|Home | About | Journals | Submit | Contact Us | Français|
We sought to assess the relationship between acculturative type and breastfeeding outcomes among low-income Latinas, utilizing a multidimensional assessment of acculturation.
We analyzed data derived from a breastfeeding peer counseling randomized trial. Acculturation was assessed during pregnancy using a modified ASRMA-II scale. Analyses were restricted to Latinas who completed the acculturation scale and had postpartum breastfeeding data (N=114). Cox survival analyses were conducted to evaluate differences in breastfeeding continuation and exclusivity by acculturative type.
Participants were classified as integrated-high (23.7%, n=27), traditional Hispanic (36.8%, n=42), integrated-low (12.3%, n=14) and assimilated (27.2%, n=31). The integrated-low group was significantly more likely to continue breastfeeding than the Traditional Hispanic, Assimilated, and Integrated-high groups (p<0.05, p<0.05, and p<0.01, respectively). The Traditional Hispanic group was marginally more likely to continue breastfeeding than the integrated-high group (p=0.06).
Breastfeeding continuation rates vary significantly between acculturative types in this multinational, low-income Latina sample. Multidimensional assessments of acculturation may prove useful in better tailoring future breastfeeding promotion interventions.
While Latinas in the United States have slightly higher rates of breastfeeding initiation than the national average (Centers for Disease Control and Prevention, 2010a), Latinas who choose to breastfeed are significantly more likely to supplement with formula in the first 2 days of life, compared to black or Caucasian US mothers (Centers for Disease Control and Prevention, 2010b). The breastfeeding practices of US Latinas are impacted by numerous factors, some of which are specific to their culture (Bunik et al., 2006, Higgins, 2000, Newton et al., 2009, Pérez-Escamilla et al., 1998). For example, US Latinas preference for a “big” baby often results in the supplementation of breastfed infants, while the traditional use of teas for colic or constipation interferes with exclusive breastfeeding (Bunik et al., 2006).
While national data frequently report Hispanics or Latinas as a single group, this ethnic category represents women from multiple countries, with varying lengths of US residence. US Census data indicated that Latinas living in the US report their ethnic origins as being from Mexico (66.9%), other central and South American countries (14.3%), Puerto Rico (8.6%), other countries (6.5%) and Cuba (3.7%) (Ramirez and de la Cruz, 2002). Thus, US Latinas are not a homogenous group, and healthcare providers need to be wary of assuming that a single breastfeeding promotion approach is effective among all Latinas. It has been shown, for example, that Puerto Rican women have lower rates of exclusive breastfeeding, compared to women from Central or South America (Pachon and Olson, 1999). Typically, Latinas who are foreign born and those who have lived in the US for a short period of time have better breastfeeding practices than those who have spent all or most of their life in the US (Gibson-Davis and Brooks-Gunn, 2006, Harley et al., 2007, Merewood et al., 2006, Pachon and Olson, 1999, Pérez-Escamilla et al., 1998, Romero-Gwynn and Carias, 1989, Singh et al., 2007).
These differences in breastfeeding practices within the Latina population are likely related to acculturation, which has been defined as “the process by which immigrants adopt the attitudes, values, customs, beliefs and behaviors of a new culture” (Abraido-Lanza et al., 2004). Frequently, lactation researchers have assessed acculturation with static, proxy measures such as birthplace (Gibson-Davis and Brooks-Gunn, 2006, Merewood et al., 2006, Newton et al., 2009, Pachon and Olson, 1999, Romero-Gwynn and Carias, 1989, Singh et al., 2007), language spoken (Gorman et al., 2007) or years in the US (Gibson-Davis and Brooks-Gunn, 2006, Harley et al., 2007, Pachon and Olson, 1999, Pérez-Escamilla et al., 1998). Some breastfeeding researchers have simultaneously assessed multiple acculturation proxies (Anderson et al., 2004, Byrd et al., 2001, Gibson et al., 2005, Gibson-Davis and Brooks-Gunn, 2006, Pérez-Escamilla et al., 1998, Rassin et al., 1994, Sussner et al., 2008), in order to identify acculturation-related variables that are associated with breastfeeding practices. To better assess acculturation, other researchers have developed linear scales, which contain combinations of these indicators that are ranked to identify a participant along a continuum ranging from a strongly Hispanic orientation through a strongly American/Anglo orientation (Burnam et al., 1987, Gibson et al., 2005, Norris et al., 1996, Rassin et al., 1993, Rassin et al., 1994, Marin et al., 1987).
