Length of gestation and postpartum visit attendance

Abstract

Introduction

Approximately 40% of obstetric patients do not attend a postpartum visit. Postpartum parents of preterm infants (<37 weeks) are often sicker than parents of full‐term infants. Nonetheless, they may forgo postpartum visits to attend to the needs of their infant when admitted to the neonatal intensive care unit. Our goal was to evaluate whether postpartum visit attendance varied with length of gestation at birth, hypothesizing after preterm birth, parents would be less likely to attend postpartum visits than after full‐term birth.

Methods

Retrospective cohort study of births in Epic Systems Cosmos research platform (2018–2024), a national electronic health record database with deidentified, patient‐level data. To increase the likelihood that postpartum visits, if attended, would be captured by Cosmos, we only included births with prenatal visits in the database. Bivariate analyses examined characteristics associated with not attending postpartum visits. Multilevel, multivariable, modified Poisson regression models calculated adjusted risk ratios (aRRs) of not attending postpartum visits after various lengths of gestation compared to full‐term birth (39–40 weeks). Models adjusted for age, race and ethnicity, insurance, residential Centers for Disease Control and Prevention Social Vulnerability Index, rural versus urban residence, smoking, body mass index (BMI), hypertension, diabetes, parity, cesarean birth, and birth year.

Results

Of the 2,403,574 included births, 653,552 (27%) parents did not attend a postpartum visit. For the several characteristics with significant differences in visit attendance rates, notably high rates of not attending a postpartum visit were observed for Black non‐Hispanic race and ethnicity (32%), smoking during pregnancy (56%), BMI > 30 kg/m² (30%), no diabetes (28%), multiparity (29%), and no cesarean birth (29%). After multivariable adjustment, preterm (<37 weeks), early‐term (37–38 weeks), and late‐ and post‐term (41–43 weeks) births were all associated with not attending a postpartum visit compared to full‐term (39–40 weeks’ gestation) birth. Periviable birth (22–23 weeks) demonstrated the highest risk (aRR, 1.21; 95% confidence interval [CI], 1.13–1.29).

Conclusions

Compared to full‐term birth, all other length‐of‐gestation categories were significantly associated with not attending a postpartum visit, with periviable birth at the highest risk. Interdisciplinary, innovative approaches to provide postpartum care to this vulnerable population are needed.

1. INTRODUCTION

The United States is facing a maternal health crisis, with a maternal mortality rate of 18 deaths per 100,000 live births in 2024—more than triple that of other comparable high‐income countries [1, 2, 3]. The majority (65%) of maternal deaths occur after birth, with 35% happening in the first 42 days postpartum [4]. It is estimated that more than 80% of pregnancy‐related deaths are preventable [4, 5]. Notably, nearly 40% of postpartum patients do not receive postpartum care [6]. Postpartum care provides an opportunity to identify and treat medical conditions common in the postpartum period, such as hypertension and depression, which can lead to substantial maternal morbidity or mortality [6, 7]. It is vital to identify factors that influence postpartum care receipt to target interventions to improve visit attendance and ultimately reduce adverse maternal outcomes.

Postpartum parents of preterm infants are often sicker than parents of full‐term infants. Parents of infants in the neonatal intensive care unit (NICU) are 2.8 times more likely to have hypertensive disorders of pregnancy and 3–4 times more likely to have postpartum depression and posttraumatic stress disorder [8, 9, 10, 11, 12, 13]. High‐risk obstetric patients with preexisting or pregnancy‐associated conditions (including both physical and mental health conditions) are more likely to have preterm infants or sick infants requiring NICU admission [14, 15, 16].

We, and others, have shown that parents of preterm infants may choose to prioritize their infant's needs over their own healthcare needs [17, 18, 19]. In a previous study using data from two Philadelphia hospitals, we found that parents of infants born before 35 weeks’ gestation were less likely to receive postpartum care than parents of full‐term infants [20]. However, it is unclear whether those findings from one city are generalizable. Thus, the goal of this work was to use a nationwide sample to determine if the length of gestation is associated with postpartum visit attendance. Our hypothesis was that the shortest lengths of gestation would be associated with not attending postpartum visits.

