Birthing, Nativity, and Maternal Depression: Australia and the United States. Presented to the INSIDE Conference Barcelona, Spain June 27, 2011

Transcription

1 Birthing, Nativity, and Maternal Depression: Australia and the United States Presented to the INSIDE Conference Barcelona, Spain June 27, 2011 Melissa L. Martinson*, Princeton University Marta Tienda, Princeton University *Please direct correspondence to Melissa L. Martinson, Office of Population Research, Wallace Hall, Princeton University, Princeton, NJ 08544,

2 Abstract This study analyzes the ECLS-B and LSAC birth cohort surveys to examine variation in maternal depression by nativity and acculturation status in Australia and the United States, two countries with long immigrant traditions, many cultural similarities, and a common language. US immigrant mothers are significantly less depressed than citizen mothers, but maternal depression does not differ by nativity in Australia. Moreover, the association between duration of residence and maternal depression is not linear recent arrivals and long-term residents exhibit the highest depression levels. Lack of English proficiency exacerbates maternal depression in Australia, but protects against depression in the United States. Differences in immigration regimes and contexts of reception likely contribute to the differing salience of nativity for maternal depression. 1

3 Birthing, Nativity, and Maternal Depression: Australia and the United States I. Introduction Especially for first-time mothers, giving birth is generally a joyful event, but one that introduces new stresses to family life. In fact, approximately 13 percent of all US mothers experience a postpartum depressive episode (O Hara & Swain, 1996). Postpartum depression is the second most common reason for hospitalization among women ages 18 to 44 (Jiang et al., 2002). Mothers with severe postpartum depression think about harming themselves or abandoning their infants (Jennings et al., 1999). Notwithstanding extensive media coverage of cases where depressed mothers harm their children, most episodes of maternal depression result in neglect rather than outright physical injury. Nevertheless, infant neglect, whether intentional or not, can have deleterious, long-term consequences for child wellbeing. Maternal depression is associated with developmental delays, behavior problems, social skills deficits, poor emotional regulation, adverse psychological symptoms and psychiatric disorders (Lyons-Ruth et al., 2000; Goodman & Gotlib, 1999; Oyserman et al., 2002). These psychosocial impairments can become evident as early as the neonatal period and may affect child development through adolescence. Early detection and treatment of maternal depression can help avert the deleterious consequences of maternal depression for child wellbeing (Miller, 2002). Despite the prevalence of this serious health issue, very few studies examine maternal depression in immigrant mothers and none consider whether foreign-born mothers acculturation status is associated with depressive episodes following a birth. In a recent review of postpartum depression among US immigrant women, Fung & Dennis (2010) draw attention to the dearth of evidence on maternal depression among the swelling foreign-born population. Moreover, the 2

4 limited research that considers nativity differentials in postpartum depression yields inconsistent results, partly due to differences in outcomes measured, target samples and analytical methods. Although many studies show that geographic movement, and international migration in particular, is a stressful process that involves myriad cultural and socioeconomic adaptations over a protracted period of time (Berry, 2006), to our knowledge, no study considers whether maternal depression varies according to duration of residence in the host society. It is conceivable that the inconclusive results about nativity and maternal depression conceal systematic variation by level of acculturation as well as the context of reception. Accordingly, we address three questions that aim to reconcile contradictory findings in the sparse literature about nativity and maternal depression using nationally representative data for two of the largest immigrant receiving nations Australia and the United States. First, does the incidence of maternal depression differ by nativity status, and if so, are the differentials similar in both countries? Second, does the level of depression depend on acculturation status, namely length of residence in the host country and English proficiency? Finally, what factors mitigate and accentuate the likelihood of a depressive episode within the first year after giving birth, and are the mediators of depression similar in both countries? As two nations with long immigration traditions, many cultural similarities, and a common language, Australia and the United States afford an instructive comparison of maternal depression because their admission regimes and their treatment of settlers from abroad represent quite different contexts of reception for immigrants. Relative to their population size, both nations receive similar annual inflows on the order of 0.4 and.05 percent (legal and illegal); however, in Australia, the foreign-born share of the total population is approximately double that of the United States, at 25 versus 13 percent, respectively (Freeman and Birrell, 2001; Walsh, 3

