²⁶ To avoid measurement errors among patients who used different health care facilities for their delivery than for their prenatal care, we further excluded patients who did not receive any prenatal care at a facility within the EHR network but delivered at a facility within the network or patients with an initial visit or pregnancy episode initiated less than 1 week before delivery. Prenatal care visit data were restricted to institutions that fully contributed data from January 1, 2022, to December 31, 2024.

NCHS Data

We obtained demographic and clinical information on the target population of all US residents with live births from January 1, 2023, to December 31, 2024, from the Centers for Disease Control and Prevention (CDC) WONDER database. Reports from CDC WONDER aggregate counts of live births to US residents derived from NCHS natality data, the primary source of US pregnancy surveillance data.²⁷ The Healthy People 2030 and March of Dimes websites provide national and state-based estimates of pregnancy surveillance metrics for all US residents with live births (including singleton and multiple births) derived from the NCHS data.^4,25 We obtained 2023 national and state-based population-representative estimates from these websites overall and by age and race and ethnicity to serve as our gold standard for external validation of our EHR-based metrics. The Healthy People 2030 website provides metrics for prenatal care adequacy. The March of Dimes website provides metrics for early care initiation. National Center for Health Statistics–based estimates included birthing people younger than 18 years in reporting aggregations.

EHR-Based Metrics of Prenatal Care Use

When measuring early and adequate care metrics using EHR data, we first had to identify initial and subsequent prenatal care visits using data elements that are available in the selected EHR database. Because population-based estimation of these prenatal metrics using EHR data, to our knowledge, has not previously been conducted and coding practices (eg, use of certain encounter types, preferred diagnosis, or procedure codes) can vary across institutions, we evaluated multiple definitions of initial and subsequent prenatal visits of varying sensitivity and specificity. Few studies have defined prenatal visits using EHR network data; thus, our definitions were largely informed by insurance claims–based definitions from the literature and clinical expertise.^26,28,29,30 For all definitions, initial and subsequent prenatal visits were limited to encounters that occurred between the pregnancy start date and birth date.

Table 1 summarizes the 6 definitions we developed, including the information used to define the initial prenatal visit and subsequent prenatal visits. First, pregnancy diagnoses defined the initial and subsequent visits as any visit with an active pregnancy-related diagnosis code (eTable 1 in Supplement 1).²⁹ Second, prenatal specialties included visits that either had a prenatal encounter type (initial prenatal or routine prenatal) or an outpatient encounter with a prenatal-providing department or clinician specialty.^28,31,32 Third, any prenatal evidence expanded the definition for prenatal specialties to also include evidence of prenatal diagnoses, procedures, or laboratory tests.²⁶

Table 1. Definitions of Initial and Subsequent Prenatal Visits Using Electronic Health Record Data.

Definition	Description
Definition	Initial prenatal visit	Subsequent prenatal visit
Pregnancy diagnoses	Active pregnancy diagnosis^a	Active pregnancy diagnosis^a
Prenatal specialties	Prenatal encounter type (initial or routine prenatal encounter) or clinician-facing encounter type and prenatal-related department or clinician specialties^b^,^c	Prenatal encounter type (initial or routine prenatal encounter) or expanded clinician-facing encounter type and prenatal-related department or clinician specialties^c^,^d
Any prenatal evidence	Prenatal encounter type (initial or routine prenatal encounter), clinician-facing encounter type and prenatal-related department or clinician specialties, or active pregnancy-related outpatient procedure or laboratory code^a^,^b^,^c	Prenatal encounter type (initial or routine prenatal encounter), expanded clinician-facing encounter type and prenatal-related department or clinician specialties, or prenatal visit diagnosis or procedure code^a^,^c^,^d
Strict encounter type	Initial prenatal encounter type	Prenatal encounter type (initial or routine prenatal encounter)
Initial encounter type	Initial prenatal encounter type	Prenatal encounter type (initial or routine prenatal encounter) or expanded clinician-facing encounter type and prenatal-related department or clinician specialties^c^,^d
Any initial evidence	Initial prenatal encounter type or initial prenatal-related diagnosis, procedure, or laboratory code and clinician-facing encounter type and prenatal-related department or clinician specialties^a^,^b^,^c	Prenatal encounter type (initial or routine prenatal encounter), expanded clinician-facing encounter type and prenatal-related department or clinician specialties, or prenatal visit diagnosis or procedure code^a^,^c^,^d

Open in a new tab

^{^a}

A list of prenatal-related diagnosis, procedure, and laboratory codes is provided in eTable 1 in Supplement 1.