However, these static measures and linear scales are not optimal, because acculturation is impacted by the degree to which immigrants maintain their former identity, as well as by the extent to which they adopt the cultural practices of their new homeland (Pérez-Escamilla and Putnik, 2007). Multidimensional assessments of acculturation have been developed (Cuellar et al., 1995, Marin and Gamba, 1996), which may provide a better understanding of how acculturation is related to health outcomes. We hypothesize that the breastfeeding practices of low-income Latinas will vary, based on both the level of assimilation into the US culture and the degree to which they retain traditional Hispanic cultural practices. The objective of this study was to assess the relationship between acculturation and breastfeeding duration (any and exclusive) among low-income Latinas, utilizing a multidimensional assessment of acculturation.
We analyzed data derived from a randomized trial of a specialized, breastfeeding peer counseling intervention promoting exclusive breastfeeding among overweight and obese, low-income women (Chapman et al., 2008). The randomized trial was conducted in Hartford, CT from September, 2006 through January, 2010. Briefly, that study involved the recruitment of 206 pregnant, overweight or obese, low-income women who were considering breastfeeding, with randomization to receive either specialized breastfeeding peer support or standard care at this Baby Friendly Hospital. Women in the intervention group received specialized breastfeeding support promoting exclusive breastfeeding, which specifically addressed issues of relevance to overweight/obese women through 3 prenatal home visits, daily in-hospital visits, up to 11 postpartum home visits, unlimited telephone support, the provision of a breastfeeding sling, and free breast pump (if needed). Standard care at this hospital included the option to receive routine breastfeeding peer counseling, which involved up to 3 prenatal visits, daily in-hospital visits, up to 7 home visits in the first year postpartum, telephone support, and limited access to a breast pump. Women in both groups received routine breastfeeding assistance from staff nurses, and an Internationally Board Certified Lactation Consultant if necessary. Data collection for all participants included a phone interview at 36 weeks gestation, medical record review after delivery, and monthly telephone calls through 6 months postpartum to assess infant feeding practices. The current analyses include data from the 114 Latinas in this sample who delivered healthy, term infants, had no contraindications to breastfeeding, and completed the acculturation assessment. Ethical approval was received from the Institutional Review Boards at the University of Connecticut, Hartford Hospital, and the Hispanic Health Council.
Acculturation was assessed during the 36 week gestation phone interview, using a modified version of the Acculturation Rating Scale for Mexican Americans, (ARSMA-II, Scale 1) (Cuellar et al., 1995). As originally described by Cuellar et al, this multidimensional scale had 30 items, divided into a Mexican Orientation subscale and an Anglo orientation subscale. These subscales assess the frequency and enjoyment of conducting several activities (ie: reading, writing, speaking, thinking) in each language (Spanish or English). The original ARSMA-II used a 5 point Likert scale, ranging from Not at all to Extremely often/Almost always for each item.
The ARSMA-II was modified based on pilot testing among 10 low-income Latinas (5 English speaking, 5 Spanish speaking). The term “Mexican” was changed to “Hispanic” to meet the needs of our multi-national Latina population. Other modifications included changing “Anglo American” to “American”, changing the statements to questions, and reducing the number of items to those that were most relevant for this population. Minor changes in word selection were made. The modified scale had 10 items assessing American orientation and 10 items evaluating Hispanic orientation. The response was changed to a 4 point Likert scale (Never, A little, Some, A lot), which was coded as scores ranging from 1 through 4.
