Documented In-Hospital Lactation Support and Exclusive Breastfeeding During the Postpartum Stay

Key Points

Question

How is hospital-based lactation support associated with exclusive breastfeeding during the hospital stay?

Findings

In this cross-sectional study of electronic health records of 22 857 patients who intended to breastfeed, more documented in-hospital breastfeeding support was associated with the probability of exclusive breastfeeding. Documentation of breastfeeding support and probability of exclusive breastfeeding was associated with patient demographic and clinical characteristics, with significant differences by feeding intention, race and ethnicity, interpreter need, age, marital status, payer, body mass index, current smoking, birth method, gestational age, and low birth weight.

Meaning

This study suggests that ensuring equitable in-hospital postpartum lactation care may close gaps in unequal breastfeeding outcomes.

Abstract

Importance

Exclusive breastfeeding (EBF) during the postpartum hospital stay is associated with longer breastfeeding duration. Electronic health record (EHR) data can be used to uncover patterns in missed and inequitable breastfeeding care that may be associated with persistent breastfeeding disparities.

Objectives

To assess the association of documented postpartum in-hospital breastfeeding care with EBF and test differences in documented care by feeding intention and demographic and clinical characteristics.

Design, Setting, and Participants

This cross-sectional study of EHR data from a single academic hospital network in Minnesota included 22 857 patients who intended to provide breastmilk and delivered singleton infants born at 37 weeks’ or more gestation between January 1, 2018, and December 31, 2022. Data preparation was completed between September 2023 and May 2024; analyses were completed between June and December 2025.

Exposures

Ratio of the number of documented lactation support events per 8-hour nursing shift to the number recommended during the postpartum stay (breastfeeding support ratio [BSR]). Current guidelines recommend a BSR of 1.00 or more.

Main Outcomes and Measures

Multivariable logistic regression with restricted cubic splines was used to estimate the probability of EBF during the hospital stay and to estimate the relative risk (RR) of a BSR less than 1.00.

Results

The sample included 22 857 dyads (mean [SD] maternal age, 31.0 [5.1] years; mean [SD] gestational age, 39.4 [1.1] weeks), with feeding intentions of 91.1% EBF (n = 20 826) and 8.9% breastmilk and formula (n = 2031). There was a nonlinear positive association between BSR and the probability of EBF, with a probability of 0.49 (95% CI, 0.44-0.54) for a BSR of 0.00, a probability of 0.78 (95% CI, 0.76-0.80) for a BSR of 1.00, and a probability of 0.91 (95% CI, 0.90-0.92), with a plateau in the association above a BSR of 2.00. The positive association of BSR with EBF was weaker, with a higher RR of a BSR less than 1.00, among American Indian or Alaska Native patients (adjusted RR, 1.94; 95% CI, 1.26-2.62); Asian, Native Hawaiian, and Other Pacific Islander patients (adjusted RR, 2.35; 95% CI, 2.08-2.62); and Black or African American patients (adjusted RR, 1.83; 95% CI, 1.60-2.07) than among White patients. There were further disparities in documented breastfeeding support identified, with higher risk of a BSR less than 1.00, by breastfeeding intention (both breastmilk and formula: adjusted RR, 4.65; 95% CI, 4.25-5.05), need for an interpreter (adjusted RR, 1.52; 95% CI, 1.32-1.72), age younger than 25 years (adjusted RR, 1.33; 95% CI, 1.14-1.51), marital status (single: adjusted RR, 1.34; 95% CI, 1.21-1.48), health insurance type (public health insurance: adjusted RR, 1.25; 95% CI, 1.06-1.45), body mass index of 30 or more (adjusted RR, 1.40; 95% CI, 1.25-1.56), current tobacco smoking (adjusted RR, 2.09; 1.72-2.47), cesarean delivery (adjusted RR, 1.16; 95% CI, 1.04-1.28), gestational age of 37 to 38 weeks (adjusted RR, 1.26; 95% CI, 1.14-1.38), and low-birth-weight status (<2500 g; adjusted RR, 1.87; 95% CI, 1.42-2.32).

Conclusions and Relevance

This cross-sectional study found that, in a large EHR-based postpartum patient sample, the frequency of documented breastfeeding support was positively associated with EBF during the hospital stay, an early predictor of breastfeeding duration. However, there were numerous disparities in the level of support documented. More frequent in-hospital breastfeeding support, above currently recommended levels, could help reduce gaps in breastfeeding outcomes.

This cross-sectional study uses electronic health record data to assess the association of documented postpartum in-hospital breastfeeding care with exclusive breastfeeding and to test differences in care by feeding intention and demographic and clinical characteristics.