5 2008). Both countries admit immigrants based on family, labor market and humanitarian criteria; however, Australia currently places much greater weight on skills that are relevant to its labor market including strict thresholds for English proficiency while the United States admits the majority of its immigrants under family preferences without regard to age, economic potential, or proficiency in English (Wasem, 2007; Freeman and Birrell, 2001; Miller, 1999). 1 Although both nations have checkered histories with respect to acceptance of immigrants (Freeman and Birrell, 2001), Australia is more forthcoming in recognizing its need for immigrants and in acknowledging their contributions to the national economy (Walsh, 2008; Australian Government Department of Immigration and Citizenship, 2010). Furthermore, based on the generosity of benefits afforded to foreign settlers, Australia appears to be more immigrant-friendly in ways that can influence the stress associated with integration. For example, following the 1996 revisions to US welfare programs, new immigrants are ineligible for social benefits, including Medicaid, Supplemental Nutrition Assistance Program (SNAP or Food Stamps), and Temporary Assistance to Needy Families (TANF), until after they have been in the country for five years. This policy context could exacerbate stress for recent US immigrants who give birth during the period that the benefit moratorium is in place. Australian permanent settlers are not restricted from access to health and social programs once admitted; however, immigrants must have resided as permanent residents for at least two years before they can access unemployment benefits, sick leave and stipends for postsecondary education (Australian Human Rights Commission, 2008). 2 1 In the United States, English proficiency is required for naturalization rather than admission; nevertheless, the proficiency level required in Australia is appreciably higher compared with the United States. 2 In addition, both countries offer a variety of community-based services that cater to the needs of new immigrants. Australian asylum seekers are denied access to such services and there are only a very small number of specific asylum seeker services catering to their needs. 4

6 Whether the salience of nativity for maternal depression actually differs between these two contexts of reception is an empirical question that we address below. Section 2 reviews existing research about the correlates of maternal depression, including the handful of studies that consider nativity status, and derives testable propositions. We describe the data and operational measures in Section 3, with due attention to issues of item comparability across surveys. The empirical analysis reported in Section 4, show that immigrant mothers are significantly less depressed than native-born mothers in the United States, but nativity differences in maternal depression are nonexistent in Australia. Lack of English proficiency exacerbates maternal depression in Australia; that it protects against maternal depression in the United States implicates ethnic networks in providing social support to immigrant mothers. Moreover, the association between duration of residence and maternal depression is not linear recent arrivals and long-term residents exhibit the highest depression levels. The final section concludes, highlighting both the insights that comparative analyses yield and directions for future research. 2. Maternal Depression: Claims and Evidence Although most cases of postpartum depression are short-lived, lasting 6 months or less (Oates, 1995), stresses associated with childbirth often lead to extended or recurrent maternal depression (Nott, 1987; Warner et al., 1996). Prior studies have identified several social and environmental factors that can exacerbate the likelihood of maternal depression. Chief among these are a history of mental illness, stressful life events, and lack of social support (Robertson et al., 2004). Mothers with low levels of education are at higher risk of a depressive episode following childbirth (McCue Horwitz et al, 2007); however, employment typically reduces the 5