^{^b}

Clinician-facing encounter types included office visits, telemedicine, nurse-only, and hospital outpatient visits.

^{^c}

Prenatal-related department or clinician specialties included obstetrics, obstetrics and gynecology, maternal and fetal medicine, perinatology, midwifery, primary care, family medicine, internal medicine, general internal medicine, nutrition, reproductive endocrinology and infertility, infertility, general practice, gynecology, nurse practitioner, and physician assistant.

^{^d}

Expanded clinician-facing encounter types included office visits, telemedicine, nurse-only visits, hospital outpatient visits, evaluation, follow-up, results follow-up, consult, education, and nutrition.

The definitions of strict encounter type, initial encounter type, and any initial evidence required direct evidence of an initial prenatal visit to qualify the patient as eligible. Strict encounter type required an initial prenatal encounter type for the first visit and an initial or routine prenatal encounter type for all subsequent visits. Initial encounter type required an initial prenatal encounter type for the first visit but widened the definition for subsequent visits to include either a prenatal encounter type or an outpatient encounter with a prenatal-providing department or clinician specialty. Any initial evidence expanded the definition for initial encounter type to also include evidence of an initial visit and subsequent visits through diagnosis, laboratory tests, or procedure codes (eTable 1 in Supplement 1).

For all 6 prenatal care use definitions, the week of initiation was calculated based on the pregnancy start date and date of the first visit that qualified as an initial prenatal visit. The total number of prenatal care encounters was counted based on distinct dates of the initial visit and any subsequent visits. Apart from the any initial evidence and any prenatal evidence definitions for the initial prenatal visit, we did not count prenatal laboratory tests or procedures (eg, ultrasonography) that occurred on dates without a typical clinician-facing encounter type (eg, office visit) as a distinct prenatal visit.

We calculated 2 metrics of prenatal care use: early initiation and adequacy of care. Early initiation was measured as the proportion of patients initiating care before the 13th week of pregnancy, and adequacy of care was measured as the proportion of patients with adequate or better prenatal care per the Adequacy of Prenatal Care Utilization Index.³³ This index categorizes prenatal care use based on timing of the initial visit and frequency of visits adjusted for the newborn’s gestational age. A classification of adequate or higher corresponded to having an initial visit by the fourth month of pregnancy and attending 80% or more of expected visits for the newborn’s gestational age. These metric definitions align with the published NCHS-based estimates that were used for external validation.^4,25

Statistical Analysis

Since the EHR data we used represent a nonrandom sample of patients, we used raking to account for demographic and clinical differences between the EHR sample and the target population of all US residents with live births. Raking weights (eTable 2 in Supplement 1) were calculated using the marginal target population distributions for the covariates of age (18-24, 25-34, and ≥35 years), race and ethnicity (Asian, Native Hawaiian, or Other Pacific Islander; Black; Hispanic or Latino; White; and other [American Indian, Alaska Native, multiracial, and other; grouped due to small sample sizes producing unstable weights]), primary insurance (Medicaid, other [self-pay, Medicare, and miscellaneous or other; non-Medicaid categories were combined due to unstable weights]), gestational diabetes, gestational hypertension, prior preterm birth, and geographic region (Northeast, Midwest, South, West). Race and ethnicity were ascertained by a combination of self-report and hospital staff and were included as a covariate to address the lack of representativeness in the EHR sample.

We had 2 a priori approaches to compare the validity and reliability of different definitions. First, we assessed external validity by comparing 2023 crude and weighted EHR estimates with published NCHS-based estimates for live births occurring in the US using the two 1-sided test of equivalence.^4,25 We assessed the validity of definitions stratified by age, race and ethnicity, and state of residence to identify potential differential measurement errors. Second, we assessed reliability over time by visualizing quarterly EHR metrics from January 1, 2023, to December 31, 2024, 2024. Analyses were performed using R, version 4.5.1 (R Project for Statistical Computing). Two 1-sided P < .05 was considered significant.

