Associations Between Prenatal Intention and Postpartum Choice: Infant Feeding and Contraception Decisions Among Inner-City Women

Abstract

Purpose:

We sought to understand how women's prenatal infant feeding and contraception intentions were related to postpartum choices.

Materials and Methods:

Expectant women ≥14 years of age receiving care at MacDonald Women's Hospital, Cleveland Ohio were previously surveyed regarding feeding and contraceptive intentions. Here, we asked: (1) What were postpartum feeding choices, and did prenatal intention predict postpartum choice?, (2) What were postpartum contraceptive choices, and did prenatal intention predict postpartum choice?, and (3) What was the relationship of postpartum contraceptive choice to postpartum feeding choice?

Results:

Of 223 women interviewed prenatally, 214 (96%) were followed to postpartum in-hospital, and 119 out of 214 (56%) were followed to the postpartum visit. The mean age was 25 years, 185 out of 206 (89.8%) were African American, and 149 out of 200 (75.0%) were multiparous. Prenatal feeding and contraceptive intent were significantly associated with postpartum feeding and contraceptive choices, respectively (both p < 0.0001). More women who initiated breastfeeding chose no contraception (54.5% for any breastfeeding versus 32.2% for exclusive formula feeding) versus long-acting reversible contraception (LARC), tubal ligation, or other contraceptive types (χ² = 9.28, p = 0.03). After adjusting for known confounders, only receipt of other contraceptive types (not LARC, not tubal ligation) was significantly associated with decreased odds of any breastfeeding (p = 0.02).

Conclusions:

Among low-income predominantly African American inner-city women, prenatal intentions were significantly associated with postnatal choices for infant feeding and contraception. After controlling for confounders, women receiving less effective types of contraception (not LARC and not tubal ligation) had reduced odds of any breastfeeding (p = 0.02).

Introduction

The health benefits of exclusive breastfeeding for both mother and child are well documented.^1–6 However, although 83.2% of infants in the United States are breastfed at some point, only 24.9% of infants are exclusively breastfed for the first 6 months of life, as recommended (2015 data).⁷ Appropriate birth-spacing can also reduce infant and child mortality and promote maternal health.^8,9 The World Health Organization (WHO) recommends at least a 2–3-year interval between pregnancies; however, more than 30% of women experience inter-pregnancy intervals of less than 18 months in the United States, and only 14% of women employ a contraceptive method immediately postpartum.^10–12 Postpartum predischarge provision of contraception improves inter-pregnancy intervals, and long-acting reversible contraception (LARC) methods have been shown to substantially improve inter-pregnancy intervals compared with other methods.^13–15 As a newer option with widely varying rates globally, and relatively low uptake among young minority populations in the United States, postplacental intrauterine device (IUD) placement is an ideal option to achieve both breastfeeding and birth spacing goals.

Little is known about the interaction between infant feeding choice and contraceptive choice. There is an overlap between risk factors for not exclusively breast feeding and those for short interval pregnancy, including socioeconomic status, educational level, race and ethnicity, and age.^16,17 Conversely, exclusive breastfeeding is associated with improved inter-pregnancy intervals.¹⁸ Although an increasing number of studies address the impact of differing contraceptive methods on breastfeeding outcomes, there is limited information regarding the interaction between maternal choice of infant feeding method and postpartum contraception.¹² A recent pilot study, aimed at examining whether low-income inner-city expectant women who intend to breastfeed make different contraceptive choices than those who intend to formula feed, found that prenatal contraceptive intentions did not differ significantly between women intending to breastfeed and those intending to formula feed.¹⁹

The actual postpartum choices of these women were not yet known, and hence the primary objective of this study is to report on the choices women made postpartum regarding both infant feeding and contraception. We also sought to understand how women's infant feeding and contraception intentions before delivery were related to their infant feeding and contraceptive choices postpartum. Specific research questions (RQs) were: (1) What was the distribution of postpartum infant feeding choice, and did prenatal feeding intention predict choice in-hospital and at the postpartum visit (PPV)?, (2) What was the distribution of postpartum contraceptive choice, and did prenatal contraceptive intention predict contraceptive choice in-hospital or at the PPV?, and (3) What was the relationship of contraceptive choice to feeding choice, both postpartum in-hospital and at the PPV?

Materials and Methods

Study design, setting, and population

We previously (November 1, 2016–January 15, 2017) surveyed 223 expectant women about infant feeding and contraceptive intentions at University Hospitals Cleveland Medical Center MacDonald Women's Hospital (MWH), a Baby-Friendly designated tertiary care academic institution serving predominantly inner-city publicly insured African American women in its on-site clinics.¹⁹ Reported here, we followed these participants to examine their postpartum choices regarding infant feeding and contraception. Expectant English-speaking mothers 14 years or older who had a confirmed intrauterine pregnancy of any gestational age were eligible for the initial survey study. University Hospitals Cleveland Medical Center Institutional Review Board approved the study, and informed consent was obtained from all participants (and guardian if <18 years old).