7 likelihood of a depressive episode, particularly when mothers work by choice rather than necessity (Hock & DeMeis, 1990; Usdansky et al., 2011). Maternal health status during pregnancy and the health status of the infant, notably low birthweight, also influence the likelihood of postpartum depression (Halbreich, 2005). Compared with citizen mothers, US immigrants have better birth outcomes, but this advantage erodes over time (Weeks & Rumbaut, 1991). Compared with native-born mothers, immigrants may experience higher levels of stress because of thinner social support networks. Because of the unique integration challenges associated with adjustment to a new society, stress associated with demands of acculturation may be particularly acute among recent immigrants (Berry, 2006). Adjustment stress should be especially pronounced for immigrants who are not proficient in English and thus encounter barriers to social institutions and employment (Berry, 2006), as well as difficulty navigating the health care system or communicating with health professionals (Escarce and Kapur, 2006). Partly due to positive selection, immigrant mothers may exhibit better mental health status than their citizen counterparts, as many studies document for Mexican immigrants to the United States (Grant et al., 2004; Alegria et al., 2008). The evidence on maternal depression among immigrant mothers is limited both because very few studies consider the salience of nativity for mental health following address this question and, owing to diverse methodologies and samples, the empirical findings are mixed. For example, Stewart (2008) and Bandyopadhyay et al. (2010) find a positive association between immigrant status and maternal depression, but other researchers (Heilemann et al.,.2004; Huang et al., 2007; Davila et al., 2008) claim that foreign-born mothers are less likely to experience a depressive episode following a birth. In Canada, Stewart (2008) finds that immigrant mothers 6

8 have about 4 times the risk of postpartum depression compared to native-born mothers using the Edinburgh Postnatal Depression Scale (EPDS). This study recruited 341 participants from hospitals in Vancouver, Montreal, and Toronto and interviewed them 7 to 10 days after birth. In addition to its limited external validity, the study included a large proportion of refugees among the immigrant sample, which likely biased the estimates of maternal depression in this study. Bandyopadhyay and associates (2010), who studied nativity differentials in maternal depression for a sample of mothers in Victoria, Australia, also show that immigrant mothers face a higher risk of depression six months after childbirth compared with their native-counterparts, particularly if they were not English proficient. This claim, which basically compared EPDS scores for native- and foreign-born mothers, did not hold up to further scrutiny. Based on a multivariate analysis of maternal mental health measured by the Short Form-36 (SF-36), the authors find no relationship between nativity status and mothers depression. In addition to several methodological limitations, the external validity of this study is questionable both because of the low response rate for immigrant women (31 percent) and because the larger intervention study produced null findings for 16 Victorian localities. Compared with Australia, there are more studies about nativity variation in maternal depression for the United States, but the empirical evidence is likewise inconsistent., Davila et al. (2008) analyze a sample of pregnant and postpartum women of Mexican descent in San Antonio using the Center for Epidemiological Studies-Depression scale (CES-D). They find significantly higher levels of depression among US-born women compared with their Mexicanborn counterparts, with foreign-born women about half as likely to be depressed. That the authors combined pregnant women with mothers who were up to one-year post-partum makes it difficult to disentangle perinatal and postnatal depression. Moreover, it is unclear whether the 7

9 experiences of Mexican immigrant women can be generalized to mothers of other national origins. San Antonio s Hispanic population is dominated by Mexicans both native and foreign born, which might reflect thicker social support networks that are not easily reproduced among new immigrant groups. Using nationally representative data for the United States, Huang and colleagues (2007) find that native-born mothers experience higher rates of moderate and severe depression on the CES-D scale than foreign-born women one year following the birth of a child; however, they note that the propensity for depressive episodes differs by race and ethnicity. Specifically, they find no differences in depression rates between non-hispanic white native- and foreign-born women, but lower rates for the Hispanic foreign-born versus with native-born mothers, and higher rates among Asian immigrants versus their US-born counterparts. These authors did not conduct multivariate analyses of depression rates, nor did they consider whether the incidence of depression varied by length of US residence or English proficiency level; however, race and ethnic variation in the incidence of maternal depression among immigrant and citizen mothers suggests that other social and acculturation factors may influence the likelihood of maternal depression. It is striking that none of these studies examine the influence of the length of duration in the host country on maternal depression. Because immigration and its associated adjustment is a stressful process, particularly for groups that are culturally dissimilar to the host society, it is conceivable that stress associated with childbirth differs by length of residence. Depending on the auspices of migration, i.e., whether newcomers are sponsored by employers, family members, or are refugees, and the generosity of welfare benefits in the host society, levels of social support will likely differ by years since migration. Recent immigrants, particularly those 8