Results

In total, there were 1 963 496 eligible patients (mean age [SD], 29.5 [5.7] years; 100% women) who had at least 1 prenatal visit at a facility in the selected EHR network through any of the tested definitions in Table 1 (27.2% of the total live births in the US) (Table 2). In the EHR group, 0.4% were American Indian or Alaska Native; 4.0%, Asian; 13.5%, Black; 18.3%, Hispanic or Latino; 0.4%, Native Hawaiian or Other Pacific Islander; 44.8%, White; 7.8%, multiracial; and 1.6%, other. In the NCHS group, 0.7% were American Indian or Alaska Native; 6.3%, Asian; 13.5%, Black; 26.7%, Hispanic or Latino; 0.3%, Native Hawaiian or Other Pacific Islander; 50.0%, White; 2.5%, multiracial; and 0%, other. Compared with the general population of people who gave birth in the US, the EHR group included a substantially lower proportion of patients who had Medicaid insurance (40.5% vs 21.1%) and a higher proportion of patients with missing race and ethnicity information (0% vs 9.1%), prior preterm birth (4.0% vs 8.8%), gestational diabetes (8.4% vs 12.3%), or gestational hypertension (10.2% vs 15.4%).

Table 2. Demographic and Clinical Characteristics of People With Live Births in the EHR and NCHS Data, 2023-2024.

Characteristic	Patients, No. (%)
Characteristic	EHR (n = 1 963 496)	NCHS (n = 7 224 951)^a
Age, y
<18	NA	73 672 (1.0)
18-19	66 555 (3.4)	208 071 (2.9)
20-24	352 037 (17.9)	1 228 770 (17.0)
25-29	560 793 (28.6)	1 977 384 (27.4)
30-34	599 917 (30.6)	2 210 461 (30.6)
≥35	384 194 (19.6)	1 526 593 (21.1)
Race and ethnicity^b
American Indian or Alaska Native	7006 (0.4)	49 449 (0.7)
Asian	79 434 (4.0)	454 831 (6.3)
Black	265 407 (13.5)	974 440 (13.5)
Hispanic or Latino	359 333 (18.3)	1 929 292 (26.7)
Native Hawaiian or Other Pacific Islander	8364 (0.4)	21 213 (0.3)
White	879 622 (44.8)	3 614 250 (50.0)
Multiracial	153 879 (7.8)	181 476 (2.5)
Other	31 088 (1.6)	NA
Unknown	179 363 (9.1)	NA
Primary insurance^c
Medicaid	414 873 (21.1)	2 924 992 (40.5)
Other	1 442 684 (73.5)	3 918 377 (54.2)
Self-pay	13 676 (0.7)	325 159 (4.5)
Unknown	92 263 (4.7)	56 423 (0.8)
US census region
Northeast	341 545 (17.4)	1 146 995 (15.9)
Midwest	508 896 (25.9)	1 471 900 (20.4)
South	848 194 (43.2)	2 954 012 (40.9)
West	264 861 (13.5)	1 652 044 (22.9)
Prior preterm birth	173 088 (8.8)	287 309 (4.0)
Gestational diabetes	241 757 (12.3)	603 709 (8.4)
Gestational hypertension	301 599 (15.4)	739 047 (10.2)

Open in a new tab

Abbreviations: EHR, electronic health data; NA, not applicable; NCHS, National Center for Health Statistics.

^{^a}

NCHS data were obtained from the Centers for Disease Control and Prevention WONDER database natality records for 2016 to 2024.²⁷

^{^b}

In the EHR data, other is a distinct category for race. In NCHS data, other and unknown are not eligible categories.

^{^c}

In the EHR data, other insurance includes Medicare and miscellaneous or other (including commercial and private insurance). In NCHS data, other is a distinct category for source of payment.

Early Initiation of Prenatal Care

Within the EHR group, the crude proportion of patients with early prenatal care was relatively stable (eg, prenatal specialties: 70.0% [95 CI, 69.9%-70.1%]; initial encounter type: 70.8% [95% CI, 70.7%-70.9%]) except when using pregnancy diagnoses to identify the initial visit, for which only 62.7% (95% CI, 62.6%-62.8%) of patients were classified as having received early care (Table 3). Incorporating raking weights slightly reduced these estimates. Across all definitions, prevalence of early care was lower than that reported by NCHS (76.1%; 95% CI, 76.1%-76.1%). For example, weighted prevalence of early care was 61.2% (95% CI, 61.0%-61.3%) for pregnancy diagnoses and 68.6% (95% CI, 68.4%-68.8%) for initial encounter type. When comparing the distribution of month of initial prenatal visit among individuals with live births, definitions that required an initial prenatal encounter type or any evidence of an initial prenatal visit more closely aligned with the NCHS distribution than definitions that did not require this evidence (eFigure 2 in Supplement 1).

