Abstract
Background:
Religion has rarely been studied as a determinant of infant feeding practices. We examined whether religious affiliation is associated with formula feeding vs breastfeeding intention and practice in women from the US Project Viva cohort.
Methods:
Between 1999 and 2002, 2,128 pregnant women were recruited in the area of Boston, Massachusetts. They reported by questionnaire their religious affiliation, and their intended and practiced infant feeding mode (exclusive formula feeding vs partial vs exclusive breastfeeding) at different time points. We examined associations of religious affiliation with infant feeding intention and practice by modified Poisson regression and multinomial logistic regression adjusted for known sociodemographic confounders.
Results:
Of 1,637 women with complete data, 52% reported being Catholic, 29% Protestant, 11% unaffiliated, 4% Jewish and 4% of other religion. Overall, 8.5% and 15.9% women intended and initiated exclusive formula feeding, respectively. Compared to unaffiliated women, Catholics were more at risk to intend to exclusively formula feed their infant at birth (risk ratio [95% CI]: 6.4 [1.6-26.0]), to exclusively formula feed after delivery (2.4 [1.3-4.2]) and 3 months postpartum (1.3 [0.98-1.8]). The odds ratio for intending and practicing partial (vs exclusive) breastfeeding did not differ by religious affiliation at most examined time points. Associations of Protestant women with infant feeding exhibited estimates closer to unaffiliated than Catholic women.
Conclusions:
Catholic women are more at risk to intend and practice exclusive formula feeding than women of other religious affiliations. Our findings may help health care providers adapt their breastfeeding promotion to the mother’s religious affiliation.
Keywords: pregnancy, breastfeeding, infant formula, religion, Christianity
INTRODUCTION
Breastfeeding is widely recognized to confer important short-term health benefits over formula feeding.1 These include lower infant mortality due to infectious diseases,2 and decreased risks of gastrointestinal and respiratory infections.3 Beneficial effects of breastfeeding on later child cognitive development have also been reported,4–7 but may diminish with age.8 The World Health Organization recommends exclusive breastfeeding for 6 months and continued partial breastfeeding to 2 years.9
These international recommendations are not followed by a substantial proportion of mothers, particularly in high-income countries, where breastfeeding initiation rates are <85% (vs >95% in low- and middle-income countries).1 Despite recent increases, those rates remain even lower in Ireland (58% in 2015),10 France (70% in 2011),11 and the USA (83% in 2015).12 Breastfeeding rates are generally higher among mothers who are older, more educated, of higher socioeconomic status, and with older children; rates may also vary by race/ethnicity and migration status, depending on the settings.13,14 Breastfeeding practice is also higher in women who declare during pregnancy their intention to breastfeed.15 Cultural factors, including social norms, attitudes and knowledge about breastfeeding, also matter.16 Deeper-rooted cultural factors such as religious culture, however, have rarely been examined. Yet, religious scriptures, including the Talmud and the Holy Bible, frequently refer to breasts and mother’s milk.
Using an ecological study design, we previously reported associations between religious affiliation and breastfeeding initiation rates between and within Western countries.17 Countries and regions with high proportions of Roman Catholics had consistently lower breastfeeding initiation rates than those with high proportions of Protestants, suggesting that Catholic women are more at risk of formula feeding in place of initiating to the breast. Inferring causality from ecological studies is suspect, however, since associations may differ at the population vs individual levels.18,19 A US study in a religiously diverse sample of 4,166 women reported that those from some Protestant affiliation subgroups had higher rates of breastfeeding initiation compared to those without religious affiliation.20 Catholic women had lower breastfeeding rates that did not differ “significantly” from the unaffiliated. Although that study is the only individual-level study of which we are aware so far on this topic, its findings may lack generalizability because of the low proportion of non-Hispanic White women in its sample. As suggested by our previous study, further examination seems warranted for women of European ancestry living in the US.
New England has a long history of immigration from both Catholic and Protestant religious backgrounds. The first European settlers were predominantly Catholic French colonists near the Canadian border and English Reformed Protestants. Several immigration waves occurred thereafter, including arrivals of Catholic Irish and Polish populations, Protestant Germans, and Catholic Italians. According to the 2011-2015 American Community Survey, a substantial proportion of the current Massachusetts population declares their ancestry as Irish (22%), Italian (14%), English (10%), French (7%) or German (6%).21 Although Massachusetts is one of the least religious states in the US (32% report no affiliation) and includes a large number of Blacks and Hispanics, it comprises substantial proportions of non-Hispanic White Catholics and Protestants thus providing a useful setting to examine associations between religion and infant feeding in women of European ancestry. If religious affiliation affects BF practice, it may help tailor breastfeeding promotion policies to the religious and cultural views of the mother-to-be.
In the present study, we examined associations of maternal religion with infant feeding intention and practice in women enrolled in Project Viva, a US pregnancy-birth cohort study based in the Boston area. We hypothesized that, compared with Protestant and unaffiliated women, Catholics would be more at risk of formula feeding in place of breastfeeding.