Study conduct and data collection

At MWH, routine prenatal care includes informing all women about the benefits and management of breastfeeding as described in Step 3 of the Baby-Friendly Hospital Initiative guidelines. Prenatal providers include resident and attending physicians and nurse midwives, and counseling regarding contraception is individualized; details were not recorded as part of the initial study. The postpartum order set includes the contraceptive options of a postplacental hormonal IUD and the Medroxyprogesterone acetate injection. Copper IUDs and the implant are not standardly available postpartum.

The initial prenatal research interview (previously reported) included socio-demographic information and questions about knowledge, beliefs, prenatal intentions, and attitudes about contraception and breastfeeding.¹⁹ It included a question about personal priorities in which women were asked to “Put the following statements in the order of how important they are to you, with 1 being most important and 3 being least important”: “Not getting pregnant right away,” “Breastfeeding my baby,” and “Being able to return to work/school as soon as possible.” For this study, we reviewed the electronic medical records of the initial study participants, and we recorded their postpartum infant feeding and contraception choices during the postpartum in-hospital stay and at the PPV. Not all surveyed women delivered at the “home hospital” because other birthing hospitals are located in close geographic proximity. Findings were entered into an REDCap database (Research electronic data capture), a secure, web-based application designed to support data capture for research.²⁰

Data collection and outcome measures

The primary outcome measures were postpartum infant feeding choice and maternal contraceptive choice. Infant feeding choice was characterized as exclusive breastfeeding, mixed feeding, or exclusive formula feeding. “Exclusive breastfeeding” was defined as breast milk only, including feeding at the breast or expressed breast milk with use of vitamin D drops but no water or formula. “Mixed feeding” was defined as feeding both breast milk and formula. “Any breastfeeding” included both exclusive breastfeeding and mixed feeding. “Exclusive formula feeding” was defined as formula milk only.

Postpartum in-hospital contraceptive choices were recorded, and they included “bridge” to the PPV, as well as “none” and “will discuss with provider later.” For the analysis, contraceptive choice was characterized as no contraceptive method versus LARC (LARC includes IUDs and the implant [Nexplanon^®]) versus tubal ligation versus all other contraceptive choices. Tubal ligation emerged as a unique category due to the unexpectedly large proportion of women (10.3%) who chose this method. LARC was defined as a unique category because LARC choice versus nonchoice was integral to the initial study RQ. Although Medroxyprogesterone acetate injection can be considered an LARC, we and others classify it separately because Medroxyprogesterone acetate injection has a clinically meaningful shorter duration of action (13 weeks versus years) compared with IUDs and the implant, and users have significantly lower rates of method continuation compared with both IUDs and the implant.^21,22 In addition, relevant to this study, controversy exists regarding Medroxyprogesterone acetate injection's impact on breastfeeding, with the WHO and CDC recommendations, aligned for IUDs and the implant, diverging regarding immediate postpartum Medroxyprogesterone acetate injection for breastfeeding mothers.^15,23

Other descriptors included infant birth weight, gestational age, single or multiple gestation, and infant sex. All delivered women are encouraged to attend a PPV as standard of care; women see a certified nurse midwife, nurse, or physician at this visit. The PPV was defined by an appointment <43 days postpartum since this was the period covered by public insurance, and we chose to use this pragmatic timing since the majority of participants were publicly insured.

Statistical analysis

We first employed descriptive, univariate statistics of key variables, including demographic characteristics, prenatal contraception and infant feeding intentions, postpartum in-hospital and PPV contraceptive and infant feeding choices. After this, we performed bivariate statistics. We compared the characteristics of the initial cohort with the subpopulation who delivered at MWH. We then compared the characteristics of all those who delivered at MWH between those who attended the PPV and those who did not. Analyses related to the postpartum in-hospital visit were conducted with the cohort who delivered at MWH, and those related to the PPV with the cohort who both delivered at MWH and attended the PPV <43 days postpartum.

To answer RQs 1 and 2, we used similar approaches. To determine whether prenatal infant feeding intention is associated with postpartum in-hospital infant feeding choice or PPV feeding choice (RQ 1), we conducted two McNemar paired exact tests. Similarly, we conducted a McNemar paired exact test between feeding choice postpartum in-hospital and at the PPV. As these are exact tests, only p-values are presented. We also calculated the phi-correlation coefficients (ψ) for prenatal infant feeding intention, postpartum in-hospital feeding choice, and PPV feeding choice. Phi-correlation coefficients are the most appropriate choice for strength of association when considering dichotomous variables. They tend to have lower thresholds for strength compared with Pearson's correlation coefficient. We conducted the same sets of analyses for contraceptive intentions and choice at the same time points (RQ 2). We also compared mothers' prenatally ranked prioritizations of the relative importance of (1) breastfeeding, (2) pregnancy prevention, and (3) return to work with their PPV choices by using Fisher's exact test.

