In this cross-sectional analysis using baseline data from Proyecto Buena Salud, we found that almost one-third of Hispanic prenatal care patients reported prenatal depressive symptoms in early pregnancy. Higher levels of education and income, and living with a spouse/partner were independently associated with lower risk of depressive symptoms while there was the suggestion that failure to discontinue cigarette smoking with the onset of pregnancy and English language preference were associated with higher risk. While single marital status, second generation in the U.S., and higher levels of alcohol consumption were associated with higher risk of depressive symptoms in univariate analyses, these results were attenuated after adjustment for other risk factors. Finally, age, number of children and adults in the household, birthplace, and psychological acculturation score were not associated with depressive symptoms.
We observed a relatively high prevalence of prenatal depressive symptoms (30%) compared to prior studies among Hispanics as well as non-Hispanic white women (8
). Among studies using the EPDS to assess prenatal depression, Rich-Edwards et al. found that Hispanic participants in Project Viva (5% of the sample) had a higher prevalence of depressive symptoms (EPDS scores >12) in mid pregnancy compared with non-Hispanic white women (16% vs. 7%) (8
). In a prospective longitudinal study of women in England (n = 8,323), Evans et al. found that 11.8% of women scored >12 on the EPDS at 18 weeks gestation (26
). In their cohort of French women, Dayan et al. found that 14.5% of participants scored >14 on the EPDS in mid pregnancy (37
Prior studies focusing on Hispanic women were conducted predominantly among Mexican American women and used either the Center for Epidemiological Studies Depression Scale (CES-D) or the Beck Scale. Unlike the EPDS which is limited to cognitive and affective symptoms, the CES-D and Beck scales also query physical symptoms (e.g., fatigue and physical discomfort) which, because they are typical complaints of pregnancy, could lead to an overestimate of depression (8
). In these studies, the prevalence of depressive symptoms among Hispanic pregnant women ranged from 32.4% (7
) to 51% (CES-D > 16) (20
). In addition, in a small sample of Puerto Rican women, Zayas et al. found that 53% reported depressive symptoms in the third trimester (Beck scale > 14) (9
Aside from differences in population characteristics and screening tools, differences in findings may reflect demographic differences that are, in turn, related to depressive symptoms (8
). We found that increasing education, income, and living with a partner were associated with lower risk of depressive symptoms in multivariate analyses. Similarly, among a cohort of predominantly non-Hispanic white women, Rich-Edwards et al. found that women who were less educated, with fewer financial resources, and without a spouse/cohabitating partner had a higher prevalence of depressive symptoms in pregnancy. Similar findings have been observed among non-pregnant Hispanic women (21
). These findings have important public health implications as 48% of all births to Hispanics in the U.S. in 2005 were to women who were single (38
The few studies of acculturation and depressive symptoms among Hispanics have been limited to the postpartum period (18
), or combined prenatal and postpartum periods (20
), and observed inconsistent findings. Authors have found either no difference in prevalence of depressive symptoms according to birth place (8
) or acculturation (19
), or a higher risk of depressive symptoms among Hispanics who were born in the U.S. (21
), spent more years in the US (20
), or had higher acculturation scores (40
). Similarly, we found a suggestion that women who preferred English or who were second generation were more likely to report depressive symptoms compared to less acculturated women. In addition, participants reported a variety of characteristics which are potential risk factors for depression. For example, we observed higher levels of perceived stress and trait anxiety in this population compared to those reported by prior studies among predominantly non-Hispanic white populations (2
While we did not observe a statistically significant trend between age and depressive symptoms, we did observe a lower risk among women aged 16–24 as compared to women aged 25–29. However, the relative risks for the older age categories (30
) were also consistent with a lower risk, although not significantly so. A recent systematic review on risk factors for depressive symptoms during pregnancy observed inconsistent findings for age (14
). Our finding may be confounded by social support, such that younger Hispanic women may be more likely to live with their extended family and have greater social support, as compared to older Hispanic women (42
This study faces several limitations. The EPDS is a self-reported measure of depressive symptoms, as opposed to a clinical diagnosis. Although the EPDS has been validated among Hispanic pregnant women (27
), minority women and women of low socioeconomic status may be particularly susceptible to social desirability bias (43
). While a score >12 on the EPDS does not confirm depression, the EPDS is widely used to indicate probable depressive disorder and factors associated with such scores have been predictive of clinically significant depression (27
). In addition, minority women of lower socioeconomic status might be less likely to receive a professional diagnosis or treatment for depression. Therefore the fact that our definition of depressive symptoms did not rely on treatment or diagnosis, could be considered a study strength. Finally, brief screening instruments have practical clinical utility in that they identify a high risk group who may benefit from intervention.