Table 3. External Validation of EHR-Based vs NCHS-Based Definitions of Early Initiation and Adequacy of Prenatal Care Among US Live Births in 2023.

Definition	Patients, No.^a	Prevalence, % (95% CI)
		Early initiation of care		Adequate or better care
		Crude	Weighted	Crude	Weighted
Pregnancy diagnoses	939 933	62.7 (62.6-62.8)	61.2 (61.0-61.3)	56.4 (56.3-56.5)	55.8 (55.6-55.9)
Prenatal specialties	719 371	70.0 (69.9-70.1)	67.7 (67.6-67.9)	60.6 (60.5-60.7)	59.4 (59.3-59.6)
Any prenatal evidence	772 617	70.2 (70.1-70.3)	67.8 (67.6-67.9)	64.4 (64.3-64.5)	63.7 (63.6-63.8)
Strict encounter type	378 601	70.8 (70.7-70.9)	68.6 (68.4-68.8)	69.5 (69.4-69.6)	66.9 (66.7-67.1)
Initial encounter type	378 601	70.8 (70.7-70.9)	68.6 (68.4-68.8)	76.9 (76.8-77.0)	74.5 (74.3-74.7)^b
Any initial evidence	462 522	70.3 (70.2-70.3)	68.0 (67.9-68.2)	77.6 (77.5-77.7)	76.0 (75.9-76.2)^b
NCHS^c	3 596 017	76.1 (76.1-76.1)	NA	75.2 (75.1-75.2)	NA

Open in a new tab

Abbreviations: EHR, electronic health record; NA, not applicable; NCHS, National Center for Health Statistics.

^{^a}

Patients who had an initial prenatal visit per the definition of prenatal care.

^{^b}

Statistically equivalent through the two 1-sided test of equivalence at 0.01 equivalence bound.

^{^c}

NCHS-based estimates were obtained from the March of Dimes and Healthy People 2030 websites.^4,25

Adequacy of Prenatal Care

The crude proportion of patients in the EHR group who received adequate care varied substantially across definitions (from 56.4% [95% CI, 56.3%-56.5%) for pregnancy diagnosis to 77.6% [95% CI, 77.5%-77.7%] for any initial evidence). Under the other definitions with no direct evidence of an initial prenatal visit, 60.6% (95% CI, 60.5%-60.7%) diagnosed by prenatal specialties and 64.4% (95% CI, 64.3%-64.5%) by any prenatal evidence received adequate care. Restricting the definitions to patients with documented initial visits (strict encounter type, initial encounter type, and any initial evidence) produced estimates closer to the NCHS data (Table 3). After incorporating raking weights, initial encounter type and any initial evidence produced adequacy estimates that were statistically equivalent to NCHS estimates (initial encounter type: 74.5% [95% CI, 74.3%-74.7%]; any initial evidence: 76.0% [95% CI, 75.9%-76.2%]; NCHS: 75.2% [95% CI, 75.1%-75.2%]; two 1-sided test, P < .001 at 0.01 equivalence bound).

We observed a bimodal distribution in number of prenatal encounters among patients with live births when using definitions that did not require evidence of an initial prenatal visit, with peaks around 1 to 3 visits and around 13 to 17 visits (eFigure 3 in Supplement 1). Definitions that required evidence of an initial prenatal visit produced a distribution that was closer to normal, with a peak around 10 to 15 visits, similar to the NCHS distribution. Ultimately, we selected any initial evidence as the optimal EHR-based definition since it produced an adequacy of care estimate that was equivalent to the NCHS estimate and captured a larger number of prenatal patients than did strict encounter type and initial encounter type, which used only an initial prenatal encounter type for identifying eligible patients (any initial evidence, 462 522 patients; vs strict encounter type and initial encounter type, 378 601 patients).