Introduction

In the US, national breastfeeding goals and best practices such as the Baby-Friendly Hospital Initiative’s Ten Steps to Successful Breastfeeding emphasize providing evidence-based breastfeeding support for infants and their parents during the postpartum hospital stay.^1,2,3 Best practices, however, might not be implemented equitably, as evidenced by continued disparities in breastfeeding rates.^4,5 These disparities align with health hazards that disproportionately affect non-Hispanic Black infants and communities of lower socioeconomic status, including higher rates of maternal pregnancy-related deaths,⁶ cardiovascular disease, certain types of cancer, childhood asthma, ear infections, gastrointestinal illness, and infant death.⁷

Although breastfeeding disparities in the US are well described, little empirical evidence exists to evaluate the association between the frequency of in-hospital breastfeeding support and breastfeeding outcomes or to quantify the optimal frequency of documented in-hospital breastfeeding support. Insufficient or inequitably provided breastfeeding support can be contextualized as missed care—any aspect of required patient care that is delayed, incomplete, or omitted^8,9 (eFigure 1 in Supplement 1). Missed breastfeeding care has been associated with patients’ race and ethnicity,¹⁰ spoken language, educational level, age, weight, insurance status, parity, and breastfeeding intention,⁸ and there is an inverse association between exclusive breastfeeding (EBF) and the frequency of missing immediate skin-to-skin contact, missing breastfeeding within 1 hour after birth, and missed care overall.¹¹ Electronic health records (EHRs) offer detailed data on breastfeeding care during the postpartum hospital stay,¹² providing an opportunity to explore missed care and the implementation of best practices in breastfeeding support at the health care system level.

This study analyzed EHR flowsheet data collected during the postpartum hospital stay from a large, socioeconomically and geographically diverse sample of lactating parents and their infants to (1) describe disparities in breastfeeding intensity and documented breastfeeding support; (2) examine the association between documented breastfeeding support and EBF during the hospital stay, including identifying the amount of documented support associated with EBF; and (3) evaluate whether patient characteristics modify the association of documented lactation support and EBF during the hospital stay. Findings cast light on gaps in hospital-based breastfeeding support and the association with early breastfeeding outcomes in healthy, full-term singleton infants.

Methods

The study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.¹³ The University of Minnesota institutional review board determined that this secondary analysis did not meet the criteria for human participants research and was exempt from institutional review board review.

Data Collection and Preparation

This cross-sectional secondary analysis uses EHR data from an academic health center data repository¹⁴ that includes Epic (Epic Systems Corporation) data from 6 urban and 3 rural hospitals providing perinatal care in Minnesota. We included records from patients who did not opt out of research and who delivered infants born at full term (≥37 weeks’ gestation) during the 5-year period between January 1, 2018, and December 31, 2022 (n = 43 876), as well as their infants’ records (n = 45 654). We excluded infants without a parent match (n = 1682) or from multiple gestations (n = 3026), patients discharged from units other than post partum (n = 622), records where no delivery occurred (n = 2) or there was no demographic information (n = 1), infants transferred to the neonatal intensive care unit (n = 2835), and patients with a postpartum length of stay of less than 12 hours or more than 120 hours (5 days) (n = 212). For patients with more than 1 birth during the period, 1 infant was randomly selected and retained in the dataset (6503 siblings excluded). Finally, we excluded infants with no feedings recorded (n = 18) and patients who intended to feed only formula (n = 1670) or had undecided or undocumented feeding intention (n = 8758). In total, 22 857 dyads were analyzed (eFigure 2 in Supplement 1). Birthing parents’ demographic information was collected by self-report or data entry by hospital employees. Race and ethnicity categories used in analysis included American Indian or Alaska Native, non-Hispanic; Asian, Native Hawaiian, or Other Pacific Islander, non-Hispanic; Black or African American, non-Hispanic; Hispanic or Latino; multiracial, non-Hispanic; and White, non-Hispanic. Race and ethnicity data were collected by the hospital system as part of their normal collection of demographic data. We summarize the demographic and clinical variables in the eTable in Supplement 1.

Primary Outcomes

Breastfeeding Intensity and Infant-Feeding Outcomes

Feeding method and breastfeeding intensity were determined by dividing the number of breastmilk feedings by the total number of feedings during the hospital stay, resulting in a number between 0.00 and 1.00 for each patient. EBF during the hospital stay, a key outcome measured by the Centers for Disease Control and Prevention, is represented by a breastfeeding intensity of 1.00. Infants were considered to have initiated breastfeeding if any feedings of breastmilk were documented.