METHODS
Study design and population
Project Viva is a longitudinal pregnancy-birth cohort study whose aims are to examine how early life exposures and events affect later health outcomes. Pregnant women were recruited from 1999 through 2002 at their first obstetric visit at eight urban and suburban sites of Atrius Harvard Vanguard Medical Associates, a multispecialty group practice in eastern Massachusetts. Exclusion criteria included multiple gestation, gestational age ≥22 weeks at recruitment, inability to answer questions in English, or plans to move away from the study area before delivery. In total, 2,341 eligible women were enrolled, of whom 2,128 gave birth to live infants. Details of the study protocol and the participants have been published elsewhere.22
All participants provided written informed consent. The institutional review boards of Harvard Pilgrim Health Care, Brigham and Women’s Hospital, and Beth Israel Deaconess Medical Centre, all located in Boston, MA, approved the study protocols. The present analysis received further approval from the Inserm Ethics Evaluation Committee under the registration number 17-402.
Women reported cultural, demographic and socioeconomic information by interview with a research assistant and on two self-administered questionnaires in early and mid-pregnancy. Information obtained included self-reported race/ethnicity, place of birth, date of birth, highest level of completed schooling, occupational status, annual household income, marital status, parity, and smoking habits before and during pregnancy. Women also reported whether they themselves were ever breastfed when they were infants (yes, no, do not know). Maternal height (m) and pre-pregnancy weight (kg) were self-reported in early pregnancy, from which we calculated pre-pregnancy body mass index as weight divided by the square of height (kg/m2).
In the present analysis, we excluded 107 women who were not of non-Hispanic White, Black, Hispanic or Asian ethnicities (n=77 mixed race/ethnicity, 1 American Indian, 5 other race/ethnicity, n=24 missing race/ethnicity) because of sparse strata in these racial/ethnic groups (Figure 1).
Figure 1.
Flow chart of the participants, Project Viva mother-child cohort, Massachusetts, 1999-2002.
Religious affiliation
Of the 2,021 women of non-Hispanic White, Black, Hispanic or Asian ethnicities who were recruited, 1,735 completed a self-administered questionnaire in early pregnancy (Figure 1). The questionnaire included a question on religious affiliation worded as follows: “What do you consider to be your religion or spiritual tradition? (Check all that apply).” Proposed answers were Catholic, Protestant, Unitarian, Jewish, Hindu, Muslim, Buddhist, none, and other (please specify). When free text was specified, religious affiliation was manually recoded. Women who declared ≥2 religions (n=47) and those who declared being undecided/not knowing (n=4) were excluded from the analysis. The final variable for religious affiliation was recoded into 5 categories: unaffiliated, Catholic, Protestant (including Anglican, Nondenominational Christian, ‘Protestant’ and Unitarian), Jewish, and other religions (including Mormon, Muslim, Christian Orthodox, Pagan, Rastafarian and Sikh). We were unable, however, to distinguish specific Protestant denominations; the Protestant category was therefore defined as all Christian Churches that departed from the Roman Catholic Church after the Protestant Reformation of 1517. Although we explored other ways to analyze the religion categories, unaffiliated women were set as the reference category, as this facilitated interpretation of the results.
Infant feeding intention and practice
We obtained information on women’s infant feeding intention and practice at three time points. In mid-pregnancy, women reported their infant feeding intention for the first week after delivery (“What kind of milk do you plan to feed your baby in the first week of life?”) and for 3 months later (“What kind of milk do you plan to feed your baby at 3 months of age?”) using 5 possible answers: 1) breast milk only, 2) mostly breast milk, some formula, 3) mostly formula, some breast milk, 4) formula only, or 5) haven’t decided yet. Categories 2 and 3 were merged into a single category labelled “partial breastfeeding”. Women in category 5 were excluded from the analysis.
During their maternity hospital stay after delivery, mothers were interviewed about whether their infant had been breastfed (“Have you breastfed your baby?” Yes or no) and had received any formula (“Has your baby received any formula?” Yes or no). Mothers additionally stated their infant feeding intention for the following week and at 3 months after delivery (with response categories worded/coded as above).
The above-described variables were recoded both into two categories (exclusive formula feeding vs ever breastfeeding) and into three categories (exclusive formula feeding, partial breastfeeding, and exclusive breastfeeding) for additional analyses based on previously published studies.17,20
At 6 months and 1 year postpartum, mothers reported the infant’s age when they stopped breastfeeding (if already weaned), from which we derived a two-category variable of infant feeding at 3 months (exclusive formula feeding vs any breastfeeding).
Covariates
Covariates used were maternal race/ethnicity, birthplace, age, education, annual household income, occupation status, marital status, pre-pregnancy body mass index, pregnancy smoking status, parity and whether the mother had been breastfed herself when she was an infant. Of note, marital status was categorized as married vs other, including living with her partner, single, separated, divorced, and widowed. Marital status was more strongly associated with declaring a religion (vs unaffiliated) when categorized this way than when categorized as ‘in a couple’ vs other.