The scoring of the modified ARSMA-II was conducted using methods described by Cuellar et al (Cuellar et al., 1995). First, a linear acculturation score (LAS) was generated. This was calculated as the mean score for the American Orientation Subscale (AOS) minus the mean score from the Hispanic Orientation Subscale (HOS). This score assesses acculturation in a linear fashion, with scores ranging from a strong Hispanic orientation (−3.0) to a strong American orientation (3.0). LAS results were used to categorize participants as “Bicultural” (LAS = mean LAS ± 0.5 SD), “More Hispanic” (LAS > 0.5 SD below the mean LAS), and “More American (LAS > 0.5 SD above the mean LAS).
The second method of scoring the ARSMA-II involved a multidimensional classification on a 2-dimensional grid, in which the American orientation and Hispanic orientation scores are plotted on the × and y-axes, respectively. Individuals scoring above the sample mean on both the American and Hispanic subscales were classified as having a high level of integration (“Integrated, high”). Those scoring below the mean on the American scale, but above the mean on the Hispanic scale are considered “Traditional Hispanic”. Those scoring below the mean on both scales are considered to have a low level of integration (“Integrated, low”). While those who score above the mean on the American scale and below the mean on the Hispanic scale are labeled “Assimilated”.
Cronbach’s alpha was used to test the reliability of the revised subscales. Chi square and ANOVA were used to evaluate potential baseline differences between acculturation groups. When the ANOVA assumption of homogeneity of variances was violated, the more conservative Welch’s F statistic was used to compare the equality of means. Conservative post-hoc tests were used to compare all possible combinations of between group comparisons evaluating group means for continuous variables. The Games-Howell post hoc test was used when the ANOVA homogeneity of variances assumption was not met. Otherwise, the Hochberg post-hoc test was used to control for differing sample sizes between comparison groups.
We used Cox survival curves to compare the percent of women breastfeeding (exclusively and to any extent) by acculturative types. Cases which were still breastfeeding at the time of the last available follow-up were right-censored. Unadjusted Cox survival models were first run evaluating the relationships between acculturative type and: a) breastfeeding duration; and b) exclusive breastfeeding duration. These analyses were then repeated, controlling for relevant covariates. In order to identify relevant covariates, bivariate analyses (Student’s t-test, Mann Whitney U test (for non-normal distributions), and Chi square analyses) were conducted to identify demographic and biomedical variables associated with: breastfeeding continuation at 2 months postpartum (based on the median) and exclusive breastfeeding at 24 hours postpartum (based on the median). Variables which were significant in these bivariate analyses but collinear with the acculturative type variable (length of time in the US, and Hispanic origins), were not included in the adjusted Cox survival models.
Exclusive breastfeeding was defined in accordance with the WHO (World Health Organization, 2002), and excluded the provision of formula, water, other supplemental liquids, and solid foods since birth. All analyses were completed using SPSS, with p<0.05 indicating statistical significance.
The acculturation scale was completed by 114 of the 125 eligible Latinas in the sample, as 11 Latinas could not be reached for the prenatal phone interview. The mean score for the modified HOS was 3.33 ± 0.59, with a Cronbach’s alpha of 0.83, indicating good scale reliability. The mean score on the modified AOS was 2.65 ± 1.04. The Cronbach’s alpha was 0.96, reflecting good reliability. The mean linear acculturation score, (mean AOS - mean HOS) was −0.69 ± 1.36, indicating an overall stronger Hispanic orientation in this population. Using the linear acculturation scores, 39.5% (n=45) were classified as More Hispanic, 25.4% (n=29) were Bicultural, and 35.1% (n=40) were More American.
Table 1 shows the demographic and biomedical characteristics of participants, classified as More Hispanic, Bicultural, and More American based on LAS. As would be expected, there were significant differences between these 3 acculturation groups. Those classified as “More American” were significantly younger (p<0.01), and had more education (p<0.01) than those who were “More Hispanic”. All of the acculturation groups were significantly different from each other with regards to the length of time in the US, with increased levels of acculturation associated with more time in the US (More Hispanic vs Bicultural, p<0.01; More Hispanic vs More American, p<0.001; Bicultural vs More American, p<0.001). There were significant between group differences in employment status, country of Hispanic origin, and the participants’ history of being breastfed as an infant. There were no significant differences in delivery mode, study group assignment, or WIC participation between these 3 acculturation groups.