Documentation of Breastfeeding Care Provided

Guidelines from the American Academy of Pediatrics¹⁵ and the Academy of Breastfeeding Medicine¹² state that breastfeeding newborns should have at least 1 breastfeeding session observed and documented by a health professional trained in breastfeeding assessment during each nursing shift and within 8 hours prior to discharge. For this analysis, documented breastfeeding support was compared against a goal of 1 instance of breastfeeding support being provided per 8-hour nursing shift. The breastfeeding support ratio was calculated by dividing the number of instances of documented breastfeeding support of any length or quality by the expected number of instances of support for the hospital stay (ie, 3 support interactions per 24 hours). A support ratio of 1.00 or more indicates that the support goal was met; a support ratio of less than 1.00 indicates that care was missed. On the final day of the dyad’s stay, we expected 1 instance of support per 8 hours admitted.

Statistical Analysis

Data preparation and analysis were performed within a university-administered data shelter using R, version 4.4.0 (2024-04-24-ucrt; R Project for Statistical Computing), and SAS software, version 9.4 (SAS Institute Inc). Data were not deidentified. Data were prepared between September 2023 and May 2024; analyses were completed between June and December 2025. Further details about the methods used to prepare data for analysis are provided in the eMethods in Supplement 1 and Gallagher et al.¹⁶ R package gtsummary¹⁷ was used for descriptive statistics (aim 1), rms¹⁸ was used for multivariable logistic regression (lrm) and predictive modeling with restricted cubic splines (aims 2 and 3), and ggplot2¹⁹ was used for data visualization. Relative risks (RRs) and 95% CIs were calculated with SAS software using proc logistic with the NLMEANS macro (aim 3). Group differences were tested using the χ² test of independence or the Fisher exact test for categorical variables and Kruskal-Wallis test for continuous variables. P values were from 2-sided tests, with P < .05 considered statistically significant, and we applied the Benjamini-Hochberg method to control for the false discovery rate at .05 for multiple comparisons.

Regression models were sequentially covariate adjusted for feeding intention (breastmilk or breastmilk and formula); clinical variables (body mass index [calculated as weight in kilograms divided by height in meters squared], current smoking, parity, delivery method, gestational age, and low-birth-weight status); and demographic variables (urban or rural hospital, race and ethnicity, interpreter needed, age, marital status, and insurance type). To test effect modification, an interaction term was added in the fully adjusted model to estimate the probability of EBF for each covariate. Analyses included missing variables coded as “not documented.” Educational level, a factor associated with EBF, was not reliably recorded in the EHR; 92.6% of cases (n = 21 168) lacked documentation, and this variable was excluded from analysis and reporting. Documentation of the time to first breastfeeding was included in the support ratio calculations and thus was not included as a variable to reduce collinearity.

Results

Descriptive Statistics (Aim 1)

The sample included 22 857 dyads (mean [SD] maternal age, 31.0 [5.1] years; mean [SD] gestational age, 39.4 [1.1] weeks; 181 American Indian or Alaska Native, non-Hispanic [0.8%]; 2689 Asian, Native Hawaiian, or Other Pacific Islander, non-Hispanic [11.8%]; 2373 Black or African American, non-Hispanic [10.4%]; 846 Hispanic or Latino [3.7%]; 278 multiracial, non-Hispanic [1.2%]; 15 275 White, non-Hispanic [66.8%]; and 1215 race or ethnicity not documented [5.3%]), with feeding intentions of 91.1% EBF (n = 20 826) and 8.9% breastmilk and formula (n = 2031) (Table 1). Most patients in this study who identified as Black or African American were born outside of the US (1463 of 2373 [61.7%]). Overall descriptive statistics of the sample, including breastfeeding intensity and support ratio by demographic and clinical characteristics, are shown in Table 1. The rate of any breastfeeding during the hospital stay was 98.3% (n = 22 461), while the rate of EBF was 71.0% (n = 16 227). Asian, Native Hawaiian, or Other Pacific Islander patients had the lowest breastfeeding intensity (mean [SD], 0.72 [0.35]), with an EBF rate of 44.0% (1184 of 2689). The mean (SD) support ratio for all patients was 2.32 (0.98), more than double the recommended support goal (ie, a support ratio ≥1.00) (Table 1). Patients who gave birth in rural hospitals were less likely than those in urban hospitals to exclusively breastfeed (2537 of 3863 [65.7%] vs 19 887 of 29 422 [67.6%]); however, rural patients had higher documented breastfeeding support than urban patients (mean [SD] support ratio, 2.66 [1.10] vs 2.24 [1.00]. The mean (SD) support ratio for patients who exclusively provided breastmilk was significantly greater than that for patients who provided breastmilk and formula (2.55 [0.88] vs 1.71 [0.97]; P < .001). Of the 396 patients who provided only formula, none had a support ratio over 1.00. White, non-Hispanic patients had the highest support ratio (mean [SD], 2.45 [0.96]), and Black or African American, non-Hispanic patients had the lowest support ratio (mean [SD], 1.78 [0.95]).