To answer RQ 3, that is, to determine whether infant feeding choice postpartum in-hospital is related to contraceptive choice during the same period, we conducted three analyses. First, we examined whether the three categories of feeding (exclusive breastfeeding, mixed feeding, and exclusive formula feeding) were related to contraceptive choice by using Fisher's exact test (due to a small cell size in one of the cells for mixed feeding). Second, we examined whether any breastfeeding (includes both exclusive breastfeeding and mixed feeding) is related to contraceptive choice by using chi-square (χ²) test of association. Finally, given the significant association between breastfeeding and contraception, we performed a logistic regression modeling the odds of any breastfeeding given the type of contraception chosen postpartum in the hospital. We fit two models, the first one with only breastfeeding type (the exposure variable of interest) and a second one including covariates that are suspected confounders, namely living status, education, maternal age, number of children, and whether the mother breastfed before this current pregnancy. We did not include other covariates, such as school or employment status or postpartum plans for school/work, for which we could not establish directionality between the exposure variable of interest and outcome with those variables. The exposure variable only model's results are discussed in the Results section, and the full model is presented in the Results section and in Table 5. We judged model fit based on the AIC (Akaike's Information Criterion) of the intercept only model with the current model. A lower AIC on the current model compared with the intercept only model indicates better fit.

Results

Population description

A total of 223 expectant women were interviewed prenatally and of these, 214 (96%) delivered at MWH and had available information; 1 subject's child did not survive to discharge, and 8 subjects did not deliver at MWH. Of the 214 mothers, the mean age was 25 years, 185 (89.8%) were African American, and 104 (48.8%) were partnered but not married (Table 1). Their infants' mean birthweight was 3,977 (SD 675), mean gestational age was 38.6 weeks (SD 2.5), and 207 (98.1%) were singletons. Of the 214 mothers, 119 (55.6%) subsequently attended their PPV. There were no significant differences in demographic descriptors or in prenatal feeding or contraceptive intentions between the initial interviewed sample, the 214 who delivered at MWH, and the 119 who attended the PPV.

Table 1.

Sample Characteristics at Prenatal, Postpartum In-Hospital, and Postpartum Visit

	Prenatal visit (N = 223)	Postpartum in-hospital (n = 214)	Postpartum visit at <43 days (n = 119)
Race, % (n)
African American	89.30 (192/215)	89.80 (185/206)	88.80 (103/116)
Caucasian	3.70 (8/215)	3.90 (8/206)	4.30 (5/116)
Other	7.00 (15/215)	6.30 (13/206)	6.90 (8/116)
Hispanic/Latina	4.70 (8/169)	4.90 (8/164)	6.50 (6/111)
Number of children, % (n)
None (primiparous)	25.00 (52/208)	23.80 (51/200)	32.40 (36/111)
1 or 2	57.20 (119/208)	54.30 (113/200)	55.80 (62/111)
3 or more	17.8 (37/208)	18.0 (36/200)	11.71 (13/111)
Marital status, % (n)
Partnered but not married	50.00 (111/222)	48.80 (104/213)	47.50 (56/118)
Single	41.90 (93/222)	32.70 (91/213)	41.50 (49/118)
Married	8.10 (18/222)	8.50 (18/213)	11.00 (13/118)
Living status, % (n)
Not living with partner/spouse	48.60 (106/218)	49.80 (104/209)	47.40 (55/116)
Living with partner/spouse	51.40 (112/218)	50.20 (105/209)	52.6 (61/116)
Education level, % (n)
Did not finish high school	18.60 (41/221)	18.40 (39/211)	14.40 (17/118)
Graduated from high school	45.70 (101/221)	46.20 (98/211)	46.60 (55/118)
Continued school after high school	35.70 (79/221)	35.40 (75/211)	39.00 (46/118)
School/employment status (prenatal), % (n)
School/work full-time	32.70 (72/220)	32.20 (68/211)	40.70 (48/118)
School/work part-time	25.90 (57/220)	25.60 (54/211)	22.00 (26/118)
Other	41.40 (91/220)	42.20 (89/211)	37.30 (44/118)
Postpartum plans for school/work, % (n)
I do not plan to return to work	2.70 (6/223)	2.80 (6/214)	0.80 (1/119)
Return to work before 3 months	41.30 (92/223)	42.50 (91/214)	47.90 (57/119)
Return to work after 3 months	41.70 (93/223)	41.10 (88/214)	37.80 (45/119)
I do not know	14.30 (32/223)	13.60 (29/214)	12.20 (16/119)
Maternal age (years)	25 (16–42)	25 (16–42)	25 (16–42)
Maternal age category (quartiles by years of age), % (n)
16 to 22	29.15 (65/223)	28.50 (61/214)	28.57 (34/119)
23 to 25	28.69 (64/223)	29.44 (63/214)	31.09 (37/119)
26 to 28	17.48 (39/223)	17.76 (38/214)	19.33 (23/119)
29 to 42	24.66 (55/223)	24.29 (52/214)	21.00 (25/119)
Gestational age (weeks) at interview (median, range)	34 (6–40)	30 (9–40)	31.00 (9–40)

Table 2 describes the 214 women's infant feeding and contraceptive choices postpartum. During in-hospital postpartum, 116 out of 213 (54.5%) women exclusively breastfed, 38 out of 213 (17.8%) gave mixed feeding, and 59 out of 213 (27.7%) were exclusively formula feeding (one record missing information). After delivery, 103 women (48. 1%) chose no contraception, 45 (21%) received a postplacental hormonal IUD, 41 (18.1%) received Medroxyprogesterone acetate injection, and 22 (10.3%) had a tubal ligation. As noted earlier, copper IUDs and the implant were not standardly available postpartum.