10 with low incomes, low proficiency in English, and few relatives in their new destination may have be at higher risk of maternal depression following the birth of a child. Several studies based on the United States show that the length of residence in the host country is an important correlate health status, but the vast majority focus on physical health. On the one hand, there is ample empirical evidence that recent immigrants have better health on a variety of outcomes, including overweight/obesity among adults, low birthweight among infants, and self-rated health (Jasso et al., 2004; Antecol & Bedard, 2006; Acevedo-Garcia & Bates, 2007). A handful of studies also show that Latino immigrants have better mental health status than the native-born population (Grant et al., 2004; Alegria et al., 2008). On the other hand, there is mounting evidence the immigrant health advantages both physical and mental deteriorate over time (Finch, Kolody, and Vega, 2000; Goel et al., 2004; Antecol & Bedard, 2006; Escarce, Morales and Rumbaut, 2006; Acevedo-Garcia et al., 2010). Because of data constraints, most studies of US immigrants focus on the Mexican origin population, to the exclusion of other Latin American origins as well as Asian immigrants (see Escarce, Morales and Rumbaut, 2006). There are reasons why the experiences of Mexicans are not representative of other groups, and especially those originating in Asia. For one, there are large educational differences between the foreign-born population from Mexico and other Latin American countries compared with immigrants from Asia, Europe and other regions. To the extent that educational attainment is associated with depression, the external validity of studies based on Mexicans is weak (Escarce, Morales and Rumbaut, 2006). That a large share of Mexican immigrants lacks legal status further compounds the stresses associated with social integration not only because of fear of deportation, but also because of the more limited access to social welfare benefits enjoyed by legal residents, and particularly those with over five years of continuous legal residence. Finally, 9

11 the Mexican origin population has a long immigration history that has produced a large nativeborn segment, which not only sponsors family members, but also provides an important source of social support for recent arrivals. Living Arrangements and Social Support In terms of social support, there is evidence that Mexican immigrant women are quick to reconstitute friend social networks in the United States, but social connectedness with family members tends to increase with time in the host country (Vega et al., 1991), thus recent migrants likely do not have the social support available to them in the United States that they had in their home countries. Vega and colleagues (1991) also report that contact with siblings and mothers provided the highest levels of emotional support to immigrant women. This is consistent with evidence that new immigrants from Mexico are more likely to live in extended family situations that include additional adults for access to instrumental social support, rather than simply due to traditional cultural reasons (Van Hook & Glick, 2007). The presence of other relatives or source country acquaintances in the host country can facilitate the adjustment of new arrivals by assisting with housing and employment as well as providing information about resources and social supports (Massey, et al, 1987; Tienda, 1980). In fact, immigrants are considerably more likely than natives to live in extended households, but the prevalence of complex households declines over time, as the foreign-born become socially anchored in their new communities (Landale, Opresa, & Bradatan, 2006). For example, Tienda and Glass (1995) find that co-residence of another female adult facilitates the labor market activity of single mothers; they argue that the family system, via inclusion of extended relatives, alleviates some of the harsher aspects economic hardship by assisting with childcare and domestic responsibilities. Extended living arrangements might also exacerbate the stresses of 10

12 daily living, especially if associated with crowding and low incomes. Under some circumstances, the presence of another family member might alleviate the stresses of adjusting to a newborn, particularly for first time mothers. Angel and Tienda s (1982) study of variation in extended living arrangements among US families reveals not only higher propensity for minority families to include other relatives, but also to compensate for the lower earnings of minority households. That similar behavior does not obtain for majority white households invokes the importance of both culture and economic need as mechanisms driving the propensity for families both to incorporate other relatives. Although they did not consider variation by nativity, a subsequent paper by Tienda and Angel (1982) finds that single mothers are significantly more likely to invite other relatives to co-reside, and particularly if they have low income. These authors do not find nativity differences in household extension behavior, nor did they consider the timing of extension relative to childbearing. Maternal Employment There is evidence that maternal employment and employment preferences are associated with depression levels. Compared with mothers who both prefer employment and are currently working, mothers who prefer to work, but stay home after the birth of the child, are more likely to experience depressive symptoms (Hock & DeMeis, 1990; Usdansky et al., 2011). Usdansky and colleagues also find that mothers in jobs deemed as high quality ( good jobs ) are less likely to be depressed, suggesting that perhaps mothers with less economic resources do not benefit as much from work. Klein and associates (1998) find that job overload is associated with maternal depression and distress. The salience of the interconnectedness of maternal employment and mental health makes employment status an important predictor to consider in the analysis of maternal depression. 11