Table 1. Breastfeeding Intensity and Support Ratio by Feeding, Demographic, and Birth Characteristics^a.

Characteristic	Overall, No. (%) (N = 22 857)	Mean (SD)b
		Breastfeeding intensity	Support ratio
Support ratio and feeding method
Support ratio for the entire hospital stay	NA	NA	2.32 (0.98)
Breastfeeding intensity (breastmilk feedings divided by total feedings)	NA	0.89 (0.24)	NA
Feeding intention
Breastfeeding	20 826 (91.1)	0.92 (0.18)	2.38 (0.96)
Breastmilk and formula	2031 (8.9)	0.52 (0.36)	1.27 (1.06)
Feeding method during hospital stay
Breastmilk only	16 227 (71.0)	1.00 (0.00)	2.55 (0.88)
Breastmilk and formula	6234 (27.3)	0.65 (0.26)	1.71 (0.97)
Formula only	396 (1.7)	0.00 (0.00)	0.40 (0.21)
Demographic characteristics
Rurality
Urban hospital	20 626 (90.2)	0.89 (0.24)c	2.24 (1.00)
Rural hospital	2231 (9.8)	0.91 (0.21)c	2.66 (1.10)
Race and ethnicity
American Indian or Alaska Native, non-Hispanic	181 (0.8)	0.79 (0.31)	2.02 (1.15)
Asian, Native Hawaiian, or Other Pacific Islander, non-Hispanic	2689 (11.8)	0.72 (0.35)	1.84 (1.12)
Black or African American, non-Hispanic	2373 (10.4)	0.74 (0.29)	1.78 (0.95)
Hispanic or Latino	846 (3.7)	0.87 (0.22)	2.15 (0.97)
Multiracial, non-Hispanic	278 (1.2)	0.87 (0.25)	2.22 (1.01)
White, non-Hispanic	15 275 (66.8)	0.95 (0.16)	2.45 (0.96)
Not documented	1215 (5.3)	0.86 (0.24)	2.14 (0.97)
Interpreter needed
No	21 119 (92.4)	0.90 (0.22)	2.32 (1.01)
Yes	1674 (7.3)	0.69 (0.32)	1.68 (0.97)
Not documented	64 (0.3)	0.91 (0.21)	2.35 (0.92)
Age, y
<25	2520 (11.0)	0.83 (0.29)	2.07 (1.05)
25-29	5743 (25.1)	0.88 (0.25)	2.28 (1.03)
30-34	8831 (38.6)	0.91 (0.22)	2.34 (1.01)
35-39	4793 (21.0)	0.90 (0.22)	2.28 (0.99)
≥40	970 (4.2)	0.88 (0.23)	2.18 (0.96)
Marital status
Married or partnered	17 015 (74.4)	0.91 (0.21)	2.35 (0.99)
Single	5723 (25.1)	0.83 (0.29)	2.06 (1.06)
Not documented	119 (0.5)	0.86 (0.25)	2.32 (1.01)
Insurance type
Private	5983 (26.2)	0.94 (0.17)	2.44 (0.96)
Public	3318 (14.5)	0.80 (0.29)	1.99 (1.02)
Not documented	13 556 (59.3)	0.89 (0.24)	2.28 (1.03)
Clinical characteristics
BMI
<30	10 017 (43.8)	0.90 (0.23)	2.33 (1.02)
≥30	4035 (17.7)	0.85 (0.27)	2.17 (1.07)
Not documented	8805 (38.5)	0.89 (0.22)	2.26 (0.98)
Current tobacco smoking
No	21 557 (94.3)	0.89 (0.23)	2.29 (1.01)
Yes	737 (3.2)	0.80 (0.32)	2.02 (1.14)
Not documented	563 (2.5)	0.90 (0.22)	2.27 (1.00)
First baby
Yes	7989 (35.0)	0.89 (0.24)d	2.29 (1.02)d
No	11 647 (51.0)	0.89 (0.23)d	2.28 (1.01)d
Not documented	3221 (14.1)	0.88 (0.24)d	2.25 (1.04)d
Delivery method
Vaginal	15 197 (66.5)	0.89 (0.24)	2.30 (1.04)
Cesarean	4987 (21.8)	0.88 (0.23)	2.17 (0.94)
Not documented	2673 (11.7)	0.89 (0.23)	2.33 (0.98)
Gestational age, wk
37-38	6267 (27.4)	0.87 (0.25)	2.18 (1.03)
39-40	14 371 (62.9)	0.90 (0.23)	2.32 (1.01)
≥41	2219 (9.7)	0.89 (0.23)	2.27 (0.97)
Low birth weight (<2500 g)
No	22 489 (98.4)	0.89 (0.23)	2.29 (1.02)
Yes	368 (1.6)	0.76 (0.30)	1.73 (0.96)
Time to first breastfeed, h
<1	14 104 (61.7)	0.93 (0.17)	2.45 (0.95)
1-2	5772 (25.3)	0.91 (0.19)	2.29 (0.89)
>2	2129 (9.3)	0.77 (0.31)	1.74 (1.00)
Not documented	852 (3.7)	0.41 (0.46)	0.75 (1.18)
Infant weight loss at discharge >7%
No	17 884 (78.2)	0.87 (0.25)	2.23 (1.06)
Yes	4926 (21.6)	0.95 (0.12)	2.45 (0.80)
Infant length of stay, mean (SD), h	47 (16)	NA	NA
Time to first formula feeding, mean (SD), h	3.5 (5.6)	NA	NA