Table 2.

Postpartum Infant Feeding and Contraceptive Choices

	Postpartum in-hospital choice (N = 214), % (n)	Postpartum in-hospital plan (if no choice made)	Postpartum visit at <43 days (n = 119), % (n)
Infant feeding choice^a
Exclusive breastfeeding	54.5 (116/213)	NA	27.6 (29/119)
Mixed feeding	17.8 (38/213)	NA	25.7 (27/119)
Exclusive formula feeding	27.7 (59/213)	NA	46.7 (49/119)
Contraceptive method choice
None	29.4 (63/214)	41.2 (42/102)	12.7 (14/110)
None but “Will Discuss with Provider Later”	18.7 (40/214)	0	0
Barrier method/condoms	0.5 (1/214)	0	0.9 (1/110)
Progestin-only contraceptive pills	0.5 (1/214)	4.9 (5/102)	12.7 (14/110)
Combination oral contraceptive pills	0	7.8 (8/102)	4.5 (5/110)
Contraceptive vaginal ring	0	0	2.7 (3/110)
Medroxyprogesterone injection	12.1 (26/214)	13.7 (14/102)	23.6 (26/110)
Medroxyprogesterone injection—as bridge	7.0 (15/214)	0	0
IUD: levonorgestrel	21.0 (45/214)	10.8 (11/102)	22.7 (25/110)
IUD: nonhormonal^b	0	1.0 (1/102)	0.9 (1/110)
Subdermal implant^b	0	16.7 (17/102)	9.1 (10/110)
Tubal ligation	10.3 (22/214)	3.9 (4/102)	9.1 (10/110)
Other unspecified	0.5 (1/214)	0	0.9 (1/110)

One record missing information.

Copper IUDs and the implant were not standardly available postpartum.

IUDs, intrauterine devices.

Results by RQs

RQ 1—Prenatal infant feeding intention was significantly associated with feeding choice both postpartum in-hospital and at the PPV (Table 3). Phi-correlation coefficients (ψ), which are the most appropriate choice for strength of association when considering dichotomous variables and tend to have lower thresholds for strength compared with Pearson's correlation coefficient, are reported for each pair in Table 3, and they reveal strong to very strong positive correlations for each relationship. Prenatal feeding intention and postpartum in-hospital feeding choice had a statistically significant strong, positive correlation of 0.61 (p < 0.0001); prenatal intention and PPV choice had a statistically significant strong, positive correlation of 0.57 (p < 0.0001); and postpartum in-hospital choice and PPV choice had a statistically significant strong, positive correlation of 0.51 (p < 0.0001). This is illustrated in the high prevalence of those who intended to exclusively breastfeed and remained exclusively breastfeeding at postpartum: of those who planned to exclusively breastfed, 75 out of 101 mothers (74.3%) exclusively breastfed postpartum in-hospital, and 24 out of 52 (46.2% of those mothers who could be followed) were still exclusively breastfeeding at the PPV. We found a similar, but stronger pattern for exclusive formula feeding, in that once a mother decides to exclusively formula feed, her decision is essentially unchanged. Of mothers whose prenatal intention was exclusive formula feeding, 87.1% (27/31 mothers) chose this postpartum in-hospital and 100% (14/14 mothers) were then exclusively formula feeding at the PPV. Among mothers who prenatally ranked breastfeeding as their #1 priority (versus pregnancy prevention or return to work), significantly more were likely to be breastfeeding at the PPV than those who had ranked breastfeeding as their #2 or #3 priority (24 [66.2%] versus 6 [38.2%] versus 2 [15.8%], respectively, p < 0.001).

Table 3.

Prenatal Feeding and Contraceptive Intentions by Postpartum In-Hospital and Postpartum Visit Choices

Feeding	Exclusive breastfeeding, n (%)	Mixed feeding, n (%)	Exclusive formula, % (n)
Prenatal visit intention ↓ by postpartum in-hospital choice → (ψ = 0.61)^a
Exclusive breastfeeding	74.3 (75)	13.9 (14)	11.9 (12)
Mixed feeding	50.7 (36)	26.8 (19)	22.5 (16)
Exclusive formula	6.5 (2)	6.5 (2)	87.1 (27)
Unsure	30.0 (3)	30.0 (3)	40.0 (4)
Prenatal visit intention ↓ by postpartum visit choice → (ψ = 0.57)^a
Exclusive breastfeeding	46.2 (24)	32.7 (17)	21.2 (11)
Mixed feeding	14.3 (5)	25.7 (9)	60.0 (21)
Exclusive formula	0	0	100 (14)
Unsure	0	25.0 (1)	75.0 (3)
Postpartum in-hospital choice ↓ by postpartum visit choice → (ψ = 0.51)^a
Exclusive breastfeeding	37.9 (25)	28.8 (19)	33.3 (22)
Mixed feeding	17.6 (3)	35.3 (6)	46.1 (11)
Exclusive formula	4.5 (1)	9.1 (2)	86.4 (19)