2. MATERIALS AND METHODS

2.1. Study population

We performed a retrospective cohort study of live births in Epic Systems Corporation's Cosmos research platform (Cosmos) from January 2018 to December 2024. Cosmos is a dataset created in collaboration with a community of Epic health systems representing more than 300 million patient records from over 1800 hospitals and 42,000 clinics as of December 2025 [21]. Cosmos includes patient data from all 50 states and Washington, DC. All Cosmos patient‐level data are deidentified. We restricted analysis to births with prenatal visits in Cosmos to maximize the likelihood that we would capture a postpartum visit, if it occurred, as most obstetric patients receive prenatal and postpartum care in the same practice. Prenatal care was defined as having at least one healthcare encounter in obstetrics or family medicine departments during pregnancy or a healthcare encounter during pregnancy that was associated with International Classification of Diseases (ICD‐10‐CM) codes indicating prenatal care (Z33, Z34, O09). We randomly selected only one birth per individual so that postpartum parents only contributed once during the study period. Births with maternal deaths and missing gestational age were excluded. Births with missing covariate data were excluded for primary analyses but included in sensitivity analyses.

2.2. Exposure—Length of gestation

The primary exposure was the length of gestation at birth obtained from the electronic medical record documentation. We used the following length‐of‐gestation categories: periviable preterm birth (22–23 weeks), extreme preterm birth (24–27 weeks), very preterm birth (28–31 weeks), moderate preterm birth (32–33 weeks), late preterm birth (34–36 weeks), early‐term birth (37–38 weeks), full‐term birth (39–40 weeks), and late‐term to post‐term birth (41–43 weeks). Length‐of‐gestation categories were chosen based on their clinical relevance and informed by established definitions from organizations such as the World Health Organization (WHO), Centers for Disease Control and Prevention (CDC), and American College of Obstetricians and Gynecologists (ACOG), which provide widely recognized gestational age subgroups [22, 23, 24].

2.3. Outcome—Postpartum visit attendance

The primary outcome was postpartum visit attendance within 12 weeks postpartum. Postpartum visits were defined as healthcare encounters in obstetrics or family medicine departments within 12 weeks of birth. Healthcare encounters were also included up to 12 weeks after birth if they were associated with the ICD‐10‐CM code indicating postpartum care (Z39).

2.4. Covariate ascertainment

We considered several individual‐level characteristics including age, race and ethnicity, insurance, smoking status, prepregnancy body mass index (BMI), hypertension (chronic hypertension and hypertensive disorders of pregnancy), diabetes (pre‐existing and gestational), parity, cesarean birth, and birth year. Variables of interest were assessed for missingness. When prepregnancy BMI was not available, we subtracted recommended weight gain from conception to the first prenatal visit with a recorded weight to estimate the prepregnancy weight and used this estimated weight to calculate BMI in kg/m² [25]. For the 0.02% with no height available, it was replaced with the average height of the cohort to calculate BMI.

The CDC Social Vulnerability Index (SVI) was utilized as a ZIP code‐level socioeconomic variable in this analysis [26]. SVI comprises variables grouped into four major areas of social vulnerability that are combined into a composite measure: socioeconomic status; household characteristics; racial and ethnic minority status; and housing type and transportation. SVI was divided into quartiles for reporting. We also utilized Rural–Urban Commuting Area (RUCA) codes for this analysis, mapped from patient ZIP codes [27]. RUCA codes 1–3 represent urban residence and RUCA codes 4–10 represent rural residence.

2.5. Statistical analysis

Bivariate analyses examined characteristics associated with not attending postpartum visits, utilizing χ ² tests for categorical variables. We elected to quantify dissimilarities across groups using standardized mean differences since small differences between groups are often statistically significant in studies with large sample sizes [28, 29]. Standardized mean differences are independent of sample size. A standardized mean difference of >0.1 indicates substantial imbalance between groups [30].

We performed multilevel, modified, Poisson regression to estimate relative risks (RRs) and 95% confidence intervals (CIs) of not attending a postpartum visit for each preterm birth category compared to full‐term births (39–40 weeks). Models were adjusted for age, race and ethnicity, insurance, CDC SVI, residence, smoking, BMI, hypertension, diabetes, parity, cesarean birth, and birth year. The primary analysis used complete cases. In sensitivity analyses, we examined postpartum visit attendance for each missing covariate and compared unadjusted RRs of not attending postpartum visits for each gestational age group before and after excluding individuals with missing covariate data.

3. RESULTS

Of the 7,117,929 live births during the study period, 6,464,091 (90.8%) had a valid gestational age documented (Figure 1). Among these births, there were 2,950,130 (46%) unique individuals with a record of prenatal care in Cosmos (expanded cohort). After excluding individuals with missing covariate data, there were 2,403,574 births in the primary analytic cohort (81.5% of the expanded cohort). Of these, 653,552 (27%) parents did not attend a postpartum visit.

FIGURE 1.