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); NA, not applicable.

^a Lactation support is represented by the support ratio, calculated by dividing the number of breastfeeding support encounters documented during the hospital stay by the number of expected support encounters. A support ratio of 1.00 indicates that the minimum recommended level of support was met.

^b P < .001.

^c Urban vs rural breastfeeding intensity, P = .002.

^d First baby yes vs no vs not documented breastfeeding intensity and support ratio, P = .20.

Association of Documented Breastfeeding Support and Probability of EBF (Aim 2)

A nonlinear, positive association was found between support ratio and the probability of EBF during the hospital stay, which remained after adjusting for clinical and demographic characteristics and feeding intention (Figure 1). Patients with a support ratio of 0.00 had a probability of EBF of 0.49 (95% CI, 0.44-0.54), those with a support ratio of 1.00 had a probability of 0.78 (95% CI, 0.76-0.80), and those with a support ratio of 2.00 had a probability of 0.91 (95% CI, 0.90-0.92); the probability of EBF plateaued above a support ratio of 2.00 (Table 2). Feeding intention was a statistically significant factor associated with EBF until a support ratio of approximately 1.50 was achieved. The interactions of support ratio with feeding intention, race and ethnicity, insurance coverage, and low-birth-weight status demonstrated statistical significance after false discovery rate adjustment (Figure 2).

Figure 1. Line Graph Showing Association Between Support Ratio and Probability of Exclusive Breastfeeding.

Association between documented postpartum lactation support, defined as the number of inpatient lactation support interactions per 8-hour nursing shift (support ratio), and the probability of exclusively breastfeeding during the hospital stay among 22 857 patients who intended to provide breastmilk to their infants. The dashed vertical line represents the current recommended amount of breastfeeding support (support ratio of 1.00). Models: unadjusted—no adjustment; clinical—adjusted for body mass index, current smoking, parity, delivery method, gestational age, and birth weight less than 2500 g; demographic and clinical—adjusted for clinical characteristics plus urban or rural hospital, race and ethnicity, interpreter needed, age, marital status, and insurance type; fully adjusted—adjusted for demographic and clinical variables, plus feeding intention.

Table 2. Comparison of Covariate-Adjusted Models of the Probability of Exclusive Breastfeeding During the Postpartum Hospital Stay, by Support Ratio.

Support ratio	Probability of exclusive breastfeeding (95% CI)a
	Clinical and demographic modelb	Fully adjusted modelc
0.00	0.40 (0.36-0.45)	0.49 (0.44-0.54)
0.50	0.54 (0.56-0.63)	0.65 (0.62-0.68)
1.00 (Current minimum recommendation)	0.75 (0.73-0.78)	0.78 (0.76-0.80)
1.50	0.86 (0.84-0.87)	0.86 (0.85-0.88)
2.00	0.91 (0.87-0.92)	0.91 (0.90-0.92)
2.50	0.93 (0.93-0.94)	0.94 (0.93-0.94)
3.00	0.95 (0.94-0.96)	0.95 (0.95-0.96)

^a Probabilities are based on sequentially covariate-adjusted models that included clinical and demographic characteristics as well as clinical and demographic characteristics plus feeding intention.

^b Clinically and demographically adjusted for urban or rural hospital, race and ethnicity, interpreter needed, age, marital status, insurance type, body mass index, current smoking, parity, delivery method, gestational age, and birth weight less than 2500 g.

^c Fully adjusted for clinical and demographic variables, plus feeding intention.

Figure 2. Line Graphs Showing Association Between Support Ratio and Probability of Exclusive Breastfeeding, by Feeding Intention, Race and Ethnicity, Insurance Type, and Low-Birth-Weight Status.

The association of documented inpatient postpartum lactation support and the probability of exclusively breastfeeding during the hospital stay differed by 4 maternal demographic and clinical characteristics (P < .001). Estimates are from a restricted cubic spline regression model that was sequentially covariate adjusted by clinical and demographic characteristics and feeding intention, including interaction terms between support ratio and feeding intention (A), race and ethnicity (B), insurance type (C), and low-birth-weight status (D).