Of the 114 Latinas in the analytical sample, 96 had complete breastfeeding data (ie. were followed up until they stopped breastfeeding). The remaining 18 participants were right censored in the Cox survival analyses, because they were lost to follow-up and breastfeeding status was only known until the point of last contact. Nine of these participants were lost to follow-up after the day 1 postpartum interview. The remaining 9 participants were lost to follow-up between 0.5 and 5.0 months (mean = 2.7 months). There were no significant differences between those with complete breastfeeding data and those lost to follow-up for the subject characteristics shown in Table 1, except for maternal education. Women lost to follow-up had completed significantly more years of school than those with complete breastfeeding data (11.7 ± 1.4 vs 10.5 ± 3.1 years, respectively; p<0.05). There were no significant differences between the acculturative type distributions (linear and multidimensional) of those with complete breastfeeding data versus those lost to follow-up.
Breastfeeding initiation was nearly universal in this sample (98.2%), as the inclusion criteria specified that participants must be planning to breastfeed or considering breastfeeding their infant. Thus, analyses using breastfeeding initiation as a dependent variable were not possible.
The median duration of breastfeeding was 2.1 months. Univariate analyses indicated that women still breastfeeding at 2 months were significantly older (26.8 ± 5.1 vs 23.7 ± 4.7 years, respectively; p<0.01) than those who were not breastfeeding at 2 months. As expected, proxies for acculturation (Hispanic subgroup, and time in US) were significantly associated with breastfeeding practices. Breastfeeding at 2 months occurred among 43%, 89%, and 68% of women identifying their origins as being from Puerto Rico, Mexico, and other countries, respectively (p<0.001). Women who breastfed at 2 months had lived in the US for significantly less time than their counterparts who were not breastfeeding (Median: 7 vs 23 years, respectively; p<0.01). There were no statistically significant differences in maternal education, delivery mode, WIC participation, employment and breastfeeding status as an infant between women who were vs were not breastfeeding at 2 months postpartum. Thus, these variables were not entered as covariates in the Cox survival models evaluating breastfeeding continuation (linear acculturative type model and multidimensional acculturative type model).
When evaluating breastfeeding duration by LAS groups, unadjusted Cox survival analyses indicate that those who were More Hispanic were significantly less likely to stop breastfeeding, compared to those who were More American (Figure 1; Hazard Ratio (HR): 0.54, Confidence Interval (CI): 0.30, 0.97, p<0.05). However, when maternal age was included in the model, acculturative type was no longer significant, but maternal age was positively associated with breastfeeding duration (p<0.05). The results of these models were unchanged when study group (intervention/control) was added as a covariate.
Exclusive breastfeeding was a rare practice, with 75.2% of participants having introduced formula in the first 2 days of life. The only covariate significantly associated with exclusive breastfeeding duration was Hispanic origins (Puerto Rico, Mexico, Other; p<0.05), but this was not entered into the Cox exclusive breastfeeding models, due to the strong correlation with acculturative type. The Cox survival models showed no significant differences in exclusive breastfeeding practices between linear acculturative type (data not shown).
Using the multidimensional assessment of acculturation, participants were classified as having a high level of integration (ie. the integrated-high group) (23.7%, n=27), traditional Hispanic (36.8%, n=42), having a low level of integration (ie. the integrated-low group) (12.3%, n=14) and assimilated (27.2%, n=31). As expected, there were significant differences in maternal age, education and time in the US when comparing these 4 acculturative types (Table 1). The Integrated-high group was significantly younger than the traditional (p<0.001) and integrated-low (p<0.05) groups. The assimilated group had a higher level of maternal education than the integrated-low group (p<0.05). The traditional and integrated-low groups had spent significantly less time in the US, compared to the assimilated and integrated high groups (p<0.001).