Contraception	No contraception	LARC	Tubal ligation	Other types
Prenatal visit intent ↓ by postpartum in-hospital visit choice → (ψ = 0.41)^a
No contraception	70.0 (42)	13.3 (8)	8.3 (5)	8.3 (5)
Unsure	46.4 (13)	35.7 (10)	0	17.9 (5)
LARC	33.3 (13)	38.5 (15)	10.3 (4)	17.9 (7)
Other types	40.2 (35)	13.8 (12)	14.9 (13)	31.0 (27)
Prenatal visit intent ↓ by postpartum visit choice → (ψ = 0.47)^b
No contraception	24.1 (7)	24.1 (7)	10.3 (3)	41.1 (12)
Unsure	0	66.7 (12)	0	33.3 (6)
LARC	23.8 (5)	38.1 (8)	9.5 (2)	28.6 (6)
Other types	4.8 (2)	21.4 (9)	11.9 (5)	61.9 (26)
Postpartum in-hospital choice↓ by postpartum visit choice (ψ = 1.23)^a
No contraception	24.6 (14)	22.8 (13)	0	52.6 (30)
LARC	0	95.7 (22)	0	4.3 (1)
Tubal ligation	0	0	100 (10)	0
Other types	0	5.0 (1)	0	95.0 (19)

McNemar exact test p < 0.0001.

McNemar exact test p < 0.01.

LARC, long acting reversible contraception.

RQ 2—Contraceptive choice follows a similar pattern to feeding choice in this sample. Phi-correlation coefficients are reported in Table 3 and reveal strong to very strong positive correlations for each relationship. Regarding the relationships between contraceptive intention and choice at different time points, prenatal contraceptive intention and postpartum in-hospital choice had a statistically significant strong, positive correlation of 0.41 (p < 0.0001); prenatal contraceptive intention and PPV choice had a statistically significant strong, positive correlation of 0.47 (p < 0.01); and postpartum in-hospital contraceptive choice and PPV choice also had a statistically significant very strong, positive correlation of 1.23 (p < 0.0001). Of those who intended to use LARC, 38.5% (15/39) chose to use LARC postpartum in-hospital and 38.1% (8/21) continued using LARC at the PPV. Of women who were unsure at prenatal visit, 35.7% (10/28) opted to get LARC postpartum in-hospital, whereas 66.7% (12/18) ended up using no contraception at PPV. Women who had prenatally ranked pregnancy prevention as their #1 priority (versus breastfeeding or return to work) were highly likely to be using any contraception at the PPV (34 [91.9%]), but this was not significantly different from those who prioritized pregnancy prevention as #2 or #3 (p = 0.23).

RQ 3—Table 4 presents test results and contingency tables for the relationship between breastfeeding and contraceptive choice postpartum in-hospital. We failed to find a statistically significant relationship between the three feeding choice variables (exclusive breastfeeding versus mixed feeding versus exclusive formula feeding) and contraceptive choice (p = 0.08, Table 4). However, we found that more women who chose any breastfeeding (54.5% for any breastfeeding versus 32.2% for exclusive formula feeding) chose no contraception, compared with LARC, tubal ligation, or other types of contraception. This relationship was statistically significant (χ² = 9.28, p = 0.03) before controlling for measured confounders.

Table 4.

Relationship of Infant Feeding Choice to Contraception Choice Postpartum In-Hospital

	No contraception, % (n)	LARC, % (n)	Tubal ligation, % (n)	Other types, % (n)
Exclusive breastfeeding^a	53.4 (62)	19.8 (23)	7.8 (9)	19.0 (22)
Mixed feeding	57.9 (22)	18.4 (7)	13.2 (5)	10.5 (4)
Exclusive formula feeding	32.2 (19)	25.4 (15)	11.9 (7)	30.5 (18)
Any breastfeeding^b	54.5 (84)	19.5 (30)	9.1 (14)	16.9 (26)
Exclusive formula feeding	32.2 (19)	25.4 (15)	11.9 (7)	30.5 (18)

Fisher's exact test p = 0.08.

χ² = 9.28, p = 0.03, any breastfeeding includes both exclusive breastfeeding and mixed feeding.

LARC, long-acting reversible contraception.

Table 5 presents the results of the logistic regression, which controls for measured confounders. Although overall, contraceptive type became not statistically significant (Type 3 omnibus p = 0.07), odds of any breastfeeding with contraception other than LARC or tubal ligation, as compared with no contraceptive type, were reduced (odds ratio 0.35, confidence interval [95% CI] 0.14–0.91, p = 0.02). Thus, women not on contraception, compared with those who received any other contraceptive type (not LARC or tubal ligation), had 2.86 times the odds of breastfeeding. Other statistically significant covariates for feeding choice were living status, number of children, and having breastfed before. Women living with a partner had 2.15 times the odds of breastfeeding compared with women not living with a partner (95% CI 1.02–4.53, p = 0.04), and those who had breastfed earlier, compared with women who had not, had 4.36 times the odds of breastfeeding (95% CI 1.81–10.49, p = 0.02). Women with more children had lower odds of breastfeeding, with every one additional child reducing by 0.47 times the odds of breastfeeding (95% CI 0.31–0.71, p = 0.0003). Education level and maternal age did not significantly alter the odds of any breastfeeding.

Table 5.