Inequities in Documentation of Provision of Minimum Breastfeeding Support (Aim 3)

There were differences in whether patients had the minimum level of breastfeeding support documented (support ratio ≥1.00) by demographic and clinical characteristics (Table 3). Patients who intended to provide both breastmilk and formula had the highest RR of a support ratio less than 1.00 (adjusted RR, 4.65 [95% CI, 4.25-5.05]; P < .001). Patients who identified as American Indian or Alaska Native (adjusted RR, 1.94; 95% CI, 1.26-2.62); Asian, Native Hawaiian, or Other Pacific Islander (adjusted RR, 2.35; 95% CI, 2.08-2.62); and Black or African American (adjusted RR, 1.83; 95% CI, 1.60-2.07) also had significantly higher RR of a support ratio less than 1.00, as were those who needed an interpreter (adjusted RR, 1.52; 95% CI, 1.32-1.72); patients younger than 25 years (adjusted RR, 1.33; 95% CI, 1.14-1.51); single patients (adjusted RR, 1.34; 95% CI, 1.21-1.48); those with public insurance plans (adjusted RR, 1.25; 95% CI, 1.06-1.45); patients with prepregnancy body mass index of 30 or more (adjusted RR, 1.40; 95% CI, 1.25-1.56); current tobacco smokers (adjusted RR, 2.09; 1.72-2.47); and patients delivering by cesarean (adjusted RR, 1.16; 95% CI, 1.04-1.28), delivering at early term (37-38 weeks’ gestation; adjusted RR, 1.26; 95% CI, 1.14-1.38), or delivering a low-birth-weight infant (<2500 g; adjusted RR, 1.87; 95% CI, 1.42-2.32).

Table 3. RR of Having Less Than the Recommended Amount of Documented In-Hospital Postpartum Breastfeeding Support^a.

Characteristic	Univariable			Multivariable		FDR-corrected P value
	RR (95% CI)	P value	RR (95% CI)		P value
Feeding intention
Breastmilk	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Both breastmilk and formula	6.12 (5.69-6.55)	<.001	4.65 (4.25-5.05)		<.001	<.001
Urban or rural hospital
Urban	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Rural	0.74 (0.63-0.85)	<.001	0.90 (0.75-1.05)		.20	.35
Race and ethnicity
American Indian or Alaska Native, non-Hispanic	3.47 (2.47-4.48)	<.001	1.94 (1.26-2.62)		.007	.02
Asian, Native Hawaiian, or Other Pacific Islander, non-Hispanic	3.87 (3.51-4.22)	<.001	2.35 (2.08-2.62)		<.001	<.001
Black or African American, non-Hispanic	3.33 (2.99-3.67)	<.001	1.83 (1.60-2.07)		<.001	<.001
Hispanic or Latino	1.86 (1.49-2.23)	<.001	1.08 (0.83-1.32)		.55	.72
Multiracial, non-Hispanic	1.91 (1.27-2.54)	<.001	1.30 (0.82-1.77)		.22	.36
White, non-Hispanic	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Not documented, non-Hispanic	1.87 (1.55-2.18)	<.001	1.24 (1.00-1.49)		.05	.10
Interpreter needed
No	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Yes	2.74 (2.48-3.00)	<.001	1.52 (1.32-1.72)		<.001	<.001
Not documented	0.69 (0.03-1.35)	.36	0.90 (0.0-1.89)		.84	.98
Age, y
<25	1.80 (1.60-2.01)	<.001	1.33 (1.14-1.51)		.001	.002
25-29	1.23 (1.10-1.35)	<.001	1.07 (0.95-1.19)		.25	.38
30-34	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
35-39	1.04 (0.92-1.16)	.49	1.00 (0.87-1.12)		.95	.98
≥40	1.19 (0.95-1.42)	.12	1.00 (0.78-1.22)		.98	.98
Marital status
Married or partnered	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Single	1.93 (1.78-2.08)	<.001	1.34 (1.21-1.48)		<.001	<.001
Not documented	1.22 (0.56-1.88)	.51	0.96 (0.28-1.63)		.90	.98
Insurance type
Private	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Public	2.70 (2.36-3.04)	<.001	1.25 (1.06-1.45)		.009	.02
Not documented	1.72 (1.53-1.91)	<.001	1.25 (1.10-1.40)		.001	.003
BMI
<30	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
≥30	1.45 (1.31-1.60)	<.001	1.40 (1.25-1.56)		<.001	<.001
Not documented	1.02 (0.93-1.11)	.66	0.98 (0.88-1.08)		.72	.88
Current smoking
No	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Yes	2.03 (1.73-2.34)	<.001	2.09 (1.72-2.47)		<.001	<.001
Not documented	0.97 (0.72-1.22)	.82	0.98 (0.70-1.26)		.89	.98
Delivery method
Vaginal	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Cesarean	1.04 (0.94-1.14)	.41	1.16 (1.04-1.28)		.01	.03
Not documented	0.91 (0.80-1.03)	.14	1.16 (1.00-1.32)		.05	.10
Gestational age, wk
37-38	1.30 (1.19-1.40)	<.001	1.26 (1.14-1.38)		<.001	<.001
39-40	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
≥41	1.07 (0.93-1.22)	.33	1.04 (0.88-1.20)		.63	.80
First baby
No	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Yes	0.99 (0.92-1.26)	.001	1.01 (0.91-1.10)		.88	.98
Not documented	1.13 (1.01-1.26)	.036	1.09 (0.95-1.22)		.22	.36
Low birth weight (<2500 g)
No	1.00 [Reference]	NA	1.00 [Reference]		NA	NA
Yes	2.29 (1.85-2.74)	<.001	1.87 (1.42-2.32)		<.001	<.001