There were significant differences in breastfeeding rates, based on the multidimensional categorizations of acculturation. The best breastfeeding outcomes were observed in the integrated-low group, which scored low on both the Hispanic and American Scales. The groups that were most likely to discontinue breastfeeding were those that scored highest on the American Orientation Scale (integrated-high and assimilated groups).
In the unadjusted Cox model, the traditional Hispanic and integrated-low groups were significantly less likely to stop breastfeeding than the integrated-high group (Hazard Ratio (HR): 0.44, CI: 0.24 – 0.80, p=0.008 and HR: 0.13, CI: 0.04 – 0.45, p=0.001, respectively). The integrated-low group was also significantly less likely to discontinue breastfeeding than the assimilated group (HR: 0.21, CI: 0.06 – 0.72, p=0.013), and marginally less likely to stop breastfeeding than the traditional Hispanics (HR: 0.30, CI: 0.09 – 1.00, p=0.051). When the Cox model was adjusted for maternal age, similar results were obtained (Figure 2). The integrated-low group was significantly less likely to stop breastfeeding than the Traditional Hispanic (HR: 0.29, CI: 0.09 – 0.97, p=0.044), Assimilated (HR: 0.23, CI: 0.08 – 0.79, p=0.020), and integrated-high (HR: 0.16, CI: 0.05 – 0.55, p= 0.004) groups. Traditional Hispanics were marginally less likely to discontinue breastfeeding, compared to the integrated-high group (HR: 0.55, CI: 0.29 – 1.03, p=0.063). Maternal age had a significant, positive association with breastfeeding duration in this model (p<0.05). Study group (intervention vs control) was not a significant predictor of breastfeeding continuation in these models, and did not change the results for the acculturation variables.
The unadjusted Cox survival model shows no significant differences in the duration of exclusive breastfeeding (since birth) between the 4 acculturative types (Figure 3). However, the Assimilated group was marginally less likely to introduce foods other than breastmilk, compared to the integrated high group (HR: 0.58, CI: 0.33 – 1.03, p=0.06). Bivariate analyses identified no relevant covariates for this model.
We have demonstrated that acculturation among a low-income, multinational sample of Latinas can be successfully assessed with a multidimensional scale, and that the resultant acculturative types are associated with significant differences in breastfeeding outcomes. To our knowledge, this is the first publication which evaluates breastfeeding practices of Latinas using a multidimensional acculturation scale, (ie. which simultaneously evaluates the degree to which they adopt American culture and the extent that they retain their Hispanic heritage).
Breastfeeding researchers have used a variety of approaches to assess acculturation among Latinas. Rassin and colleagues (Rassin et al., 1993, Rassin et al., 1994) used 14- and 20-item linear scales to assess acculturation. They demonstrated that acculturation was inversely associated with breastfeeding intentions (Rassin et al., 1993, Rassin et al., 1994), initiation (Rassin et al., 1993, Rassin et al., 1994), and breastfeeding continuation at 2–3 weeks postpartum (Rassin et al., 1993). Our linear acculturation results support their findings regarding breastfeeding continuation. Anderson (Anderson et al., 2004) used an additive score based on 37 acculturation variables among low-income Latinas in Hartford, CT. Contrary to Rassin’s results, Anderson demonstrated no association between breastfeeding initiation and acculturation. It is possible that Anderson’s additive score may have masked the relationship between key acculturation variables and breastfeeding initiation.
Our multidimensional analyses provide a more comprehensive picture of the relationship between acculturation and breastfeeding, than can be obtained from linear scales or proxy measures. In general, our linear acculturation scores are in agreement with previous research based on fewer acculturation items (Byrd et al., 2001, Gibson et al., 2005, Sussner et al., 2008) or proxy measures (Celi et al., 2005, Gibson-Davis and Brooks-Gunn, 2006, Gorman et al., 2007, Harley et al., 2007, Merewood et al., 2006, Newton et al., 2009, Pachon and Olson, 1999, Pérez-Escamilla et al., 1998, Romero-Gwynn and Carias, 1989, Singh et al., 2007), which showed that less acculturated women have better breastfeeding outcomes.