Logistic Regression Modeling Odds of Any Breastfeeding Given Contraception Choice and Other Covariates Postpartum In-Hospital, n = 196

	β	Error	OR	95% CI	p
Intercept	−1.12	1.18			0.34
LARC^a	−0.21	0.33	0.64	0.25–1.63	0.52
Other contraception methods^a	−0.81	0.35	0.35	0.14–0.91	0.02
Tubal ligation^a	−0.79	0.47	1.75	0.47–6.52	0.09
Living with a partner/spouse^b	0.38	0.38	2.15	1.02–4.53	0.04
Graduated from high school/GED^c	−0.37	0.25	0.73	0.28–1.86	0.14
Continued school after high school^c	0.42	0.30	1.60	0.53–4.80	0.16
Maternal age	0.05	0.05	1.05	0.95–1.16	0.36
Number of children	−0.76	0.21	0.47	0.32–0.71	0.0003
Breastfed before—yes^d	0.77	0.42	4.36	1.81–10.50	0.02
Breastfed before—no, first pregnancy^d	−0.07	0.33	1.88	0.55–6.45	0.86

Reference category is no contraception.

Reference category is not living with a partner/spouse.

Reference category is not graduated from high school.

Reference category is breastfed before—no.

CI, confidence interval; LARC, long-acting reversible contraception; OR, odds ratio.

Discussion

Findings and interpretation

In this group of low-income inner-city predominantly African American mothers, we found that postpartum in-hospital more than half of women (54.5%) exclusively breastfed and just less than half (48.1%) chose to receive no contraception. Prenatal infant feeding intention was significantly associated with postpartum infant feeding choice both in-hospital and at the PPV (both p < 0.0001). Similarly, prenatal contraceptive intention was significantly associated with contraceptive choice both in-hospital before discharge and at the PPV (both also p < 0.0001). The identified association between any breastfeeding and no contraceptive was no longer statistically significant after controlling for measured confounders. However, the most noteworthy finding of this study is that choice of other less effective types of contraception (not LARC or tubal ligation), compared with no contraception, remained significantly associated with decreased odds of any breastfeeding (p = 0.02) after controlling for confounders.

Comparison to similar studies

It has been previously shown that prenatal breastfeeding intention is an important predictor of postpartum breastfeeding initiation and duration. In a large cohort, Donath et al. found that intended duration of breastfeeding alone correctly predicted 91.4% of breastfeeding initiation and 72.2% of breastfeeding at 6 months.^24,25 In a predominantly African American inner-city population comparable to our sample, Hundalani et al. demonstrated a similar though weaker relationship, in that 75% of women who intended to breastfeed were doing so postpartum.²⁶

Less information is available about the relationship between prenatal contraceptive intention and postnatal contraceptive choice. Several studies, including ones in populations similar to ours, have focused on factors impacting contraceptive intention without recording postnatal choice, have examined factors associated with postnatal contraceptive intent, or have examined the relationship between postnatal intention and subsequent contraceptive receipt, without recording prenatal intention.^27–30 A retrospective study by Brownell et al. demonstrated that among low-income women who received depot Medroxyprogesterone immediately postpartum, 72.4% had not planned to use this contraceptive method.³¹ This study was limited to women who initiated breastfeeding and was specific to Medroxyprogesterone acetate injection administration. In another retrospective study, 3,063 pregnancies in a racially diverse population were compared: Although >90% of women had a documented prenatal contraceptive plan, there was a significant decrease from prenatal plan to postpartum discharge and the PPV in those receiving the most effective contraception (Tier 1—IUD, sterilization, implant, vasectomy).³² Thus, our findings do not fully align with prior literature.³³

Contraceptive methods that are safe to use postpartum in breastfeeding mothers are well described.^13,15 However, few studies examine the relationship between contraceptive intention and choice as it relates to infant feeding intention and choice. A retrospective study examined the association between postpartum contraception and breastfeeding at 4 months among women who intended to breastfeed: the authors demonstrated that those who selected progestin-only pills were most likely, and those who selected combined oral contraceptive pills were least likely, to achieve their breastfeeding intentions.³³ However, this cohort was largely white, college educated, affluent, and married, and it included only those who intended to breastfeed; prenatal contraceptive plans were not recorded. Loewenberg et al. investigated postpartum contraceptive intentions and choice among breastfeeding women.³⁴ They reported that 91% intended to use postpartum contraception, but just 21% reported considering the impact of their contraception method on breastfeeding when deciding on a postpartum contraceptive.³⁴ Study participants were largely highly educated, Non-Hispanic white, employed, and married. Finally, among the women who are the subject of this article, we previously found no difference in prenatal contraceptive intentions between those who intended to breastfeed and those who intended to feed formula.¹⁹ Thus, our finding that any breastfeeding postpartum, as compared with formula feeding, was significantly associated with not using the most effective contraceptive methods (LARC, tubal ligation) adds to the literature.