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); FDR, false discovery rate; NA, not applicable; RR, relative risk.

^a RR of having less than the recommended amount of documented in-hospital postpartum breastfeeding support (<1.00 interaction per 8-hour nursing shift), by demographic and clinical characteristics, among 22 857 patients who intended to provide breastmilk in a large US hospital system (2018-2022). Multivariable log-binomial analysis included all the other variables as covariates.

Discussion

Principal Findings

This cross-sectional study used EHR flowsheet data to examine the breastfeeding support documented for patients with a recorded intention to provide breastmilk to their full-term singleton infants at a large hospital system in Minnesota (2018-2022), as well as the dyads’ in-hospital breastfeeding outcomes. Findings reveal a nonlinear positive association between the amount of documented breastfeeding support and the probability of EBF during the hospital stay, independent of feeding intention and demographic and clinical characteristics. Interaction effects were found between patient characteristics and breastfeeding support, as well; even at the same level of lactation support, rates of EBF were lower among higher-risk demographic and clinical groups. Finally, we identified disparities in the frequency of documented breastfeeding support by breastfeeding intention, race and ethnicity, need for an interpreter, age, marital status, health insurance type, current tobacco smoking, delivery method, gestational age, and low-birth-weight status. This research provides evidence that in-hospital breastfeeding support is associated with increased breastmilk provision for all patients, that support is inequitably provided, and that lower EBF rates among multiple patient groups may be due to missed care, possibly associated with health care professionals’ bias. The findings also suggest that the current recommendation for the minimum amount of breastfeeding support provided during the hospital stay (ie, 1 instance of breastfeeding support being provided per shift) is insufficient to support EBF for all patients.

Findings in Context

There are points of discordance and alignment with existing research revealed in this analysis. Consistent with other studies showing lower breastfeeding rates in rural areas,^20,21 patients in this dataset who gave birth in rural hospitals were less likely to exclusively breastfeed. However, in contrast to research demonstrating that rural hospitals are less likely to provide supportive breastfeeding care²² and the significant barriers to reproductive health care faced by rural residents,²³ rural patients in this study had higher documented breastfeeding support than urban patients. Because the hospitals in this analysis were from a single hospital system, there may be system-wide policies in place that codify higher levels of support, staffing ratios that allow more time for support, or community-level differences in breastfeeding education and support.

The exclusive breastfeeding rate of Asian, Native Hawaiian, or Other Pacific Islander patients in this study was 44.0%, in contrast to published estimates of the prevalence of formula supplementation within 2 days post partum between 21.4% and 30.1% for this group.²⁴ This gap may be associated with the large local Hmong population. In a study of Minnesota WIC (Special Supplemental Nutrition Program for Women, Infants, and Children) participants, Hmong patients demonstrated lower EBF rates and shorter breastfeeding duration than non-Hmong Asian WIC participants.²⁵ Similarly, most patients in this study who identified as Black or African American were born outside of the US (1463 of 2373 [61.7%]). Overall, Black immigrant patients are more likely to initiate breastfeeding than African American patients,²⁶ which may mean that breastfeeding initiation rates are overestimated for African American patients in this sample. Future analysis may include further examination of these subgroups to elucidate differences in breastfeeding outcomes and the support received, considering factors such as migrant status and acculturation that may be associated with breastfeeding rates.^27,28