Another limitation of the study is its cross sectional design. Because we assessed both depressive symptoms and potential risk factors at the time of enrollment, we cannot ensure the temporality of the relationship, particularly for the lifestyle variables of smoking and alcohol consumption. However, for the socioeconomic variables (i.e., income, education, insurance) and acculturation factors (i.e., generation in US, language) we can be more assured that these factors preceded depressive symptoms during pregnancy. Our study did not collect information on other potential risk factors for depressive symptoms (e.g., unplanned pregnancy, history of depression, and social support) although, unlike some prior studies, we collected perceived stress and anxiety.
Marital status, generation in the U.S., and alcohol consumption were associated with higher risk of depressive symptoms in univariate analyses but were attenuated after adjustment for other variables. However, because many adverse life circumstances co-exist, these factors may be mediators or on the causal pathway for onset of depressive symptoms. For example, older generation in the U.S. may reflect higher acculturation which, in turn, was associated with higher alcohol consumption and smoking which, in turn, was associated with depressive symptoms in the final model. However, our goals were to identify correlates of prenatal depressive symptoms which could then be used to target women at risk who might benefit from intervention. In light of these goals, even results from the unadjusted analyses are important for targeting at-risk individuals and developing public health strategies.
Approximately 44% of women did not know their annual household income. This may have been due, in large part, to the young age of participants (70% less than age 24) and the fact that many lived at home with older family members, in unconventional living situations, and received public assistance. Women missing information on income did not differ significantly from women not missing this information in terms of their risk of depression. However, including these women in the multivariable regression models, while having the advantage of retaining a consistent sample size between unadjusted and adjusted analyses, had the potential limitation of including a category with a heterogeneous grouping of income. Finally, a total of 87 women were excluded from the final model due to missing information on one or more of the predictors of interest; however these women did not differ from those retained in the model with respect to risk of depressive symptoms (p=0.61).
Because participants were recruited at prenatal care visits (up to 20 weeks gestation), we excluded, by definition, high-risk women who did not attend prenatal care. However, our study population included a sizeable proportion of women who were at high risk of depression based on socioeconomic factors and ethnicity. In addition, 2006 vital statistics data for births in Springfield, Massachusetts indicate that 93.3% of Hispanics begin prenatal care by the second trimester (44
). Because our population was limited to women of Puerto Rican or Dominican heritage, our findings cannot be generalized to other Hispanic subgroups. Finally, 4% of women were excluded for preexisting disease (e.g., type 2 diabetes). Risk factors for depression may differ among these women, and therefore findings cannot be generalized to this group.
The study also has several strengths. It addresses a gap in prior studies of depressive symptoms by focusing on Hispanics of Puerto Rican/Dominican origin during the perinatal period. As compared to prior studies, its sample size is relatively large with adequate numbers of both foreign-born and U.S.-born women. A further strength is its use of a standardized instrument, the EPDS, which has been the focus of extensive psychometric research and is not confounded by the physical symptoms of pregnancy. The study also documents the prevalence of depressive symptoms among a low-income, minority sample in Western Massachusetts.
In summary, Proyecto Buena Salud represents a group of Hispanic women with higher levels of depressive symptoms as compared to predominantly non-Hispanic white cohorts. Our findings support the potential role of routine screening for depression and for integrating psychosocial services into primary care practice. The impact of such intervention is likely to be greatest for Hispanics, given the high rates of fertility and births to single women among this population. Finally, because women are motivated to seek health care services during the prenatal period, embedding preventive interventions within obstetrical care settings can increase access to psychosocial services and improve the lives of both mothers and children.