Breastfeeding

Finally, we identified expected associations with breastfeeding, in that living with a partner and having breastfed previously are known to increase odds of initiating and continuing breastfeeding, whereas increasing parity is associated with decreasing odds of breastfeeding initiation.^35–37

Strengths and limitations of the study

Strengths of this study include its homogeneous but understudied population at increased risk of both short interval pregnancy and not breastfeeding. The longitudinal design with follow-up of women interviewed prenatally allowed us to examine the relationship between prenatal intention and postnatal choice, and the interaction between feeding and contraceptive decisions. Our study is unique in investigating these issues prospectively among low-income, inner-city predominantly African American women. Finally, follow-up of the initial cohort was good, with 214 (96%) followed to postpartum in-hospital. Although only 119 mothers (55.6%) could be followed to the PPV, this is representative of clinicians' experiences in our setting, and there were no significant demographic differences between those who were and were not followed.

Study limitations include that we did not interview women postpartum and no information on feeding or contraception was gathered beyond the PPV, which could have extended the findings. Women were interviewed prenatally at a wide range of gestational ages in the initial study, and thus some women may have had more time for decision making or changing than others; interview timing, however, was not significantly related to intent or choice, limiting its impact on study results. A woman's choice of postpartum contraception, and possibly also feeding choice, may be impacted by prenatal provider counseling and by local method availability. We do not have information about whether prenatal providers' contraceptive counseling addressed individual maternal feeding intention or breastfeeding compatibility of contraception in general. It is possible that the greater local availability of the hormone-releasing IUD, versus the copper IUD and the implant, could have shaped and impacted maternal choice. We do not know whether postplacental intrauterine methods were offered only after vaginal delivery or after cesarean section, or both, which is a study limitation, as this may have had a bearing on uptake rates. The small sample size in total and by subgroup of contraceptive choice impacted statistical power, and it is possible that despite controlling for measured confounders, the significantly reduced odds of any breastfeeding among women receiving less effective contraceptive types (not LARC and not tubal ligation) is spurious. Finally, the generalizability of results is limited to the population studied.

Conclusion

We found, in this population of low-income inner-city predominantly African American women at high risk for both not breastfeeding and short interval pregnancy, that those receiving contraception other than LARC or tubal ligation, versus no contraception, had reduced odds of any breastfeeding (p = 0.02). Thus, mothers who were breastfeeding at all, as compared with those who were feeding formula exclusively, were less likely to have effective contraception. Breastfeeding without effective contraception puts mothers at risk for unintended pregnancy, and it represents a missed opportunity to prevent rapid repeat pregnancy and its associated negative maternal and infant outcomes.^8,9 These findings highlight the importance of comprehensive counseling of expectant mothers on contraception and breastfeeding, so they may choose the method that best aligns with their feeding and family planning goals. Future research may include confirming the association we have identified between feeding method and contraceptive choice, as well as understanding what women's personal barriers are to choosing and accepting contraception if breastfeeding.

Footnotes

Acknowledgments

The authors thank the women who participated in this study, and the staff and faculty of the MWH Women's Health Center for their support and enthusiasm.

Disclosure Statement

No competing financial interests exist.

References

Hoddinott

, Tappin

, Wright

. Breast feeding. BMJ, 2008; 336:881–887.

Sankar

, Sinha

, Chowdhury

, et al. Optimal breastfeeding practices and infant and child mortality: A systematic review and meta-analysis. Acta Paediatr, 2015; 104:3–13.

Thompson

JMD

, Tanabe

, Moon

, et al. Duration of breastfeeding and risk of SIDS: An individual participant data meta-analysis. Pediatr, 2017; 140. pii: e20171324.

Chowdhury

, Sinha

, Sankar

, et al. Breastfeeding and maternal health outcomes: A systematic review and meta-analysis. Acta Paediatr, 2015; 104(Suppl 467):96–113.

Victora

, Bahl

, Barros

, et al.; Lancet Breastfeeding Series Group. Breastfeeding in the 21st century: Epidemiology, mechanisms, and lifelong effect. Lancet, 2016; 387:475–490.

Bartick

, Schwarz

, Green

, et al. Suboptimal breastfeeding in the United States: Maternal and pediatric health outcomes and costs. Matern Child Nutr, 2017; 181:49–55.e6.

Breastfeeding Rates, National Immunization Survey. Percentage of U.S. Children who were Breastfed by Birth Date, National Immunization Survey, United States. Available at www.cdc.gov/breastfeeding/data/nis_data/results.html (accessed February 1, 2019).

Conde-Agudelo

, Rosas-Bermúdez

, Kafury-Goeta

. Birth spacing and risk of adverse perinatal outcomes: A meta-analysis. JAMA, 2006; 295:1809–1823.

Conde-Agudelo

, Rosas-Bermúdez

, Kafury-Goeta

. Effects of birth spacing on maternal health: A systematic review. Am J Obstet Gynecol, 2007; 196:297–308.

10.

Marston

. World Health Organization. Report of a WHO Technical Consultation on Birth Spacing. Geneva, Switzerland, 2005. Available at http://apps.who.int/iris/bitstream/handle/10665/69855/WHO_RHR_07.1_eng.pdf;sequence=1 (accessed June 19, 2018).

11.

Healthy People 2020, 2020 Topics and Objectives, Family Planning. FP-5: Reduce the proportion of pregnancies conceived within 18 months of a previous birth. Available at www.healthypeople.gov/node/4461/data_details (accessed June 19, 2018).

12.

Cleland

, Conde-Agudelo

, Peterson

, et al. Contraception and health. Lancet, 2012; 380:149–156.