13 In sum, the existing research provides mixed evidence about whether and how nativity status is associated with maternal depression. On the one hand, foreign-born mothers, and especially recent arrivals, may be less proficient in the host country language, have less knowledge about resources available to new mothers, and have fewer economic resources than citizens. On the other hand, immigrants are more likely than citizens to rely on informal resources for social support, and at least in the United States, have better birth outcomes than native-born women. Whether similar differentials exist in Australia, where immigrants are positively selected based on market skills and proficiency in English is an empirical question that we address below. We improve on existing studies in several ways. First, we use nationally representative surveys that address the weak external validity of most existing studies. Second, we disaggregate nativity status by duration to test whether the heterogeneity of the foreign-born population by length of residence is responsible for the inconsistent association between maternal depression and nativity observed in other studies. Third, because English proficiency is strongly related with integration prospects, and is a crucial skill to navigate medical and social care systems (Escarce and Kapur, 2006), we evaluate the salience of English proficiency in predicting the maternal depression among new mothers. Finally, we provide the first comparative evidence about the association between maternal depression and immigration both to illustrate how mother s mental health status presents in Australia and the United States, two industrialized nations with long immigration traditions, and to call attention to the importance of this relatively understudied problem with potentially serious consequences for children. 12

14 3. Data and Methods We use two nationally representative surveys to compare and contrast the association between immigration status and maternal depression. For Australia we use the Longitudinal Study of Australian Children (LSAC), and for the United States, the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B) survey. The ECLS-B is a nationally representative sample of approximately 10,000 children born in 2001 to mothers who are 15 years of age or older and who are alive and residing in the United States at 9 months of age (ECLS-B, 2010). The ECLS-B survey is conducted by the National Center for Education Statistics and focuses on early child wellbeing with rich measures on health, home life, and education. The primary parent was interviewed, and all analyses are limited to interviews with the focal child s birth mother. The US analyses are based on a subsample of 7,700 mothers 3. The response rate was 74 percent. The LSAC, which is a nationally representative study of Australian infants born in , is based on a sample of approximately, 5,000 births taken from the Health Insurance Commission Medicare database (LSAC, 2009). Children from very remote postcodes were excluded from the study. The approach to data collection is quite similar to the ECLS-B. For the Australian analyses, the sample is limited to cases where the birth mother is the primary parent interviewed, and all analyses are based on a subsample of 3,944 mothers. The response rate was 64 percent. Our analyses use the first wave of the both the ECLS-B and the LSAC. The infants are age approximately 10 months in the ECLS-B and 9 months in the LSAC at the first interview. All analyses are weighted using the svy procedures in Stata SE 11.0 to account for the complex sampling designs and response rates in both surveys. 3 Please note that all sample sizes for the US have been rounded to the nearest 50 as per National Center of Education Statistics contract regulations. 13