Statistical analysis
We analyzed women’s characteristics by religious affiliation and compared them using chi-square test. We examined associations of maternal religious affiliation with infant feeding intention and practice at three time points (antepartum intention to exclusively formula feed after delivery, postdelivery practice of exclusive formula feeding, and practice of exclusive formula feeding at 3 months postpartum) using multivariable Poisson regression with robust error variance that allows estimating adjusted risk ratios (aRRs).23 We used the largest outcome group (ever breastfeeding) as reference, i.e., we estimated aRRs (95% CI) of exclusive formula feeding intention and practice. Models were adjusted for the above-described covariates except occupation status as it was not associated with religion or breastfeeding in our dataset. To explore the potential effect-modifier role of the mother’s own feeding mode when she herself was an infant on feeding mode of her infant, we explored the influence of adding/removing this covariate on the model estimates. To assess whether the associations differed across race/ethnicity strata, we tested interactions of religious affiliation with race/ethnicity and attempted stratified analyses. Only the analyses restricted to non-Hispanic White women were possible, however, since models among women of other races/ethnicities failed to converge, owing to insufficient sample size.
Last, to distinguish among exclusive vs partial breastfeeding, we further examined associations of maternal religious affiliation with infant feeding intention and practice as three-category variables (exclusive breastfeeding, partial breastfeeding, exclusive formula feeding). As Zou’s Poisson regression method applies to binary outcomes only, we used multivariable multinomial (non-ordered) logistic regression using the generalized logit function to estimate the odds ratios (ORs) of intending/practicing partial breastfeeding and exclusive formula feeding (exclusive breastfeeding as the reference group). We restricted this specific analysis to Catholic, Protestant and unaffiliated women because the two remaining religious affiliation subgroups (Jewish and other religions) were too small. All statistical analyses were carried out with SAS v9.4 (SAS Institute Inc, Cary, NC, USA).
RESULTS
Of the 2,021 women of non-Hispanic White, Black, Hispanic White or Asian races/ethnicities, 1,655 had data on religious affiliation, of whom 1,637 had data on all covariates and at least 1 outcome of breastfeeding intention or practice. Compared to women excluded from the present analysis, those included were older, more likely of White race/ethnicity, and had a higher household income, higher education level and lower BMI (results not shown). Most women were non-Hispanic White (75%) and born in the US (80%) (Table 1). About one-half (52%) were Catholic, 29% Protestant, 4% Jewish, 4% of other religions and 11% unaffiliated. All women’s characteristics except occupational status differed by religious affiliation. Of note, Catholic women were more likely to be non-Hispanic White (84%) than were women of Protestant (65%), unaffiliated (61%) or of other (42%) religions. They were also more likely to be born in the US (84%) than women of other religions (43%) or those who were unaffiliated (71%), and less likely (27%) to have obtained a graduate degree than all other religious affiliations. Catholic women were more likely to have been formula fed as infants (64%) than women of other religious affiliations (Protestants: 43%; Jewish: 43%; other religion: 35%; unaffiliated: 34%). During pregnancy, although only 8.5% of women declared intending to exclusively formula feed after delivery, 15.9% actually did so, and a third (31.6%) exclusively formula fed at 3 months postpartum (Table 2).