Heterogeneity

We assessed heterogeneity in the optimal definition by age, race and ethnicity, state, and calendar quarter in which the patient gave birth. When stratified by age or race and ethnicity, patterns in weighted metrics generally aligned with known disparities per NCHS data. Specifically, the proportion of patients with early (Figure 1A) or adequate (Figure 1C) care increased with age until 30 years or older and then plateaued. Additionally, Asian, White, and multiracial patients had higher levels of early (Figure 1B) and adequate (Figure 1D) care, while Native Hawaiian or Other Pacific Islander patients had the lowest levels. Electronic health record–based metrics showed greater variability at the state level than NCHS metrics for early care (EHR: median proportion, 69.4% [IQR, 65.4%-74.4%]; NCHS: median proportion, 77.6% [IQR, 74.5%-81.3%]) and adequate care (EHR: median proportion, 76.0% [IQR, 67.8%-80.6%]; NCHS: median proportion, 78.1% [IQR, 74.3%-80.6%]) (Figure 2). There was greater underestimation of early and adequate care in the Southwest and Midwest regions of the US (eFigures 4 and 5 in Supplement 1). Electronic health record–based estimates of early and adequate care were consistent across time (eFigure 6 in Supplement 1) and by age and race and ethnicity (eFigure 7 in Supplement 1).

Figure 1. — The EHR-based estimates were based on the any initial evidence definition. The NCHS-based estimate for adequacy of care in those younger than 20 years includes birthing people aged 15 to 19 years, while the EHR-based estimate includes birthing people aged 18 to 19 years due to misalignment between national reporting aggregations and the EHR sample. Error bars indicate 95% CIs.

Discussion

In this study, the selected EHR network produced reliable and valid estimates of adequacy of prenatal care that aligned with known demographic patterns and were stable over time if defined with consideration of documentation of an initial prenatal visit and heterogeneity in coding practices across institutions. However, EHR network data consistently underestimated the proportion of patients with early initiation of prenatal care compared with NCHS data. While EHR network data demonstrated potential for enhancing or supplementing existing pregnancy surveillance systems, metrics must be carefully constructed to account for potential sources of bias and heterogeneity in these data.

One key source of bias in EHR networks is the potential for missing encounters caused by care received at institutions that do not participate in the network.^16,19 As a result, definitions that perform well in claims data, which provide a near comprehensive view of care among individuals with continual enrollment, may not be directly translatable to an EHR network. For example, the broader definitions used in these analyses were largely built on the claims-based literature,^26,28,29,30 and they produced bimodal distributions of the number of prenatal visits. These findings suggest that many of these patients may have switched clinicians or health systems during pregnancy, resulting in an underestimation of care use. By contrast, restricting eligible prenatal patients to those with evidence of an initial visit, whether through a specific encounter type variable or through laboratory and procedure codes that are generally used in the initial visit, produced visit distributions and adequacy estimates that were more comparable with NCHS-based estimates.

Variability in documentation and coding practices poses another significant challenge when using EHR network data for surveillance purposes.¹⁹ Inconsistencies in how visit types or specialties are classified over time or across systems can complicate the enumeration of prenatal visits.³⁴ In this study, we found that 378 601 patients had a visit with an initial prenatal encounter type, but an additional 83 921 patients had evidence of an initial prenatal visit in procedure or laboratory codes. Moreover, only 69.5% (95% CI, 69.4%-69.6%) of patients were classified as having received adequate care when using a strict encounter type definition to identify prenatal visits. Incorporating specialties, diagnoses, and procedure codes consistent with prenatal care increased the adequacy metric to 77.6% (95% CI, 77.5%-77.7%). To address variability in documentation and coding practices across EHR networks, researchers should accommodate diverse approaches to data entry and coding and assess differences in performance by health system.

Strengths and Limitations

This study has several strengths. We used data from one of the largest EHR networks in the US. The size and geographic representativeness of the network allowed for precise estimation in small demographic subgroups across the country. Additionally, we developed a novel adaptation of prenatal care measures for use with EHR data and tested several definitions. Finally, while our study focused on prenatal care use visit adequacy, these data also have the potential to be used to explore postpartum care use or quality of care indicators, both of which are important for maternal health.