13.

Sridhar

, Salcedo

. Optimizing maternal and neonatal outcomes with postpartum contraception: Impact on breastfeeding and birth spacing. Matern Health Neonatol Perinatol, 2017; 3:1.

14.

Harney

, Dude

, Haider

. Factors associated with short interpregnancy interval in women who plan postpartum LARC: A retrospective study. Contraception, 2017; 95:245–250.

15.

Curtis

, Tepper

, Jatlaoui

, et al. U.S. Medical eligibility criteria for contraceptive use, 2016. MMWR Recomm Rep, 2016; 65:1–104.

16.

Hogue

, Menon

, Dunlop

, et al. Racial disparities in preterm birth rates and short inter-pregnancy interval: An overview. Acta Obstet Gynecol Scand, 2011; 90:1317–1324.

17.

Anstey

, Chen

, Elam-Evans

, et al. Racial and geographic differences in breastfeeding—United States, 2011–2015. MMWR Morb Mortal Wkly Rep, 2017; 66:723–727.

18.

Singh

, Tripathi

, Singh

, et al. Breastfeeding as a time-varying-time-dependent factor for birth spacing: Multivariate models with validations and predictions. World Health Popul, 2012; 13:28–51.

19.

Fuell Wysong

, Tossone

, Furman

. Expectant inner-city women: Attitudes about contraception given infant feeding choice. Eur J Contracept Reprod Health Care, 2017; 22:369–374.

20.

Harris

, Taylor

, Thielke

, et al. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42:377–381.

21.

Hubacher

, Spector

, Monteith

, et al. Long-acting reversible contraceptive acceptability and unintended pregnancy among women presenting for short-acting methods: A randomized patient preference trial. Am J Obstet Gynecol, 2016; 216:101–109.

22.

Usinger

, Gola

, Weis

, et al. Intrauterine contraception continuation in adolescents and young women: A systematic review. J Pediatr Adolesc Gynecol, 2016; 29:659–667.

23.

Medical eligibility criteria for contraceptive use; A WHO Family Planning Cornerstone, Fifth Edition 2015. Department of Reproductive Health and Research World Health Organization. Available at http://apps.who.int/iris/bitstream/handle/10665/181468/9789241549158_eng.pdf;jsessionid=406E48260C747EB164C9E18B41C12230?sequence=1 (accessed December 31, 2018).

24.

Donath

, Amir

; ALSPAC Study Team. Relationship between prenatal infant feeding intention and initiation and duration of breastfeeding: A cohort study. Acta Paediatr, 2003; 92:352–356.

25.

Nelson

, Li

, Perrine

, et al. Changes in mothers' intended duration of breastfeeding from the prenatal to neonatal periods. Birth, 2018; 45:178–183.

26.

Hundalani

, Irigoyen

, Braitman

, et al. Breastfeeding among inner-city women: From intention before delivery to breastfeeding at hospital discharge. Breastfeed Med, 2013; 8:68–72.

27.

Chacko

, Wiemann

, Buzi

, et al. Choice of postpartum contraception: Factors predisposing pregnant adolescents to choose less effective methods over long-acting reversible contraception. J Adolesc Health, 2016; 58:628–635.

28.

Zerden

, Tang

, Stuart

, et al. Barriers to receiving long-acting reversible contraception in the postpartum period. Womens Health Issues, 2015; 25:616–621.

29.

Di Giacomo

, Sbarlati

, Bagnasco

, et al. Woman's contraceptive needs and preferences in the postpartum period: An Italian study. J Clin Nurs, 2013; 22:3406–3417.

30.

Simmons

, Edelman

, Li

, et al. Personalized contraceptive assistance and uptake of long-acting, reversible contraceptives by postpartum women: A randomized, controlled trial. Contraception, 2013; 88:45–51.

31.

Brownell

, Lussier

, Dozier

, et al. The discordance between planned use and actual receipt of immediate postpartum depot medroxyprogesterone among low-income women. Breastfeed Med, 2014; 9:290–293.

32.

French

, Albanese

, Gossett

. Postpartum contraceptive choice after high-risk pregnancy: A retrospective cohort analysis. Contraception, 2016; 94:173–180.

33.

Goulding

, Wouk

, Stuebe

. Contraception and breastfeeding at 4 months postpartum among women intending to breastfeed. Breastfeed Med, 2018; 13:75–80.

34.

Loewenberg Weisband

, Keder

, Keim

, et al. Postpartum intentions on contraception use and method choice among breastfeeding women attending a university hospital in Ohio: A cross-sectional study. Reprod Health, 2017; 14:45.

35.

Mahesh

PKB

, Gunathunga

, Arnold

, et al. Effectiveness of targeting fathers for breastfeeding promotion: Systematic review and meta-analysis. BMC Public Health, 2018; 18:1140.

36.

Blyth

, Creedy

, Dennis

, et al. Effect of maternal confidence on breastfeeding duration: An application of breastfeeding self-efficacy theory. Birth, 2002; 29:278–284.

37.

Sutherland

, Pierce

, Blomquist

, et al. Breastfeeding practices among first-time mothers and across multiple pregnancies. Matern Child Health J, 2012; 16:1665–1671.