Table 1.
Women’s sociodemographic characteristics according to religious affiliation, Project Viva mother-child cohort, Massachusetts, 1999-2002.
| Catholic N=847 n (%) |
Protestant N=480 n (%) |
Jewish N=68 n (%) |
Other religion N=69 n (%) |
Unaffiliated N=173 n (%) |
|
|---|---|---|---|---|---|
| Race/ethnicity | |||||
| Non-Hispanic White | 712 (84.1) | 314 (65.4) | 66 (97.1) | 29 (42.0) | 106 (61.3)*** |
| Black | 34 (4.0) | 128 (26.7) | 0 (0.0) | 11 (15.9) | 34 (19.7) |
| Hispanic White | 82 (9.7) | 23 (4.8) | 0 (0.0) | 1 (1.4) | 1 (0.6) |
| Asian | 19 (2.2) | 15 (3.1) | 2 (2.9) | 28 (40.6) | 32 (18.5) |
| Birthplace | |||||
| US | 715 (84.4) | 387 (80.6) | 58 (85.3) | 30 (43.5) | 122 (70.5)*** |
| Foreign country | 132 (15.6) | 93 (19.4) | 10 (14.7) | 39 (56.5) | 51 (29.5) |
| Age category | |||||
| <25 years | 44 (5.2) | 37 (7.7) | 0 (0.0) | 7 (10.1) | 16 (9.2)** |
| 25-29 years | 172 (20.3) | 104 (21.7) | 9 (13.2) | 21 (30.4) | 29 (16.8) |
| 30-34 years | 388 (45.8) | 191 (39.8) | 31 (45.6) | 33 (47.8) | 74 (42.8) |
| ≥35 years | 243 (28.7) | 148 (30.8) | 28 (41.2) | 8 (11.6) | 54 (31.2) |
| Education | |||||
| High school degree or less | 76 (9.0) | 48 (10.0) | 0 (0.0) | 6 (8.7) | 14 (8.1)*** |
| Some college | 221 (26.1) | 83 (17.3) | 2 (2.9) | 14 (20.3) | 26 (15.0) |
| College degree | 325 (38.4) | 202 (42.1) | 25 (36.8) | 20 (29.0) | 46 (26.6) |
| Graduate degree | 225 (26.6) | 147 (30.6) | 41 (60.3) | 29 (42.0) | 87 (50.3) |
| Occupational status | |||||
| Full time | 530 (62.6) | 315 (65.6) | 48 (70.6) | 48 (69.6) | 116 (67.1) |
| Part time | 195 (23.0) | 112 (23.3) | 13 (19.1) | 13 (18.8) | 35 (20.2) |
| Unemployed | 115 (13.6) | 49 (10.2) | 6 (8.8) | 7 (10.1) | 22 (12.7) |
| Annual household income | |||||
| ≤$40,000 | 84 (9.9) | 70 (14.6) | 1 (1.5) | 11 (15.9) | 23 (13.3)*** |
| $40,001 to $70,000 | 200 (23.6) | 109 (22.7) | 13 (19.1) | 18 (26.1) | 20 (11.6) |
| >$70,000 | 521 (61.5) | 271 (56.5) | 51 (75.0) | 31 (44.9) | 117 (67.6) |
| Do not know | 42 (5.0) | 30 (6.3) | 3 (4.4) | 9 (13.0) | 13 (7.5) |
| Marital status | |||||
| Married | 746 (88.1) | 387 (80.6) | 65 (95.6) | 61 (88.4) | 136 (78.6)*** |
| Not married | 101 (11.9) | 93 (19.4) | 3 (4.4) | 8 (11.6) | 37 (21.4) |
| Prepregnancy BMI category | |||||
| <18.5 kg/m2 | 24 (2.8) | 21 (4.4) | 1 (1.5) | 6 (8.7) | 12 (6.9)* |
| 18.5-24.9 kg/m2 | 515 (60.8) | 285 (59.4) | 46 (67.6) | 43 (62.3) | 109 (63.0) |
| 25.0-29.9 kg/m2 | 192 (22.7) | 95 (19.8) | 16 (23.5) | 11 (15.9) | 35 (20.2) |
| ≥30 kg/m2 | 116 (13.7) | 79 (16.5) | 5 (7.4) | 9 (13.0) | 17 (9.8) |
| Pregnancy smoking status | |||||
| Current smoker | 104 (12.3) | 47 (9.8) | 2 (2.9) | 8 (11.6) | 20 (11.6)*** |
| Past smoker | 211 (24.9) | 72 (15.0) | 12 (17.6) | 11 (15.9) | 45 (26.0) |
| Never smoker | 532 (62.8) | 361 (75.2) | 54 (79.4) | 50 (72.5) | 108 (62.4) |
| Parity | |||||
| 0 | 384 (45.3) | 259 (54.0) | 42 (61.8) | 40 (58.0) | 88 (50.9)*** |
| 1 | 315 (37.2) | 154 (32.1) | 18 (26.5) | 23 (33.3) | 75 (43.4) |
| ≥2 | 148 (17.5) | 67 (14.0) | 8 (11.8) | 6 (8.7) | 10 (5.8) |
| Feeding mode as an infant | |||||
| Formula fed | 544 (64.2) | 208 (43.3) | 29 (42.6) | 24 (34.8) | 59 (34.1)*** |
| Do not know | 55 (6.5) | 39 (8.1) | 5 (7.4) | 4 (5.8) | 9 (5.2) |
| Breastfed | 248 (29.3) | 233 (48.5) | 34 (50.0) | 41 (59.4) | 105 (60.7) |
BMI, body mass index; US, United States.
Statistical significance of chi-square tests:
P < 0.05;
P < 0.01;
P < 0.001.
Table 2.
Number and rates of women intending and practicing exclusive breastfeeding, partial breastfeeding and exclusive formula feeding according to time of assessment (prepartum intention, postpartum practice, postpartum intention) and religious affiliation, Project Viva mother-child cohort, Massachusetts, 1999-2002.
| Exclusive breastfeeding n (%) |
Partial breastfeeding n (%) |
Exclusive formula feeding n (%) |
|
|---|---|---|---|
| Antepartum intention of feeding mode during the 1st week postpartum | |||
| Catholic | 527 (73.6) | 96 (13.4) | 93 (13.0) |
| Protestant | 323 (80.1) | 61 (15.1) | 19 (4.7) |
| Jewish | 56 (90.3) | 3 (4.8) | 3 (4.8) |
| Other | 45 (78.9) | 11 (19.3) | 1 (1.8) |
| Unaffiliated | 120 (83.9) | 21 (14.7) | 2 (1.4) |
| Antepartum intention of feeding mode at 3 months postpartum | |||
| Catholic | 270 (38.6) | 277 (39.6) | 152 (21.7) |
| Protestant | 182 (47.2) | 171 (44.3) | 33 (8.5) |
| Jewish | 33 (55.9) | 21 (35.6) | 5 (8.5) |
| Other | 27 (50.9) | 21 (39.6) | 5 (9.4) |
| Unaffiliated | 78 (57.8) | 51 (37.8) | 6 (4.4) |
| Actual feeding mode after delivery | |||
| Catholic | 433 (53.1) | 228 (28.0) | 154 (18.9) |
| Protestant | 255 (54.8) | 159 (34.2) | 51 (11.0) |
| Jewish | 49 (77.8) | 11 (17.5) | 3 (4.8) |
| Other | 37 (56.9) | 25 (38.5) | 3 (4.6) |
| Unaffiliated | 98 (58.0) | 60 (35.5) | 11 (6.5) |
| Postdelivery intention of feeding mode upon next week | |||
| Catholic | 538 (64.7) | 129 (15.5) | 165 (19.8) |
| Protestant | 313 (66.3) | 116 (24.6) | 43 (9.1) |
| Jewish | 57 (89.1) | 4 (6.3) | 3 (4.7) |
| Other | 48 (72.7) | 15 (22.7) | 3 (4.5) |
| Unaffiliated | 120 (71.0) | 38 (22.5) | 11 (6.5) |
| Postdelivery intention of feeding mode at 3 months postpartum | |||
| Catholic | 335 (41.2) | 290 (35.7) | 188 (23.1) |
| Protestant | 219 (47.4) | 175 (37.9) | 68 (14.7) |
| Jewish | 45 (70.3) | 14 (21.9) | 5 (7.8) |
| Other | 33 (51.6) | 20 (31.3) | 11 (17.2) |
| Unaffiliated | 81 (49.7) | 61 (37.4) | 21 (12.9) |
Risk of exclusive formula feeding vs ever breastfeeding
As compared to unaffiliated women, Catholic women were more at risk to intend to exclusively formula feed their infant at birth (aRR [95% CI]: 6.67 [1.72‒25.93]), to practice exclusive formula feeding after delivery (2.50 [1.42‒4.39]), and to continue exclusive formula feeding up to 3 months postpartum (1.46 [1.07‒2.01]) (Table 3). aRRs of Protestant and Jewish women were consistently of lower magnitude than those of Catholic women and their 95% CIs overlapped the null. In sensitivity analyses, adding/removing the women’s own feeding mode as an infant had a negligible impact on the strength of the observed associations between religious affiliation and exclusive formula feeding intention and practice. When restricting the analyses to non-Hispanic White women only, aRRs for formula feeding intention reduced importantly (2.81 [0.86‒9.17), but aRRs for practice of formula feeding after delivery (2.33 [1.01‒5.38]) and at 3 months postpartum (1.49 [0.96‒2.32]) remained of similar magnitude as in the overall sample. The aRRs (95% CI) of intending and practicing exclusive formula feeding (vs ever breastfeeding) for all other women’s sociodemographic characteristics are shown in a supplemental table published online only.
Table 3.
Associations of women’s religious affiliation with the risk to exclusively formula feed at three time points (antepartum intention, postdelivery practice, and 3-month postpartum practice, Project Viva mother-child cohort, Massachusetts, 1999-2002.
| Antepartum intention to exclusively formula feed after delivery N=1381 aRR (95% CI)a |
Postdelivery practice of exclusive formula feeding N=1619 aRR (95% CI) |
Practice of exclusive formula feeding at 3 months postpartum N=1351 aRR (95% CI) |
|
|---|---|---|---|
| Model 1, no adjustment for women’s feeding mode when they were infantsb | |||
| Catholic | 6.67 (1.72–25.93) | 2.50 (1.42–4.39) | 1.46 (1.07–2.01) |
| Protestant | 3.17 (0.83–12.13) | 1.55 (0.86–2.77) | 1.08 (0.78–1.50) |
| Jewish | 5.15 (0.94–28.31) | 1.51 (0.55–4.15) | 1.14 (0.63–2.06) |
| Other religion | 1.70 (0.18–15.70) | 0.93 (0.31–2.78) | 0.61 (0.30–1.26) |
| Unaffiliated | reference | reference | reference |
| Model 2, with adjustment for women’s feeding mode when they were infantsc | |||
| Catholic | 6.36 (1.56–26.02) | 2.36 (1.34–4.16) | 1.34 (0.98–1.83) |
| Protestant | 3.38 (0.84–13.50) | 1.58 (0.89–2.81) | 1.06 (0.77–1.46) |
| Jewish | 5.65 (0.93–34.43) | 1.57 (0.56–4.36) | 1.12 (0.61–2.04) |
| Other religion | 1.79 (0.19–16.61) | 0.94 (0.32–2.76) | 0.61 (0.30–1.25) |
| Unaffiliated | reference | reference | reference |
| Model 2, among non-Hispanic White women only | |||
| Catholic | 2.81 (0.86–9.17) | 2.33 (1.01–5.38) | 1.49 (0.96–2.32) |
| Protestant | 1.26 (0.34–4.65) | 1.42 (0.59–3.43) | 1.13 (0.70–1.81) |
| Jewish | 2.50 (0.48–12.99) | 1.67 (0.51–5.53) | 1.15 (0.58–2.29) |
| Other religion | 1.07 (0.14–8.40) | 2.03 (0.63–6.61) | 0.68 (0.27–1.70) |
| Unaffiliated | reference | reference | reference |
Values are risk ratios (95% CI) resulting from multivariable Poisson regression with robust error variance.
Models were adjusted for race/ethnicity, birthplace, age, education, annual household income, marital status, pre-pregnancy body mass index, pregnancy smoking status and parity.
Models were further adjusted for feeding mode when the woman was an infant.
Bold values denote statistical significance.
Odds of exclusive formula feeding and partial breastfeeding vs exclusive breastfeeding
Figure 2 shows the adjusted associations of women’s religious affiliation with infant feeding intention and practice as 3-category outcomes (exclusive formula feeding, partial breastfeeding and exclusive breastfeeding) in the overall sample. Compared to unaffiliated women, Catholic women were more likely to intend to exclusively formula feed in the first week postpartum (aOR [95% CI]: 9.31 [1.86‒46.49]) and at 3 months postpartum (7.10 [2.51‒20.08]), and of practicing formula feeding after delivery (2.78 [1.30‒5.94]). Infant feeding intentions declared after delivery showed similar results. Partial breastfeeding intention and practice, however, were not as consistently related to women’s religious affiliation.
Figure 2.
Adjusted associations of religious affiliation with intention/practice of infant feeding, Project Viva mother-child cohort, Massachusetts, 1999-2002.
The X-axis shows the odds ratios (95% CI) for partial breastfeeding (dark blue diamond) and exclusive formula feeding (bright blue square), with exclusive breastfeeding as reference. The Y-axis shows the odds ratios of Catholic and Protestant women, compared to unaffiliated women, across five time points of infant feeding intention/practice. BF, breastfeeding.
DISCUSSION
In this US pregnancy cohort with a majority of non-Hispanic White women, being of Catholic affiliation vs unaffiliated was associated with having themselves been formula fed as infants and with intention and practice of exclusive formula feeding as mothers of their own infants. These associations hold even after adjusting for whether the women were themselves breastfed and were consistent when restricting the analyses to non-Hispanic White women. Associations were less evident, however, when examining partial (vs exclusive) breastfeeding intention and practice. Being of Protestant, Jewish or other religious affiliation was not associated with infant feeding intention and practice, compared to being unaffiliated.
These findings are consistent with our previous work, based on a different research design. In an ecological study using data aggregated at regional and national levels, we reported that the higher the Catholic proportion in the population, the lower the breastfeeding initiation rates in that population.17 This correlation held in Western countries only, where breastfeeding initiation rates show a wider variability than elsewhere in the world. We also showed that in the US, the positive unadjusted correlation between state-level proportion of Catholics and rates of breastfeeding initiation was confounded by maternal race/ethnicity, education and socioeconomic status. After accounting for those factors, the state-level proportion of non-Hispanic White Catholics was negatively correlated with breastfeeding initiation rate among non-Hispanic White women, suggesting that the causal effect, if any, might be true only for non-Hispanic White women. Ecological studies can misrepresent associations at the individual level,18 but the similar findings from our current analysis of individual-level data from Project Viva, a US pregnancy-birth cohort comprising a majority of non-Hispanic White women with substantial proportions of Catholic, Protestant and unaffiliated religious backgrounds, now adds another layer of evidence. It is worth noting that the magnitude of the associations we observed is large (aRRs from 2 to 6; aORs from 2 to 9) when comparing the risk/odds of exclusive formula feeding in Catholic vs unaffiliated women. Our results also indicate that whether the women were themselves breastfed as infants does not modify or mediate the associations between religious affiliation and infant feeding practice, since the risk ratios were of similar magnitude with and without adjusting for this variable. Associations with antepartum infant feeding intention were, however, substantially reduced after adjusting for this variable, suggesting that antepartum intention to breastfeed is driven by family history.
Previously, a multi-center study by Burdette and Pilkauskas reported that women from Muslim, Conservative Protestant, and other Christian and non-Christian affiliations were more likely to ever breastfeed than women with no religious affiliation; findings were inconclusive regarding other affiliations, including Mainline Protestants, Baptists and Catholics.20 The population in that study, however, had a very different race/ethnicity mix from ours (47% non-Hispanic Black, 27% Hispanic and 22% non-Hispanic White women), and no ethnic-specific associations were observed. In a recent data analysis of the US National Longitudinal Survey of Youth 1979, Stroope et al. reported that women religiously unaffiliated or of other religion were more likely to initiate breastfeeding compared to Conservative Protestants.24 They found no differences with Catholic women after adjusting for educational attainment; they underline, however, a marginally significant interaction with Hispanic ethnicity, suggesting that ethnic-specific associations might exist. We are aware of no other studies reporting infant feeding intention and practice by religious affiliation.
Reasons why White Catholic women are less likely to breastfeed than White Protestant and unaffiliated women remain unclear. This difference was previously observed in European countries as early as the sixteenth century, long before waves of Europeans emigrated extensively to what became the USA.25 From the seventeenth century, the employment of wet nurses may have perpetuated the difference between Catholic and Protestant families on both sides of the Atlantic ocean. Indeed, the Catholic Church encouraged the hiring of wet nurses as a solution to overcome the perceived incompatibility between sexual intercourse and lactation, while English Puritans avoided wet nursing for both religious and sociocultural reasons.26 European Union official statistics show that women’s employment rate is higher in countries where Protestantism is widespread (73-80% in Germany, Switzerland, the Netherlands, Denmark and Sweden, vs 53-70% in Italy, France Spain, Portugal, Ireland and Poland). However, these statistics also indicate a gender gap in part-time employment rate that is wider in countries of Protestant background (i.e., women occupy more part-time jobs than men), especially when the number of children increases. Although women’s part-time employment could favor breastfeeding, our study does not support this explanation, since occupational status was not linked to religious affiliation or to infant feeding intention and practice. Other possible explanations include less societal acceptance of breastfeeding in public in predominantly Catholic countries, social representation of the breast as a nurturing vs sexual organ in those countries and differing beliefs regarding milk as a bodily fluid.27–30 Qualitative research in the US and other Western countries is required to test these and other hypothesized explanations.
Strengths of our study include its large sample of pregnant women of European ancestry and its longitudinal design, with information on religious affiliation collected in early pregnancy, i.e., prior to and separately from information on infant feeding. Many potential confounders were accounted for, and we were able to examine infant feeding intention separately from its practice and to differentiate exclusive breastfeeding, partial breastfeeding and exclusive formula feeding. The religious affiliation question made the important distinction between the two main Christian affiliations in a US state where both Catholicism and Protestantism are common among Whites.
Our study also has several limitations, however. First, we could not distinguish among Protestant subgroups, although our questionnaire included a specific checkbox for Unitarians. Many Protestant women specified their specific subgroup by text, but we have no information on the completeness of that reporting, and the sample sizes for subgroups are small. Another limitation is the small number and diverse religions and ethnicities of the non-White women in our study sample. Interaction tests and stratified analyses therefore had limited statistical power. Last, our findings may not be generalizable. Our study sample was not representative of the US or New England populations, and our data were collected nearly 20 years ago. Over that timeframe, both breastfeeding rates and the proportion of religiously unaffiliated people have increased in the US.12,31,32
In conclusion, our study comprising a majority of non-Hispanic White women confirms that Catholic women are more likely to exclusively formula feed their infants than women of Protestant, Jewish, other or no religious affiliation. This finding may help policymakers and health care professionals adapt their breastfeeding promotion policies to women’s sociocultural background. For example, midwives and lactation consultants may increase their support for non-Hispanic Catholic women by being more attentive to their potential concerns towards breastfeeding practice, and adapting the form and the content of their breastfeeding advice and guidance to better meet their sociocultural expectations with regard to infant feeding practices. Our findings would benefit from replication in populations from the US and other Western countries. Qualitative investigations are necessary to identify the sociocultural values and beliefs that make Catholic women more at risk of exclusive formula feeding in place of initiating breastfeeding.
Supplementary Material
Acknowledgment
We are indebted to the Project Viva mothers, children and families for their ongoing participation.
Funding
This work was supported by no specific funding. Project Viva was supported by grants from the US National Institutes of Health (R01 HD 034568, UG3 OD023286). Emmanuel Cohen was supported by the South African DST/NRF Centre of Excellence in Human development at the University of the Witwatersrand, Johannesburg, South Africa. The other authors were funded by their respective institutions.
Footnotes
Competing Interests
The authors declare no competing interests
REFERENCES
- 1.Victora CG, Bahl R, Barros AJ, et al. Breastfeeding in the 21st century: epidemiology, mechanisms, and lifelong effect. Lancet. 2016;387(10017):475–490. [DOI] [PubMed] [Google Scholar]
- 2.Sankar MJ, Sinha B, Chowdhury R, et al. Optimal breastfeeding practices and infant and child mortality: a systematic review and meta-analysis. Acta Paediatr. 2015;104(467):3–13. [DOI] [PubMed] [Google Scholar]
- 3.Horta BL, Victora CG, World Health Organisation. Short-term effects of breastfeeding: a systematic review on the benefits of breastfeeding on diarrhoea and pneumonia mortality. World Health Organization;2013. [Google Scholar]
- 4.Horta BL, de Mola CL, Victora CG. Breastfeeding and intelligence: systematic review and meta-analysis. Acta Paediatr. 2015. [DOI] [PubMed] [Google Scholar]
- 5.Bernard JY, Armand M, Peyre H, et al. Breastfeeding, Polyunsaturated Fatty Acid Levels in Colostrum and Child Intelligence Quotient at Age 5–6 Years. J Pediatr. 2017;183:43–50 e43. [DOI] [PubMed] [Google Scholar]
- 6.Belfort MB, Rifas-Shiman SL, Kleinman KP, et al. Infant feeding and childhood cognition at ages 3 and 7 years: Effects of breastfeeding duration and exclusivity. JAMA pediatrics. 2013;167(9):836–844. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kramer MS, Aboud F, Mironova E, et al. Breastfeeding and child cognitive development: new evidence from a large randomized trial. Arch Gen Psychiatry. 2008;65(5):578–584. [DOI] [PubMed] [Google Scholar]
- 8.Yang S, Martin RM, Oken E, et al. Breastfeeding during infancy and neurocognitive function in adolescence: 16-year follow-up of the PROBIT cluster-randomized trial. PLoS Med. 2018;15(4):e1002554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.World Health Organization. Global strategy for infant and young child feeding. World Health Organization; 2003. [Google Scholar]
- 10.Purdy J, McAvoy H, Cotter N. Breastfeeding on the island of Ireland. 2017; https://www.publichealth.ie/sites/default/files/Breastfeeding%20on%20the%20island%20of%20Ireland.pdf. Accessed Oct 31st, 2019.
- 11.Wagner S, Kersuzan C, Gojard S, et al. Durée de l’allaitement en france selon les caractéristiques des parents et de la naissance. résultats de l’étude longitudinale française elfe, 2011. [Breastfeeding duration in France according to parents and birth characteristics. Results from the ELFE longitudinal French study, 2011]. Bull Epidemiol Hebdo. 2015;29:522–532. [Google Scholar]
- 12.Centers for Disease Control and Prevention. Breastfeeding Report Card – United States, 2018. CDC National Immunization Surveys 2016 and 2017, among children born in 2015 2018; https://www.cdc.gov/breastfeeding/pdf/2018breastfeedingreportcard.pdf. Accessed Oct 31st, 2019.
- 13.Yngve A, Sjostrom M. Breastfeeding determinants and a suggested framework for action in Europe. Public Health Nutr. 2001;4(2B):729–739. [DOI] [PubMed] [Google Scholar]
- 14.Celi AC, Rich-Edwards JW, Richardson MK, Kleinman KP, Gillman MW. Immigration, race/ethnicity, and social and economic factors as predictors of breastfeeding initiation. Arch Pediatr Adolesc Med. 2005;159(3):255–260. [DOI] [PubMed] [Google Scholar]
- 15.Donath SM, Amir LH. Relationship between prenatal infant feeding intention and initiation and duration of breastfeeding: a cohort study. Acta Paediatrica. 2007;92(3):352–356. [PubMed] [Google Scholar]
- 16.Stuebe AM, Bonuck K. What predicts intent to breastfeed exclusively? Breastfeeding knowledge, attitudes, and beliefs in a diverse urban population. Breastfeed Med. 2011;6(6):413–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bernard JY, Cohen E, Kramer MS. Breast feeding initiation rate across Western countries: does religion matter? An ecological study. BMJ Glob Health. 2016;1(4):e000151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Messerli FH. Chocolate consumption, cognitive function, and Nobel laureates. N Engl J Med. 2012;367(16):1562–1564. [DOI] [PubMed] [Google Scholar]
- 19.Maurage P, Heeren A, Pesenti M. Does chocolate consumption really boost nobel award chances? The peril of over-interpreting correlations in health studies. J Nutr. 2013;143(6):931–933. [DOI] [PubMed] [Google Scholar]
- 20.Burdette AM, Pilkauskas NV. Maternal religious involvement and breastfeeding initiation and duration. Am J Public Health. 2012;102(10):1865–1868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.U.S. Census Bureau. American Community Survey (ACS), 2011-2015 ACS 5-year Estimates. 2016. Accessed August 1st, 2018.
- 22.Oken E, Baccarelli AA, Gold DR, et al. Cohort profile: project viva. Int J Epidemiol. 2015;44(1):37–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zou G A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702–706. [DOI] [PubMed] [Google Scholar]
- 24.Stroope S, Rackin HM, Stroope JL, Uecker JE. Breastfeeding and the Role of Maternal Religion: Results From a National Prospective Cohort Study. Ann Behav Med. 2018;52(4):319–330. [DOI] [PubMed] [Google Scholar]
- 25.Castilho SD, Filho AdAB. The history of infant nutrition. Jornal de Pediatria. 2010;86(3):179–188. [DOI] [PubMed] [Google Scholar]
- 26.Golden J A social history of wet nursing in America: From breast to bottle. Ohio State University Press; 2001. [Google Scholar]
- 27.Scott JA, Kwok YY, Synnott K, et al. A comparison of maternal attitudes to breastfeeding in public and the association with breastfeeding duration in four European countries: results of a cohort study. Birth. 2015;42(1):78–85. [DOI] [PubMed] [Google Scholar]
- 28.Amir LH. Breastfeeding in public: “You can do it?”. Int Breastfeed J. 2014;9(1):187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Dettwyler KA. Beauty & the Breast: The Cultural Context of Breastfeeding in the United States. Breastfeeding: Biocultural Perspectives: Aldine Transaction; 1995:167–215. [Google Scholar]
- 30.Dowling SJ. Exploring the experiences of women who breastfeed long-term. Bristol, UK: Faculty of Health and Applied Sciences, University of the West of England; 2013. [Google Scholar]
- 31.Centers for Disease Control and Prevention. Breastfeeding Report Card – United States, 2010. National Immunization Survey (2010) 2010; http://www.cdc.gov/breastfeeding/pdf/breastfeedingreportcard2010.pdf.
- 32.Pew Research Center. In U.S., smaller share of adults identify as Christians, while religious ‘nones’ have grown. 2019; https://www.pewforum.org/2019/10/17/in-u-s-decline-of-christianity-continues-at-rapid-pace/pf_10-17-19_rdd_update-00-020/. Accessed Nov 11, 2019.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.