15 Operational Measures Maternal depression. For the United States, a modified version of the Center for Epidemiologic Studies Depression Scale (CES-D) is used in the ECLS-B to measure maternal depression. The CES-D has been validated for use among various demographic groups, including those with immigrant and low literacy backgrounds (Radloff, 1977; Finch et al., 2000; Grzywacz et al., 2006). There are 12 questions included in this scale. The ECLS-B asks these questions in the parent self-administered questionnaire. Each scale item asks about the way the respondent felt or behaved during the past week with 4 response options: rarely or never (less than 1 day), some or a little (1-2 days), occasionally or moderate (3-4 days), and most or all (5-7 days). The CES-D is scored from 0-36 according to the ECLS-B instructions, and we use this continuous CES-D score. The Kessler 6 (K6) scale was used to assess maternal depression for the Australian sample in the LSAC. The K6 is a validated 6 question scale asking how the respondent felt over the past 4 weeks with the following response options: none of the time, a little of the time, some of the time, most of the time, or all of the time (Furukawa et al., 2003; Kessler et al., 2003). The K6 is scored from 0-24 according to the LSAC instructions, and like the CES-D, we use the continuous K6 score. Both the modified CES-D and K6 have demonstrated validity as screening measures for depression, rather than providing a clinical diagnosis of depression, and due to their brevity are ideal measures for inclusion in large surveys such as the ECLS-B and LSAC. Nativity status and length of residence. In both the ECLS-B and LSAC we construct a dichotomous variable for nativity status (foreign or native-born). Using mother s current age and age at arrival in the United States, we construct a continuous measure of duration for immigrant mothers in the ECLS-B. In the LSAC, we use interview year and arrival year in Australia to 14

16 construct the continuous duration measure. Based on this continuous measure, four categories are created for the final duration in the host country variable: 0-4 years, 5-9 years, years, and 15 years or more. These categories are based on the distribution of the data as well as the fact that immigrants in the United States are prohibited from accessing many social programs and benefits for the first 5 years in the country. Other correlates. The theoretical discussion identified several factors that not only differ between immigrant and native mothers, but also influence the risk of depression following a birth. Therefore, our multivariate models include several additional covariates, including mother s age (at the time of childbirth), education, poverty status, living arrangements and several attributes of the index child, including age (in months), parity, and birth weight. Low birth weight (<2500 grams) is included as a proxy measure for infant health, which we expect to increase maternal stress, other things equal. To measure household structure, we use mothers current relationship status (married, cohabiting, or single) and a constructed indicator of family type (nuclear or extended). Extended households include one or more adult relatives other than the mother s partner. Maternal education is measured categorically in order to maximize comparability between the school systems of both countries. We use four categories to represent variation in mothers educational attainment: less than high school, high school diploma, some college, and bachelor s degree or higher. These categories are included in the US data, but the education categories included in the LSAC require recoding to render them equivalent. LSAC data include a measure for years of primary and secondary school completed as well as a qualifications indicator, which includes an option for a vocational diploma or a certificate. Following Choi and associates (2011), we coded the combinations in the following hierarchical manner: 11 years or 15

17 less of schooling is equal to less than a high school diploma (whether or not a certificate was completed), 12 years of school is equal to a high school diploma (without a certificate or diploma), 11 years or more with a diploma and 12 years or more with a certificate are equal to some college, and the final category of bachelor s degree or higher is equivalent to the US category. Additional maternal and household attributes known to influence maternal depression include mother s employment and household poverty status. Maternal employment combines work status before and after the birth of the index child: work both before birth and at time of survey; work before birth only; work at the time of survey only; and never worked before birth or at the time of the interview. Families in the lowest quartile of each country s weighted income distribution are designated as poor. Both surveys measure household income in bands. The advantage of designating the lowest quartile as poor is two-fold: first, the lowest quartile is the most comparable cut-point for both surveys; second, this threshold aligns closely with the widely used measure of income poverty 50 percent of median income that is common in international comparisons. Table 1 summarizes sample characteristics for mothers interviewed for the LSAC and ECLS-B surveys. Australian mothers are nearly are three years older, on average, than US mothers, and the foreign-born are about 1.5 years older than their native counterparts. In the United States there are no age differences by nativity. Table 1 About Here Nativity differentials in the educational attainment of new mothers reflect the different skill emphasis of the two countries admission regimes. In Australia over one-third of recent mothers lack a high school diploma, compared with less than 20 percent in the United States, but 16