This study also has several limitations. First, we could not perform an internal validation to confirm completeness and accuracy of information housed within medical records since the EHRs used represent a deidentified database with no access to full medical records or the ability to link birth certificate data. Instead, we validated these EHR-based estimates against an external gold standard source using NCHS natality data. We acknowledge that NCHS natality data have distinct data quality issues, such as overreporting or underreporting of prenatal care,^12,35 but they remain the current gold standard and contained data on the full target population. Second, we excluded patients without evidence of prenatal care prior to delivery because we could not determine whether they truly had no care or they received care at a health system outside of a facility in the selected EHR network. However, NCHS data report that only 2% of birthing persons received no prenatal care from 2023 to 2024,²⁷ suggesting that this exclusion was likely to have resulted in a minor overestimation of care use in the sample in our study. The study sample was also limited to adults aged 18 years or older; this age limitation may have slightly overestimated prenatal care measures compared with NCHS estimates, which include births among individuals younger than 18 years (1% of births²⁷). Additionally, we acknowledge that this study focused on a coarse adequacy metric that included live births only, combined high-risk and low-risk pregnancies, and masked heterogeneity in quality of care. Individuals with pregnancies that did not result in a live birth were excluded, and those with high-risk pregnancies (eg, multiple pregnancies, patients with diabetes) may be recommended to have more visits. However, our goal was to make direct comparisons with published national estimates of well-established surveillance measures. We also observed demographic and clinical differences between patients in the EHR group and the target population of US birthing persons, including substantial underrepresentation of individuals insured by Medicaid. Factors affecting selection into the EHR sample may have influenced prenatal care use estimates (eg, higher use among those with higher-risk pregnancies or those with greater resources and/or income to attend these visits). We attempted to address these biases through raking; however, there may have been residual biases in the EHR-based estimates, particularly at the state level. Finally, results may not be generalizable to other EHR-based data sources, including smaller networks or single health care systems.

Conclusions

In this cohort study, EHR data reliably informed surveillance of prenatal care adequacy after adjusting for nonrepresentativeness of the patient population. These findings suggest that EHR network data offer a promising, near real-time source for monitoring prenatal care adequacy, but methodologic rigor is critical. Restricting analyses to patients with direct evidence of initial visits, accounting for institutional variability in data documentation, and adjusting for nonrepresentativeness of the patient population may enhance reliability and reduce bias. These data may inform policy efforts, public health strategies, and clinical initiatives aimed at enhancing prenatal care use.

Supplement 1.

eFigure 1. Study cohort flowchart

eTable 1. Diagnosis, procedure, and lab codes used for prenatal visit

eTable 2. Distribution of raking weights under the six tested

eFigure 2. Distribution of month of the initial prenatal visit in NCHS live births (A) vs Cosmos live births under various definitions for the initial prenatal visit (B-F)

eFigure 3. Distribution of number of prenatal visits among NCHS live births (A) vs Cosmos live births under various definitions for identifying prenatal visits (B-G)

eFigure 4. Absolute difference in the proportion of 2023 live births with early initiation of prenatal care by state, Cosmos weighted estimates vs NCHS estimates

eFigure 5. Absolute difference in the proportion of 2023 live births with adequate or higher prenatal care by state, Cosmos weighted estimates vs NCHS estimates

eFigure 6. Cosmos EHR-based estimates of early initiation and adequacy of prenatal care by quarter of birth, 2023 US live births

eFigure 7. Cosmos weighted EHR-based vs NCHS-based estimates of early initiation and adequacy of prenatal care by age and race/ethnicity, 2023 US live births

jamahealthforum-e261295-s001.pdf^{(950.3KB, pdf)}

Supplement 2.