Cohort creating flow diagram. BMI, body mass index; RUCA, Rural–Urban Commuting Area; SVI, Centers for Disease Control and Prevention Social Vulnerability Index.

In bivariate analyses, several categories of characteristics had significant differences (standardized mean difference >0.1) in postpartum visit attendance rates including race and ethnicity, smoking, BMI, diabetes, parity, and mode of birth (Table 1). Notably higher rates of not attending a postpartum visit were observed among individuals with the following characteristics: Black non‐Hispanic race and ethnicity (32%), smoking during pregnancy (56%), BMI > 30 kg/m² (30%), no diabetes (28%), multiparity (29%), and no cesarean birth (29%). Over the study period from 2018 to 2024, the proportion of parents who did not attend a postpartum visit decreased by 3% (28% to 25%) (Table S1).

TABLE 1.

Characteristics of individuals who did and did not attend a postpartum visit attendance by 12 weeks, Epic Cosmos, 2018–2024 (n = 2,403,574).

	Postpartum visit		No postpartum visit
	n = 1,750,022 (73%)		n = 653,552 (27%)		Standardized mean difference *
Characteristic	n	(row %)	n	(row %)
Age (years)					0.09
<25	307,523	69	136,591	31
25–<35	1,022,657	74	355,671	26
35+	419,842	72	161,290	28
Race and ethnicity					0.12
Asian, non‐Hispanic	81,526	74	28,107	26
Black, non‐Hispanic	255,831	68	119,795	32
Hispanic	243,763	72	95,399	28
Multiracial/Other	262,615	72	103,284	28
White, non‐Hispanic	906,287	75	306,967	25
Insurance					0.08
Private	1,339,585	74	476,396	26
Public	410,437	70	177,156	30
SVI					0.08
Quartile 1	317,711	74	114,445	26
Quartile 2	375,719	74	129,602	26
Quartile 3	421,476	74	148,572	26
Quartile 4	635,116	71	260,933	29
Residential RUCA codes					0.08
Rural (4–10)	243,043	77	72,518	23
Urban (1–3)	1,506,979	72	581,034	28
Smoking					0.17
Never	1,679,717	73	611,219	27
Smoked before pregnancy	52,194	73	19,176	27
Smoked during pregnancy	18,111	44	23,157	56
Prepregnancy BMI (kg/m2)					0.13
<25	592,125	75	195,703	25
25–<30	486,509	74	166,797	26
30+	671,388	70	291,052	30
Hypertension a					0.09
No	1,287,550	72	506,810	28
Yes	462,472	76	146,742	24
Diabetes a					0.12
No	1,349,495	71	534,548	28
Yes	400,527	77	119,004	23
Parity					0.14
Primiparous	733,449	76	231,025	24
Multiparous	1,016,573	71	422,527	29
Cesarean birth					0.11
No	1,148,515	71	463,035	29
Yes	601,507	76	190,517	24

Abbreviations: BMI, body mass index; RUCA, Rural–Urban Commuting Area; SVI, Centers for Disease Control and Prevention Social Vulnerability Index.

^a Hypertension and diabetes were each a composite of pregestational and gestational.

^*All p values obtained for bivariate analysis of characteristics were <0.001. Standardized difference >0.1 considered substantial differences.

Rates of not attending postpartum visits by length‐of‐gestation category are presented in Table 2 and ranged from 27% after full‐term birth (39–40 weeks) to 34% after a periviable preterm birth (22–23 weeks). In unadjusted models, compared to full‐term birth (39–40 weeks), all lengths of gestation <39 weeks were significantly associated with not attending a postpartum visit. After multivariable adjustment, every length‐of‐gestation category, including 41–43 weeks, was associated with significantly higher risk of not attending a postpartum visit compared to full‐term birth (Figure 2). The highest adjusted risk of not attending a postpartum visit was after periviable preterm birth (22–23 weeks) (aRR, 1.21; 95% CI, 1.13–1.29).

TABLE 2.

Postpartum visit attendance after various lengths of gestation, Epic Cosmos, 2018–2024 (n = 2,403,574).

	Postpartum visit		No postpartum visit
	n = 1,750,022 (73%)		n = 653,552 (27%)		Standardized mean difference
	n	(row %)	n	(row %)
Length of gestation (weeks)					0.03
22–23	1119	66	566	34
24–27	6935	72	2744	28
28–31	15,462	72	5910	28
32–33	46,888	71	18,860	29
34–36	105,559	71	42,238	29
37–38	523,344	73	198,033	27
39–40	968,886	73	355,293	27
41–43	81,829	73	29,908	27

FIGURE 2.