18 nativity differences in maternal education differ between countries. Australian mothers are less likely than immigrants to have completed high school, but in the United States foreign-born mothers are less likely than citizen moms to graduate from high school. The skill emphasis of Australia s immigrant admission system is further evident both in the shares of foreign-born mothers with college credentials nearly 35 percent in Australia compared with 22 percent of foreign-born mothers residing in the United States. Furthermore, owing to Australia s English requirement as a condition of admission, only about 20 percent of recent foreign-born mothers indicated they were not fluent in English, compared with over half of US mothers. US mothers also exhibit higher attachment to the labor market than their Australian counterparts. Only 20 percent of American mothers reported that they had never worked compared with 31 percent of Australian mothers. At the other extreme, nearly half of US mothers worked both before and after a birth, but only one-third of Australian mothers did so. In both nations labor force participation is higher among citizen mothers compared with immigrants. Table 1 presents several indicators representing the context of maternal childbearing that are potentially associated with maternal depression, including child birth weight, the number of prior children, and living arrangements. The prevalence of low birth weight is higher in the United States compared with Australia, but the nativity differentials are reversed. The US differentials are consistent with a large body of evidence showing better birth outcomes for immigrant mothers compared with natives (Landale, Oropesa, & Gorman 1999), but in Australia the healthy birth advantage corresponds to citizen mothers. Although there is little between country variation in the proportions of first-time mothers, a slightly higher share of US mothers had three or more additional children 9 versus 7 percent. 17

19 Nearly 11 percent of Australian immigrant mothers had three or more prior children compared with only 6 percent of citizen mothers, but the comparable nativity differential for the United States is negligible. Prior childbearing is reflected in both country and nativity differentials in household size in both nations, immigrant mother households are slightly larger than those of native mothers. Finally, there are notable nativity and country differences in living arrangements of new mothers. Irrespective of mothers marital status, extended living arrangements are more prevalent in the United States compared with Australia. Estimation Methods Weighted means of maternal depression are calculated for both the US (CES-D) and Australia (K6). Multivariate regression models are estimated using the continuous maternal depression measure as the outcome. All analyses are conducted using the svy commands in Stata SE 11.0 in order to adjust for the complex sampling design in both the ECLS-B and LSAC. 4. Empirical Results Table 2 presents mean depression scores for the mothers of both Australian and American birth cohorts. Because the surveys use different metrics to assess maternal depression, absolute differences in the magnitude of between-country differences in depression scores are not meaningful. Therefore, in highlighting patterns of differentials, we focus on variation relative to the national averages 3.6 and 4.9 for Australia and the United States, respectively and among the covariates theorized to attenuate and accentuate maternal depression. Table 2 About Here Despite the different metrics used across surveys, there are several noteworthy similarities between countries, such as the significantly higher incidence of depression among 18

20 teen mothers compared with mothers in their prime reproductive years (25-34) and among poor compared with nonpoor mothers. These differentials in maternal depression are consistent with an extensive literature about the stresses associated with teenage childbearing (Deal & Holt, 1998; Birkeland, Thompson, and Phares, 2005). An additional similarity between countries concerns the higher prevalence of maternal depression among mothers of low birth weight infants compared with women who give birth to normal birth weight infants (Halbreich, 2005). Of particular substantive interest is the nativity variation in maternal depression between countries. The descriptive tabulations suggest that Australia s foreign-born mothers are not more depressed than their citizen counterparts, with the exception of mothers who have resided in the country between 10 and 14 years. This is not the case in the United States: with the exception of recent arrivals, foreign-born mothers are significantly less likely to be depressed than their citizen counterparts. That recent arrivals are more depressed than mothers who resided in the United States for five or more years possibly is due to the additional stresses associated with their integration into a new society, lack of access to social benefits and programs, and the challenges of navigating an unfamiliar health care system. In fact, lack of English proficiency is associated with significantly higher maternal depression relative to proficiency or native fluency in both nations. Two additional dimensions of between-country variation in maternal depression are noteworthy. One is the apparent lack of an association between mother s educational attainment and maternal depression in Australia. This contrasts sharply with the clear monotonic association evident for the United States, where college-educated mothers are least likely, and mothers without high school diplomas most likely to be depressed, following a birth. Second is the different association between household structure and maternal depression between 19