Data Sharing Statement

jamahealthforum-e261295-s002.pdf^{(14.8KB, pdf)}

References

1.Partridge S, Balayla J, Holcroft CA, Abenhaim HA. Inadequate prenatal care utilization and risks of infant mortality and poor birth outcome: a retrospective analysis of 28,729,765 US deliveries over 8 years. Am J Perinatol. 2012;29(10):787-793. doi: 10.1055/s-0032-1316439 [DOI] [PubMed] [Google Scholar]
2.Holcomb DS, Pengetnze Y, Steele A, Karam A, Spong C, Nelson DB. Geographic barriers to prenatal care access and their consequences. Am J Obstet Gynecol MFM. 2021;3(5):100442. doi: 10.1016/j.ajogmf.2021.100442 [DOI] [PubMed] [Google Scholar]
3.Liu TC, Chen B, Chan YS, Chen CS. Does prenatal care benefit maternal health? a study of post-partum maternal care use. Health Policy. 2015;119(10):1382-1389. doi: 10.1016/j.healthpol.2015.06.004 [DOI] [PubMed] [Google Scholar]
4.Increase the proportion of pregnant women who receive early and adequate prenatal care—MICH-08. US Dept of Health and Human Services . Accessed March 5, 2026. https://odphp.health.gov/healthypeople/objectives-and-data/browse-objectives/pregnancy-and-childbirth/increase-proportion-pregnant-women-who-receive-early-and-adequate-prenatal-care-mich-08/data
5.Office of the Surgeon General . The Surgeon General’s Call to Action to Improve Maternal Health. US Dept of Health and Human Services; 2020. [Google Scholar]
6.Lee J, Howard KJ, Greif A, Howard JT. Trends and racial/ethnic disparities in prenatal care (PNC) use from 2016 to 2021 in the United States. J Racial Ethn Health Disparities. 2025;12(5):3095-3106. doi: 10.1007/s40615-024-02115-9 [DOI] [PubMed] [Google Scholar]
7.Ahn R, Gonzalez GP, Anderson B, Vladutiu CJ, Fowler ER, Manning L. Initiatives to reduce maternal mortality and severe maternal morbidity in the United States: a narrative review. Ann Intern Med. 2020;173(11)(suppl):S3-S10. doi: 10.7326/M19-3258 [DOI] [PubMed] [Google Scholar]
8.American College of Obstetricians and Gynecologists Committee on Clinical Consensus—Obstetrics . Tailored prenatal care delivery for pregnant individuals: ACOG clinical consensus No. 8. Obstet Gynecol. 2025;e145(5):565-577. [DOI] [PubMed]
9.Gregory ECW, Martin JA, Argov EL, Osterman MJK. Assessing the quality of medical and health data from the 2003 Birth Certificate Revision: results from New York City. Natl Vital Stat Rep. 2019;68(8):1-20. [PubMed] [Google Scholar]
10.Josberger RE, Wu M, Nichols EL. Birth certificate validity and the impact on primary cesarean section quality measure in New York State. J Community Health. 2019;44(2):222-229. doi: 10.1007/s10900-018-0577-y [DOI] [PubMed] [Google Scholar]
11.Northam S, Knapp TR. The reliability and validity of birth certificates. J Obstet Gynecol Neonatal Nurs. 2006;35(1):3-12. doi: 10.1111/j.1552-6909.2006.00016.x [DOI] [PubMed] [Google Scholar]
12.Reichman NE, Hade EM. Validation of birth certificate data: a study of women in New Jersey’s HealthStart program. Ann Epidemiol. 2001;11(3):186-193. doi: 10.1016/S1047-2797(00)00209-X [DOI] [PubMed] [Google Scholar]
13.Handler AS, Johnson K, Rankin KM, Velonis AJ, James AR, Kotelchuck M. Shuttering the Pregnancy Risk Assessment Monitoring System (PRAMS): a dangerous attack on US mothers and infants. Am J Public Health. 2025;115(6):848-850. doi: 10.2105/AJPH.2025.308107 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Marshall J. CDC shutters PRAMS program on maternal and infant health. Talking Points Memo. Accessed February 12, 2026. https://talkingpointsmemo.com/edblog/cdc-shutters-prams-program-on-maternal-and-infant-health
15.Gourevitch RA, Peahl AF, McConnell M, Shah N. Understanding the impact of prenatal care: improving metrics, data, and evaluation. Health Affairs Forefront. 2020. Accessed April 3, 2026. https://www.healthaffairs.org/content/forefront/understanding-impact-prenatal-care-improving-metrics-data-and-evaluation