However, these seemingly straight-forward relationships become more complex when levels of both Hispanic and American orientation are assessed separately. For example, as shown in Figure 2, there are no significant differences in the breastfeeding continuation rates of the traditional Hispanic and assimilated groups. If we were to rely on a linear acculturation score, this similarity in breastfeeding practices would suggest a similar level of acculturation. However, these 2 groups are very different in their acculturative typology. The traditional Hispanic group scored high on the Hispanic scale and low on the American scale, while the assimilated group was the exact opposite (high on the American scale and low on the Hispanic scale). Similarly, linear acculturation scores would predict that the groups that scored lowest on the American scale (traditional Hispanics and integrated-low) would have similar breastfeeding outcomes. However, there is a nearly 2-fold difference in the 6 month breastfeeding rates of the traditional and integrated-low groups. We were unable to identify the reason why the integrated low group was significantly more likely to continue breastfeeding than the traditional Hispanic group. There were no significant differences between these groups with regards to maternal age, education level, time in the United States, Hispanic origins, marital status, delivery mode, employment status, WIC participation exclusive breastfeeding status on day 1 postpartum, or history of being breastfed as an infant (data not shown). Further qualitative research is necessary to better understand the striking difference in breastfeeding continuation rates between these acculturative types.
These findings highlight the limitations of simply relying on linear assessments of acculturation and emphasize the need to better understand the complex relationship between acculturation and breastfeeding among US Latinas. The application of a brief multidimensional assessment of acculturation in prenatal clinics will likely be useful to properly target those in need of breastfeeding support with different acculturative types. Although this brief assessment will take more effort than simply asking about acculturation proxies, this investment will likely be worthwhile, because it can clearly identify subgroups of Latinas who are at highest risk for early breastfeeding termination.
There are a few limitations inherent in this research. The sample size of the integrated-low group was small (n=14), and it would be useful to assess the breastfeeding patterns of a larger group of women who share this low level of orientation to both the Hispanic and American cultures. Approximately 16% of participants were right censored in the Cox survival analyses, either because they were still breastfeeding or because they were lost to follow-up before they stopped breastfeeding. However, due to the similarities between the groups with and without complete breastfeeding data, the potential for selection bias seems low. Finally, our results may not be generalizable to Latina populations composed of other Latina subgroups.
In conclusion, the modified ARSMA-II was a useful tool for the assessment of acculturation in our multi-ethnic Latina population, and revealed distinct patterns of breastfeeding continuation by acculturative type. Our results indicate that acculturation is not a simple linear process of progressing from one’s native culture to the adoption of the new culture. Rather, individuals vary in the levels to which they retain the culture of their homeland and adopt the practices of their new environment. A multidimensional assessment of acculturation may provide better breastfeeding support program targeting, as it will help to identify Latinas at increased risk for short breastfeeding duration.
The authors are grateful to Katherine Morel, MS, Angela Bermúdez-Millán, Ph.D., MPH, Grace Damio, MS, CDN, Sara Young, RN, IBCLC and Nan Kyer, IBCLC, for their assistance in coordinating and managing the randomized trial. We also appreciate the administrative support provided by Lisa Phillips and Khara Leon.
This project was supported by the Patrick and Catherine Weldon Donaghue Medical Research Foundation and by award P20MD001765 from the National Center on Minority Health and Health Disparities. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center on Minority Health and Health Disparities or the National Institutes of Health.
Donna J. Chapman, Associate Research Scientist, Yale School of Public Health, 135 College Street, Suite 200, New Haven, CT 06510, Email: email@example.com, Phone: 203-785-7353, Fax: 203 737-4591.
Rafael Pérez-Escamilla, Professor of Epidemiology & Public Health, Director, Office of Community Health, Yale School of Public Health, 135 College Street, Suite 200, New Haven CT 06510, Email: firstname.lastname@example.org, phone: 203 737-5882, fax: 203 737-4591.