Unadjusted and adjusted risk ratios (95% confidence intervals) of no postpartum visit by 12 weeks in Epic Cosmos, 2018–2024 (n = 2,403,574), after various lengths of gestation compared to 39–40 weeks. Covariates included age, race and ethnicity, insurance, residential Centers for Disease Control and Prevention Social Vulnerability Index, rural versus urban residence, smoking, body mass index, hypertension, diabetes, parity, cesarean birth, and birth year.

In the expanded, sensitivity analysis cohort, in addition to all the same variables in the primary analysis, insurance and hypertension differed significantly with respect to rates of not attending postpartum visits (Table S2). Rates of postpartum visit attendance by length of gestation at birth were similar to the primary analysis with the highest rate (34%) of not attending visits among those with periviable preterm birth (22–23 weeks) (Table S3). In addition, unadjusted RRs were similar in the primary analysis and sensitivity analysis (Table S4).

4. DISCUSSION

In a large, nationwide electronic medical record database, we found that lengths of gestation, both shorter and longer than 39–40 weeks, were all associated with not attending postpartum visits. The highest risk was after periviable preterm birth (22–23 weeks).

Our findings are largely consistent with prior studies of postpartum care receipt. In 2025, we analyzed data from two hospitals in our Philadelphia health system, and while we reported a higher overall proportion of individuals not attending postpartum visits (32%) in that study compared to this study's rate of 27%, we also found that the earlier the preterm birth, the higher the risk, with the highest risk after birth at <24 weeks’ gestation [20]. In another 2020 study of 594,888 mother–infant dyads with Medicaid coverage from 12 US states, preterm birth was associated with lower odds of attending any adult preventative visit in the year following birth [31]. Our finding that even after early‐term (37–38 weeks) and late‐term to post‐term (41–43) births, parents are less likely to attend postpartum visits was inconsistent with our hypothesis, but may be related to the subtly increased risk of neonatal morbidity in these groups [32, 33].

Not all studies have found an association of preterm birth with not attending postpartum visits. A 2007 study of over 4000 patients in Los Angeles County reported that while low income, public or no insurance, and insufficient prenatal care were all associated with not attending a postpartum visit, preterm birth was not [34]. Those findings may have differed from ours due to differences in the underlying population, which included individuals without prenatal care, or their smaller sample size. In our study, among parents who gave birth to infants at 22–23 weeks of gestation, 34% did not attend a postpartum visit, an absolute difference of 7% and relative risk of 1.21, compared to parents of term infants. Although the absolute difference in the proportion of patients not attending a postpartum visit after periviable and term births may appear small, this should be interpreted cautiously. These postpartum individuals have a higher risk of comorbidities than parents of term infants and also may be at particularly high risk of mental health challenges given their infants are critically ill. In addition, our cohort included only individuals who received prenatal care. This approach may underestimate the total number of parents who did not attend postpartum visits. The differences in postpartum visit attendance between individuals with preterm and term births may be larger in the general population.

Qualitative literature highlights that parents may not attend postpartum visits due to competing priorities, as well as lack of access, desire, and recognition of the benefits of accessing postpartum care [34]. In addition to these reasons, parents of preterm infants have specific barriers to postpartum care access. Studies suggest that they desire to spend as much time as possible at their infants’ bedside in the NICU, instead of addressing their own healthcare needs [9, 18, 19]. Additional reasons parents of hospitalized infants in the NICU report choosing not to access postpartum care include not having health concerns or being too busy [35]. NICU parents are also more likely than parents of well infants to face socioeconomic disadvantages that make accessing postpartum care more difficult [36, 37]. In addition, preterm infants and periviable infants are at increased risk of neonatal death [38]. In the setting of infant loss, grief and psychological distress could affect parental desire to interact with the healthcare system and dissuade postpartum visit attendance, despite the continued risk for maternal morbidity [6, 39].

The leading causes of postpartum death are cardiovascular conditions and mental health conditions [4, 40, 41]. Cardiovascular complications (including hypertensive disorders of pregnancy, cardiomyopathy, cerebrovascular accident) account for 30% to 40% of deaths within a year of pregnancy. Mental health conditions such as suicide, substance use disorder, and drug overdose account for 22.7% of deaths. These are potentially preventable conditions that could be addressed in a postpartum visit. Notably, parents of preterm infants have a higher risk than parents of term infants of both cardiovascular disease (2.8 fold) and mental health (three‐ to fourfold) disorders [8, 9]. Postpartum care provides opportunities for early recognition and management of these medical conditions that can lead to significant maternal morbidity or mortality. A 2014 study of 115,502 individuals found that parents of infants born at 23–27 weeks were more than nine times as likely to have severe maternal morbidity than parents of infants born at 39–40 weeks [42]. Furthermore, there are large racial disparities in preterm birth and maternal morbidity and mortality; Black individuals have three times the risk of both extreme preterm birth (<28 weeks) and maternal mortality compared to White individuals [3, 43]. Focusing on postpartum care provision for parents of preterm infants may help narrow maternal health disparities.