Clinical, Policy, and Research Implications

The decision to breastfeed is a personal one that occurs within a social-ecological system that includes immediate influences as well as institutional and cultural ones.²⁹ These influences may perpetuate not only systems of support and encourage self-efficacy but may compound experiences of bias, racism, and distrust in health care systems.³⁰ Although parents’ intention to breastfeed, high self-efficacy, and positive family attitudes can increase breastfeeding behaviors,^29,31 implicit racial bias and personal bias of health care professionals perpetuate inequity in the provision of breastfeeding support.^32,33,34 In this study, patients identified as American Indian or Alaska Native; Asian, Native Hawaiian, or Other Pacific Islander; and Black or African American were less likely to exclusively breastfeed and had a higher rate of experiencing missed care, which is reflective of the high maternal mortality rate among these groups within the US.⁶ Rather than targeting interventions solely toward patients, it is worthwhile to consider interventions directed toward health care professionals, such as training in trauma-informed care,³⁵ as well as other systemic changes, such as increasing health professionals’ use of interpreter services to improve patients’ health care experiences.³⁶

A crucial aspect of providing breastfeeding care is ensuring that people who wish to breastfeed are supported in doing so.¹² We demonstrate an increased probability of EBF with an increase in documented support, but patients in our study who expressed an intention to provide both breastmilk and formula had the highest RR of having a support ratio less than 1.00 (adjusted RR, 4.65 [95% CI, 4.25-5.05]; P < .001). It is unclear if patients with more documented breastfeeding support were more likely to exclusively breastfeed because of the increased support, exclusively breastfeeding patients were more likely to have breastfeeding support provided and documented, or patients perceived to have a lower intention to breastfeed were given less support (ie, experienced missed care). In this hospital system, patients’ feeding plans may be documented prenatally in a clinical note or as a postpartum flowsheet entry in the parent’s medical record but are not automatically linked to the infant’s medical record. Clear documentation of feeding intention and breastfeeding support offered and declined, and thorough linkage between care plans discussed at prenatal visits and the hospital EHR environment, will allow for continuity of care and improved tracking of longitudinal outcomes.^37,38

The American Academy of Pediatrics and the Academy of Breastfeeding Medicine recommend that all breastfeeding newborns have at least 1 breastfeeding session observed and documented during each nursing shift,^12,15 but the scientific basis of this recommendation is unclear. Our analysis indicates that following the minimum recommendation may not be sufficient to support EBF for all patients. Despite all patients in this dataset intending to breastfeed at least in part, there was a weaker association of lactation support with the probability of EBF among groups at higher risk of not meeting their breastfeeding goals, suggesting a need for a higher level of in-hospital support for patients at risk. Although the EHR did not capture the quality of breastfeeding support and we could not assess it directly, the quality of perinatal care is a known factor associated with postpartum maternal health,³⁹ and culturally responsive, tailored breastfeeding support spanning the perinatal period has a positive association with breastfeeding outcomes.⁴⁰

Strengths and Limitations

This study has some strengths, including the expertise of the research team and the large sample size. The lead researcher has experience providing breastfeeding care and using EHR as a lactation consultant within a large hospital; other experts include a nurse-midwife working within the hospital system; a nurse informaticist, a statistician, and a data analyst skilled in using EHR data for research; a maternal-child health epidemiologist; and a maternal-child health policy expert. Using systematically prepared EHR data resulted in a large sample size, enabling analysis in greater detail, including small groups and rare events, while being robust to outliers.

This study also has some limitations, including its limited ability to evaluate the timing and accuracy of EHR documentation, quality and duration of breastfeeding support provided, and temporality between documented breastfeeding support and breastfeeding behavior. We could not identify whether documented breastfeeding support was provided by registered nurses, nurses with specialized lactation training, or dedicated lactation consultants, and we did not consider who provided perinatal care (ie, physicians or midwives) and how this may have been associated with breastfeeding outcomes. This analysis was also limited by data that did not include information on staffing levels, reasons for formula introduction, or declined care. Flowsheet names were accessed and their data requested using a tool designed for that purpose, which may have limited our ability to receive complete records. Documentation was not uniform between hospitals within this system, and each hospital may have flowsheets, processes, and standards of which we were unaware. The records that comprise this study span the COVID-19 pandemic, and we do not have information on how breastfeeding support practices may have changed during this time.

Conclusions

Unequal lactation care is an underrecognized factor associated with postpartum maternal and infant health disparities. In this cross-sectional study of 22 857 birthing patients intending to breastfeed, greater documented in-hospital breastfeeding support was associated with EBF during the hospital stay, with benefits extending beyond current recommendations and varying across patient groups. These findings suggest that achieving equitable breastfeeding outcomes will require attention to both the quantity and quality of lactation support, along with improved documentation of this care.

Supplement 1.

eMethods.

eTable. Summary of Demographic and Clinical Variables

eFigure 1. Missed Care in Breastfeeding Support Model

eFigure 2. Cohort Development Flowchart

eReferences.

Supplement 2.

Data Sharing Statement