16.Al-Sahab B, Leviton A, Loddenkemper T, Paneth N, Zhang B. Biases in electronic health records data for generating real-world evidence: an overview. J Healthc Inform Res. 2023;8(1):121-139. doi: 10.1007/s41666-023-00153-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Conderino S, Anthopolos R, Albrecht SS, et al. Addressing information biases within electronic health record data to improve the examination of epidemiologic associations with diabetes prevalence among young adults: cross-sectional study. JMIR Med Inform. 2024;12:e58085. doi: 10.2196/58085 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Conderino S, Thorpe LE, Divers J, et al. Addressing selection biases within electronic health record data for estimation of diabetes prevalence among New York City young adults: a cross-sectional study. BMJ Public Health. 2024;2(2):e001666. doi: 10.1136/bmjph-2024-001666 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Bower JK, Patel S, Rudy JE, Felix AS. Addressing bias in electronic health record-based surveillance of cardiovascular disease risk: finding the signal through the noise. Curr Epidemiol Rep. 2017;4(4):346-352. doi: 10.1007/s40471-017-0130-z [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Le Meur N, Gao F, Bayat S. Mining care trajectories using health administrative information systems: the use of state sequence analysis to assess disparities in prenatal care consumption. BMC Health Serv Res. 2015;15(1):200. doi: 10.1186/s12913-015-0857-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Hung P, Yu J, Harrison SE, et al. ; National COVID Cohort Collaborative Consortium . Racial and ethnic and rural variations in the use of hybrid prenatal care in the US. JAMA Netw Open. 2024;7(12):e2449243-e2449243. doi: 10.1001/jamanetworkopen.2024.49243 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Putnam KE, Biel FM, Hoopes M, et al. Landscape of pregnancy care in US community health centers. J Am Board Fam Med. 2023;36(4):574-582. doi: 10.3122/jabfm.2023.230025R1 [DOI] [PubMed] [Google Scholar]
23.Epic Cosmos. Epic Cosmos Corporation . Accessed April 12, 2026. https://cosmos.epic.com/
24.Son M, Culhane JF, Louis JM, et al. Severe maternal morbidity rates in a US-based electronic health record database, 2018-2022. J Perinatol. 2023;43(10):1316-1318. doi: 10.1038/s41372-023-01765-7 [DOI] [PubMed] [Google Scholar]
25.Prenatal care. March of Dimes . Accessed February 19, 2025. https://www.marchofdimes.org/peristats/data?top=5&lev=1&stop=29&ftop=32&reg=99&sreg=36&obj=1&slev=4
26.Gourevitch RA, Natwick T, Chaisson CE, Weiseth A, Shah NT. Variation in guideline-based prenatal care in a commercially insured population. Am J Obstet Gynecol. 2022;226(3):413.e1-413.e19. doi: 10.1016/j.ajog.2021.09.038 [DOI] [PubMed] [Google Scholar]
27.CDC WONDER: about natality, 2016-2024 expanded. Centers for Disease Control and Prevention. Accessed October 15, 2025. https://wonder.cdc.gov/natality-expanded-current.html
28.Bennett WL, Chang HY, Levine DM, et al. Utilization of primary and obstetric care after medically complicated pregnancies: an analysis of medical claims data. J Gen Intern Med. 2014;29(4):636-645. doi: 10.1007/s11606-013-2744-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Swartz JJ, Hainmueller J, Lawrence D, Rodriguez MI. Oregon’s expansion of prenatal care improved utilization among immigrant women. Matern Child Health J. 2019;23(2):173-182. doi: 10.1007/s10995-018-2611-1 [DOI] [PubMed] [Google Scholar]
30.2025 Core set of children’s health care quality measures for Medicaid and CHIP (Child Core Set). Centers for Medicare & Medicaid Services. 2025. Accessed April 14, 2026. https://www.medicaid.gov/medicaid/quality-of-care/downloads/2025-child-core-set.pdf
31.Madden N, Emeruwa UN, Friedman AM, et al. Telehealth uptake into prenatal care and provider attitudes during the COVID-19 pandemic in New York City: a quantitative and qualitative analysis. Am J Perinatol. 2020;37(10):1005-1014. doi: 10.1055/s-0040-1712939 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Ferrara A, Greenberg M, Zhu Y, et al. Prenatal health care outcomes before and during the COVID-19 pandemic among pregnant individuals and their newborns in an integrated US health system. JAMA Netw Open. 2023;6(7):e2324011. doi: 10.1001/jamanetworkopen.2023.24011 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and a proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health. 1994;84(9):1414-1420. doi: 10.2105/AJPH.84.9.1414 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Heaman MI, Newburn-Cook CV, Green CG, Elliott LJ, Helewa ME. Inadequate prenatal care and its association with adverse pregnancy outcomes: a comparison of indices. BMC Pregnancy Childbirth. 2008;8:15. doi: 10.1186/1471-2393-8-15 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Roohan PJ, Josberger RE, Acar J, Dabir P, Feder HM, Gagliano PJ. Validation of birth certificate data in New York State. J Community Health. 2003;28(5):335-346. doi: 10.1023/A:1025492512915 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials