Significance

What is already known on this subject? The decrease in U.S.-Hispanics’ breastfeeding behaviors might be attributed to acculturation. However, the published evidence that associates acculturation with a decrease in Hispanics’ breastfeeding behaviors is not conclusive.

What does this study add? To date a variety of methodologies have been used in lactation studies. For example, authors have used different definitions of breastfeeding, have not utilized optimal scales to assess acculturation, in multivariable analyses have missed important covariates, and not all studies differentiated between Hispanic subgroups. Future studies should utilize bi/multidimensional scales when studying breastfeeding, in particular, in relation to Hispanics’ intention and exclusive breastfeeding at 6 months, and preferably by country of origin.

Introduction

The World Health Organization (WHO) recommends that “newborns be exclusively breastfed for at least 6 months with continued breastfeeding along with appropriate complementary foods up to 2 years of age or beyond” (WHO 2011). In the United States (U.S.), although the breastfeeding initiation rate is high (82.5%), the exclusive and long-term breastfeeding rates are low, as approximately 75% of newborns are not exclusively breastfed for at least 6 months (Centers for Disease Control and Prevention [CDC] 2014). Hispanics’ breastfeeding rates have increased in the past decade and have reached U.S. national rates; 84.8% of Hispanic children born in 2014 were ever breastfed and only 24.5% were exclusively breastfed until 6 months (CDC 2014). However, observational studies indicated that breastfeeding rates for Hispanics deteriorate with increasing years living in the U.S. (Harley et al. 2007) and that Hispanic immigrants are significantly more likely to breastfeed than are Hispanics born in the U.S. (Kimbro et al. 2008). Gibson-Davis and Brooks-Gunn (2006) found that for each additional year living in the U.S., the odds of ever breastfeeding decreased by 4% for Hispanics. Numerous studies have partly attributed this decrease to acculturation (Gibson et al. 2005; Gorman et al. 2007), a process whereby individuals from one culture adopt the beliefs and behaviors of another culture (Chakraborty and Chakraborty 2010).

Three models in the literature have attempted to conceptualize the acculturation process: the unidimensional model, which assumes a straight-line process toward assimilation; the bidimensional model, which subsumes individuals who adopt certain aspects of the host culture while continuing to value and retain the norms of their original culture (Berry 2003); and the multidimensional model, which includes cultural changes in several life dimensions such as practices, values, and identification (Chakraborty and Chakraborty 2010). In trying to assess the acculturation process in observational studies of Hispanics, several scales were developed ranging from proxy measures to more complex ones. In breastfeeding studies some scholars use proxy measures to roughly assess some degree of acculturation such as Hispanics’ language preference (Ahluwalia et al. 2012) and/or birth place (Balcazar et al. 1995), whereas other scholars assess deeper layers of acculturation, such as Hispanics’ attitudes about their gender roles and their cultural engagement (Kimbro et al. 2008). Moreover, other researchers combined measures (Sussner et al. 2008) such as nativity, generation status, time lived in the US, and language use.

Several review studies synthesized and summarized the impact of acculturation on Hispanics’ health behaviors (Arandia et al. 2012; Lara et al. 2005), and suggested integrating cultural elements into health promotion programs and provided recommendations for future scholars. However, a summary of the current knowledge on acculturation and breastfeeding in Hispanics has not yet been published. With the goal of filling this need, we summarized all the quantitative studies that aimed to examine the association between Hispanics’ breastfeeding behaviors and acculturation, focusing on advancing our understanding of the complexity of this link, shedding light on the significant gaps in this research area, and suggesting directions for future studies.

Methods

Search Strategy

This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses and used the PRISMA checklist, in reporting our systematic review. The PRISMA checklist was developed to provide clear, practical, and systematic guidance in researching and writing systematic reviews (Moher et al. 2015). Comprehensive electronic searches, with the help of a professional librarian to identify suitable publications, were conducted from January 9 to 11, 2018 using the main health databases: Medline-EBSCO, Academic Search Complete-EBSCO, PubMed, and other relevant databases such as PsycInfo-EBSCO, and the Cochrane library (Table 1). Additionally, we manually searched for articles published in selected professional journals, revised the list of references in selected articles, and chose the first 100 most relevant citations in Google scholar (Table 1). The main key concepts are based on MeSH keywords for ‘breastfeeding,’ ‘Hispanics,’ and ‘acculturation,’ which were mentioned in the title and/or abstract and/or the full text. Hispanics followed the definition of Hispanic-Americans given by MeSH: “Persons living in the U.S. of Mexican (Mexican-Americans), Puerto Rican, Cuban, Central or South American, or other Spanish culture or origin. The concept does not include Brazilian-Americans or Portuguese-Americans”. Table 2 includes all the “search terms” that were employed by inserting them simultaneously in each of the electronic databases. Within term groups (i.e., acculturation) the search terms were divided by ‘or’, whereas the two term groups were connected by ‘and’.

Table 1 Systematic review of the main database and key journals
Full size table
Table 2 Search terms used for the systematic review by main key concepts
Full size table

Eligibility

All citations were stored in the reference manager database Refworks (Hendrix 2004). Eligibility criteria were created by all the authors (GB, AW, AP, and NH) and are presented in Table 3. After the duplicate citations were removed, the screening and selecting processes were as follows: in the first round, articles were excluded based upon their title, then, in the second round, articles were excluded based on their abstract. In the final round, full-text papers were obtained and were carefully assessed. All three rounds were conducted independently at two separate times, during November–December 2015 (at that time generating 16 reviewed articles) and during January 2018 by the leading author, GB. The selected reviewed studies and data extraction were confirmed by all the authors and any disagreement was settled by a discussion (Bown and Sutton 2010). The literature search and article selection are presented in a flow chart (Fig. 1). Overall, 1943 citations were downloaded and 686 duplicates were removed. Based on the inclusion and exclusion criteria (Table 3), 18 articles met the eligibility criteria and 1239 citations were excluded, as described in detail in Fig. 1.

Table 3 Inclusion exclusion criteria for selecting potential articles for the final review
Full size table
Fig. 1
figure 1

Article selection process following PRISMA guidelines

Full size image

Data Extraction and Quality Assessment

We developed an electronic data extraction form and extracted information from each reviewed study based on predefined data fields, as described in the headings in Table 4.

Table 4 Summary of studies examining the association of acculturation and breastfeeding behaviors among Hispanic Americans by acculturation proxy measure and models and their final quality assessment score
Full size table

For quality assessment, we used an adapted version of the Quality Assessment Tool for Quantitative Studies developed by the Effective Public Health Practice Project (Thomas et al. 2004). This tool consists of eight key domains (Table 5) with a score ranging from low (1) to high (3). The reviewer determined an overall rating for each reviewed study: strong, moderate, or weak final scores (Table 4).

Table 5 Quality assessment tool for quantitative studies
Full size table

Results

Quality and Characteristics of the Reviewed Studies

The quality evaluation results are presented in Fig. 2 and Table 4. None of the reviewed papers were ranked as a ‘weak’ in their final score and therefore, they were not eliminated from the final review. The 18 reviewed studies utilized a prospective study design (nine articles), a cross-sectional design (eight articles), and one utilized a retrospective study design. Overall 12 studies employed primary data from Hispanic communities, mainly low-income communities, and six studies analyzed secondary data from national surveys. Mexican-Americans were the most studied Hispanic subgroup; 12 studies included Mexican-descent populations, four studies included also Puerto Ricans, three studies included Dominicans, and three studies did not distinguish Hispanics by their country of origin.

Fig. 2
figure 2

Review of authors’ judgments about each methodological quality item presented as percentages across all included studies

Full size image

We organized and presented the study findings (Table 4) according to proxy measures and main models of acculturation: unidimensional, bidimensional, and multidimensional, which were to some extent linked to the following breastfeeding behaviors: intention, initiation, and duration. Overall, seven studies (39%) used proxy measures to assess acculturation and included the following items: language preferences, nativity, and years living in the U.S. or a combination of those items. The acculturation scales were designed and validated for the Hispanic-American population and were assessed as a unidimensional model in eight studies (44%); two studies (11%) assessed the bidimensional model, and only one study (6%) employed the multidimensional model of acculturation.

In the multivariable models the majority of the reviewed studies included well-established covariates such as maternal age and education, parity, marital status, delivery mode, and household income or other associated factors such as receiving food stamps or participating in the Special Supplemental Nutrition Program for Women Infants and Children (WIC). Some studies also included the type of prenatal care and the number of visits, birth weight, working status, and smoking experience (Table 4). Overall, 13 studies (73%) showed a significant association between acculturation and at least one of the following breastfeeding behaviors: intention, initiation, and duration. However, five studies (27%) did not reveal any significant associations, of those studies; two studies included predominantly Dominican-Republicans (Glassman et al. 2014; Thiel de Bocanegra 1998), and one comprised only Puerto-Ricans (Anderson et al. 2004).

Study Findings

Acculturation and Breastfeeding Intention

Hispanics’ intention to breastfeed and their relationship with acculturation was examined in six studies: three studies used the unidimensional model of acculturation, three studies examined a proxy measure, and none utilized the bi/multidimensional models. Among the three studies that found that high-acculturated women were significantly less likely to intend to breastfeed compared to low-acculturated women, two studies also revealed mixed findings when acculturation was assessed utilizing more than one measure. For example, Byrd et al. (2001), who utilized four separate proxy measures to assess acculturation among Mexican-origin women, found that women’s nativity and the country in which they completed their formal education were associated with their intention to breastfeed, whereas language proficiency was not significantly associated with intention to breastfeed. Barcelona de Mendoza et al. (2016), who conducted a study among Puerto Rican women, only obtained significant findings when acculturation was measured as a proxy measure (e.g., generation status, language proficiency) but not when acculturation was assessed as a unidimensional scale.

Acculturation and Breastfeeding Initiation

The association between acculturation and Hispanics’ breastfeeding initiation was examined in 11 studies; four studies utilized the unidimensional model, six studies used proxy measures, and only one study used the multidimensional model to assess acculturation. Overall, eight studies showed significant and similar results, namely, that high-acculturated Hispanic women were less likely to initiate breastfeeding compared with their low-acculturated counterparts. However, the breastfeeding initiation definition varied between studies. For example, Rassin et al. (1993) defined breastfeeding initiation as success in breastfeeding up to 2–3 weeks after the delivery, whereas Gorman et al. (2007) defined breastfeeding initiation as exclusive breastfeeding upon discharge from the hospital, and another defined breastfeeding initiation as having breastfed the last infant once (Hawkins et al. 2014). Some studies showed mixed results when more than one measure of acculturation was examined. For example, Sussner et al. (2008) showed that Hispanic women who spoke only in their native language were more likely to initiate breastfeeding compared with Hispanic women who did not, whereas the years of U.S. residence and women’s nativity were not significantly associated with breastfeeding initiation. Kimbro et al. (2008), who employed the multidimensional model in a Mexican–American sample, found that women who had a higher score in cultural engagement and church attendance were more likely to initiate breastfeeding, whereas women who held more traditional gender views were less likely to initiate breastfeeding, and no significant differences were observed in breastfeeding initiation rates between women who completed the interview in Spanish vs. English.

Acculturation and Breastfeeding Duration

Acculturation and its association with breastfeeding duration by Hispanic women was evaluated in eight studies: acculturation was assessed in three studies as a bi/multidimensional model, and four studies used the unidimensional or proxy measures. Seven studies showed that shorter breastfeeding duration could be attributed to acculturation. For example, Gibson et al. (2005) used the Short Acculturation Scale to assess the unidimensional model of acculturation in which Hispanic women were categorized as having either a low or high acculturation level. After controlling for maternal education, age, and the household poverty income ratio, high-acculturated women were 77% less likely to breastfeed, compared with low-acculturated women. In addition, the duration of exclusive breastfeeding, in accordance to the WHO definition (WHO 2008), meaning that the infant received only breast milk, was analyzed in four of the studies reviewed (Ahluwalia et al. 2012; Chapman and Perez-Escamilla 2013; Glassman et al. 2014; Harley et al. 2007), and of these, only two examined exclusive breastfeeding up to 6 months (Chapman and Perez-Escamilla 2013; Harley et al. 2007). Furthermore, Chapman and Perez-Escamilla (2013) assessed acculturation among low-income Hispanics in two ways: as a unidimensional and bidimensional model utilizing the Acculturation Rating Scale for Mexican-Americans (ARSMA-II). In the unidimensional model, ARSMA-II was modified to a linear score and no significant findings were found. However, when ARSMA-II was divided into four distinct subgroups to assess the bidimensional model, breastfeeding duration rates varied significantly between the following subgroups: integrated high (23.7%, n = 27), traditional Hispanic (36.8%, n = 42), integrated low (12.3%, n = 14), and assimilated (27.2%, n = 31). Similar findings were observed when the bidimensional model was utilized by Luecken et al. (2017) to assess the relationship with breastfeeding at 3 months for a Mexican-origin population.

Discussion

The current systematic review is the first study to summarize all the published literature and the quantitative evidence explicitly aimed at examining the relationship between acculturation and breastfeeding behaviors in Hispanics. We retrieved 18 studies that met the eligibility criteria, and with the exception of five studies, all other reviewed studies reported overall similar findings; high-acculturated Hispanic women were less likely to intend to or breastfeed their newborn, compared with low-acculturated Hispanic women. Our review is in line with other diet and nutritional reviews, which repetitively demonstrated the cultural transition that Hispanics experience while immigrating to the U.S., which results in decreased consumption of healthy and nutritional foods in favor of fatty and processed foods (Arandia et al. 2012; Lara et al. 2005). We assume that the impact of acculturation on breastfeeding might be partially explained as a result of Hispanics’ widespread exposure to substitutes for human-milk via the American media and/or through indirect promotion of formula feeding in hospitals, which highly acculturated Hispanics might perceive as the preferred American infant feeding method (Office of the Surgeon General 2011). Other aspects that could be contributing and deserve further consideration are job-related demands and the availability of family and/or peer support.

However, a closer look at the reviewed studies, in particular, at the methods and findings, raised some issues and suggestions that we would like to address. First, these studies assessed acculturation in various ways, challenging our ability to completely understand the mechanisms by which acculturation influences Hispanics’ decision to breastfeed and for how long. In addition, 15 studies used a static measure (proxy measure) or a unidimensional model (linear-scales), which is not optimal, and only three studies captured more comprehensive aspects of acculturation by utilizing the bi/multidimensional scales. Recent studies have criticized the unidimensional model, claiming that it does not differentiate between individuals who are equally acculturated to both their origin and host cultures and individuals who are acculturated to neither culture (Berry 2003). Andrews et al. (2013), with the aim of studying the most appropriate model of acculturation in Hispanics, found that the bi/multidimensional models are a better fit for describing the acculturation process than is the unidimensional model.

Second, most of the reviewed studies that adopted a multivariable model, in which household income and other associated factors were controlled for, concluded that the impact of acculturation on breastfeeding could not be attributed to improved socioeconomic status, as it might be argued among other scholars. However, the studies did not adjust for other key risk factors or confounders such as previous breastfeeding experience, social support from practitioners and family members, prenatal education about breastfeeding, and preterm delivery.

Third, from the reviewed findings, the extent of influence that acculturation has on Hispanics’ decision to breastfeed or use formula, is not yet completely clear. Future studies might consider using bi/multidimensional models of acculturation and the relationship with the intention to breastfeed, since they are likely to result in a better understanding when in the acculturation process Hispanic women decide not to breastfeed. In addition, the U.S. national rate for exclusive breastfeeding at 6 months is still low, including Hispanics (CDC 2014). Interestingly, only two scholars have explored Hispanics’ exclusive breastfeeding at 6 months (Chapman and Perez-Escamilla 2013; Harley et al. 2007), which highlights the need for more studies to investigate, preferably prospectively, the cultural barriers to exclusively breastfeed at 6 months among Hispanics.

Fourth, five reviewed studies did not show significant findings (Anderson et al. 2004; Balcazar et al. 1995; Flores et al. 2016; Glassman et al. 2014; Thiel de Bocanegra 1998) and explained their results due to the following reasons. For example, Flores et al. (2016) collected data from Mexican-Americans living in El Paso, a city on the border with Mexico, and found that in their study sample breastfeeding habits and rates were similar to breastfeeding habits and rates in Mexico compared to U.S -Hispanics and therefore, no significant differences in breastfeeding rates appeared across levels of acculturation. Thiel de Bocanegra (1998), in their bivariate analysis, showed that high-acculturated women were two times less likely to decide to breastfeed than were low-acculturated women. However, the influence of acculturation lessened when the independent variables of “support by friends and family members” and “tobacco use” were included in the multivariable analysis. Lastly, in most of the studies that did not attribute variations in breastfeeding rates to acculturation, their study population predominantly originated from the Dominican Republic and/or Puerto Rico. We assume that Hispanic subgroups might experience a different immigrant trajectory, which has implications on their breastfeeding behaviors and therefore, requires the use of a more refined acculturation scale appropriate for the different Hispanic subgroups.

Lastly, recent studies show that U.S.-Hispanics and specifically, Mexican-Americans and Puerto Ricans, prefer their babies to be big because they consider it as a sign of having a healthy baby and therefore, do not consider exclusive breastfeeding to be a viable option for feeding their infants (Flores et al. 2016). Hispanics believe that complementary food (i.e., infant formula, and cereal) at an earlier stage is essential for ensuring that their baby will grow and develop healthily (Glassman et al. 2014). Therefore, U.S.-Hispanic women choose the ‘los-dos’ (“best of both”) feeding practices, which is a combination of both formula and breast milk, as being best for their babies. Combining breastfeeding with formula is a very common feeding practice among Latinas including in their countries of origin and one of the causes of the low exclusive breastfeeding rates, in Mexico for example (Gonzalez de Cosio et al. 2013). In the currently reviewed studies, however, the ‘los-dos’ feeding practice was explicitly analyzed in only two studies (Barcelona de Mendoza et al. 2016; Flores et al. 2016).

Strengths and Limitations

One strength of this review relates to the focus of the paper, which was to identify and review all published peer-reviewed evidence that explicitly examined the relationship between acculturation and breastfeeding among Hispanic-Americans. The searching process was comprehensive; it used leading databases, which are all from the U.S. National Library of Medicine. Another strength of this review is that we summarized the findings according to acculturation models to reveal the lack of a standard methodology in quantifying the acculturation levels. We found that each breastfeeding behavior (intention, initiation, and duration) requires a different research method approach and that some infant feeding practices require further investigation such as the intention to breastfeed, exclusive breastfeeding at 6 months, and ‘los-dos’.

A limitation of this review relates to the fact that all of the reviewed studies focused only on those studies that explicitly aimed to examine breastfeeding behaviors and acculturation and excluded studies that examined an association between breastfeeding behaviors and language spoken (Lizarraga et al. 1992), nativity (Singh et al. 2007), years of U.S. residency (Gibson-Davis and Brooks-Gunn 2006), and country of education (Hendrick et al. 2017), which are all valid proxy measures for acculturation. However, since those studies were not explicitly aimed at examining the process of acculturation, they were not selected in this review. In addition, we did not include publications in Spanish that might contain important and pertinent information about breastfeeding among Hispanics. Lastly, this systematic review used a less robust strategy for study selection and data extraction (Bown and Sutton 2010), which was performed by a single researcher at two separate times, whereas the gold standard requires multiple individuals (Liberati et al. 2009). However, all the authors participated in establishing the eligibility criteria and approved the final study selection, and any discrepancies regarding data extraction, study results, and conclusions were resolved by consensus.

Conclusions

Our review supports the hypothesis that higher acculturation leads to lower breastfeeding rates, and it appears that this association is independent of income levels. Along with the Surgeon General’s call to improve national breastfeeding rates, we recommend that future lactation studies use bi/multidimensional scales while following the WHO definitions for breastfeeding, to generate more valuable findings on the association between acculturation and Hispanics’ breastfeeding behaviors. Understanding which of the cultural domains influence the following Hispanic women’s breastfeeding behaviors: intention, initiation, and duration, will help improve the design of culturally appropriate interventions tailored to prenatal care and each breastfeeding phase. In addition, further studies are needed to address acculturation with respect to breastfeeding among all the U.S.-Hispanic subgroups in order to present an overall picture of breastfeeding rates in the U.S. Lastly, acculturation studies should address breastfeeding outcomes, particularly the intention to breastfeed, exclusive breastfeeding at 6 months, and the ‘los-dos’ approach which is a typical and common infant feeding practice in Hispanic women.