While the proportion of postpartum individuals not attending postpartum visits subtly decreased throughout the study period, a quarter still did not receive care. Innovative ways to administer this care are needed. Pediatric settings such as outpatient clinics or inpatient floors and NICUs could welcome postpartum clinicians or facilitate telemedicine [44]. A study of mother–infant dyads demonstrated that parents were more likely to attend their infant's pediatric visit over their own healthcare visit [31]. Currently, there is a movement for pediatricians to screen for postpartum depression during well‐child visits, which is reimbursed by Medicaid [45]. Successful screening for tobacco use, multivitamin supplementation with folic acid, and contraception use has also been demonstrated in well‐child visits [46]. As such, there is precedent to integrate some aspects of parental healthcare in the pediatric setting that could be extended to include other components of postpartum visits like hypertension screening. Additionally, considering the long hospitalizations of infants born moderately preterm and earlier, NICUs may be well‐positioned to facilitate postpartum care delivery. Recent data from our pilot randomized controlled trial of doula‐coordinated and midwife‐delivered postpartum care in the NICU demonstrate the feasibility and effectiveness of co‐located health services for postpartum parents [18, 47].

One strength of our study is the dedicated analysis of a particularly high‐risk population. Parents of preterm infants are understudied [48]. This is the first study to use the large Epic Cosmos dataset to study postpartum visit attendance. Cosmos is a diverse, nationwide dataset that incorporates individual‐ and community‐level covariates mapped from electronic health records. These features support the generalizability of our findings and their relevance to national maternal healthcare delivery. One limitation in this study is that if a postpartum visit did not occur in an Epic Cosmos contributing site, the visits would not be captured and a patient would be determined to not have a postpartum visit. To mitigate this, we restricted our analysis to births with prenatal visits in Cosmos to increase the likelihood that we would capture a postpartum visit, if it occurred, as most obstetric patients receive prenatal and postpartum care in the same practice. We speculate that our decision only to include individuals who received prenatal care in Cosmos may underestimate the total number of parents who did not attend postpartum visits as others have shown that low prenatal care attendance is associated with not receiving postpartum care [34]. Cosmos is not a population‐based dataset, but relies on data from health systems that are Epic customers and have elected to participate in Cosmos. Nonetheless, Cosmos captures a large number of births; vital statistics reveal that there were 25,680,960 births in the United States born at ≥ 20 weeks’ gestation during the study period [49]. Cosmos had at least some data on 28% of these births and sufficient data for this analysis for 9.4% of all US births. In addition, electronic health record data are imperfect and can have missing or erroneous data. For example, not all individuals had a prepregnancy weight recorded, thus prepregnancy BMI categories could have had some misclassification. While we mitigated missingness by using the earliest recorded weight during pregnancy to estimate a prepregnancy BMI, true prepregnancy BMI was unknown. Nonetheless, we are reassured that findings in unadjusted and adjusted models were similar, and analogous to those from our single‐center study [20].

5. CONCLUSION

We found that various lengths of gestation, both shorter and longer than 39–40 weeks, are significantly associated with not attending a postpartum visit, with periviable preterm births at the highest risk. Our findings highlight the urgent need for innovative ways to address postpartum care gaps among all parents, but especially parents of extremely preterm infants. Interdisciplinary, innovative strategies are needed. Investigation into the deployment of healthcare navigators, community health workers, doulas, among others, who can collaborate with maternal healthcare clinicians during the postpartum period and across different healthcare settings may reveal promising approaches. These collaborations are likely to enhance and increase postpartum visit attendance for high‐risk parents and improve maternal health outcomes.

AUTHOR CONTRIBUTIONS

All authors participated in the design of the study and interpretation of the data. Evan Miller performed the data analysis. Kristan Scott drafted the manuscript. All authors revised the manuscript for key intellectual content. All authors read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHIC STATEMENT

The Children's Hospital of Philadelphia Institutional Review Board deemed this non‐human subjects research per federal regulation 45 CFR 46.104.

Supporting information

Supporting information