21 countries. In Australia, household extension appears to be unrelated to maternal depression except among units headed by married couples, where mothers are more rather than less likely to be depressed. For the United States, married mothers who live with extended relatives also average significantly higher depression than married mothers who live in nuclear families. It is unclear, however, whether the other relatives join these units because the mothers are depressed, or whether the presence of other relatives represents an additional stressor for new mothers. Among single US mothers, the presence of extended relatives is associated with lower rates of maternal depression compared with nuclear family living arrangements. A similar pattern of association between extension and maternal depression among single mothers also obtains in Australia, except that the gaps are not statistically significant. Multivariate Results Table 3 presents the multivariate regression results for maternal depression separately for Australia and the United States. The first specification considers the baseline association between nativity and maternal depression, with a control for age at birth and its square. Model 2 adds all child, household and maternal characteristics hypothesized to influence postpartum depression except for English proficiency, which is added in model 3. Compared to native-born mothers, immigrants in Australia appear to have a similar (if not slightly higher) risk of maternal depression, but foreign-born mothers face significantly lower risk of depression than US citizens. Moreover, these differentials hold across all three specifications. In the United States the point estimates for foreign-born mothers support the healthy immigrant effect that is pervasive for physical health outcomes and is robust to controls for a wide variety of indicators focused on maternal, child, and household wellbeing. Table 3 About Here 20

22 Model 3 reveals the salience of English proficiency on maternal depression in both countries. That lack of English proficiency significantly increases the risk of maternal depression in Australia is consistent with the importance of language as an admission requirement. By contrast, lack of English proficiency is actually protective against maternal depression in the United States! This result seems counterintuitive at first blush, but is consistent with the dominance of family sponsorship as a driver of admissions. Including English proficiency in the model reduces the magnitude of the coefficient on foreign-born status, but not reduce its statistical significance. Speaking English is an important marker of acculturation, and these results suggest that acculturation protects mothers against depression in Australia, but exacerbates depression risk in the United States consistent with the extensive literature on the negative health consequences of the immigrant acculturation process. There are other noteworthy differences in maternal depression risk in Australia and the United States. Most striking are the different education gradients in maternal depression between countries, which are presaged by descriptive results presented in Table 2. In the United States, higher levels of education protect mothers against depression; mothers with a bachelor s degree or more are significantly less likely than high school dropouts to suffer from maternal depression. This is an unsurprising result supported by a large and robust body of evidence on education gradients in health in this country. Yet the Australian education gradient is very clearly reversed, with college-educated women facing a significantly higher risk of maternal depression. The reasons for this result are not obvious, but may reflect the a-typicality of highly educated women in a context where significant pluralities achieve high school or less. As hypothesized, maternal employment status is also associated with the mental health of recent mothers. In Australia, mothers who never worked are significantly more depressed than 21

23 mothers who worked consistently before and after the birth. In the United States, this group of mothers also scores significantly higher on the maternal depression scale than mothers who consistently worked. Yet, mothers who worked before the birth but are not currently working also have a much higher risk of depression compared with these always working mothers. Conceivably, the higher levels of depression among mothers who worked before but not since the birth might signal their loss of labor market rewards as they strive to find fulfillment in childcare, more simply, formerly employed mothers prefer employment over staying home with their child (Usdansky et al., 2011). Like the descriptive results, household structure is associated with the risks of maternal depression, although the causal direction is not obvious. In Australia, all cohabitating mothers and single mothers who do not live with extended relatives face significantly higher risk of depression compared with married mothers in nuclear families. In the United States, married mothers residing in nuclear families face significantly lower risks of post partum depression compared with new mothers in all household types; however, mothers in single nuclear families face the highest risk of depression. The high depression risk among single household lone mothers suggests that the lack of social support for these mothers heightens their vulnerability for maternal depression. Table 4 presents multivariate regression estimates of duration on depression for both countries, following the same additive model sequence as Table 3. Especially noteworthy are the country differences in the association between years since arrival and maternal depression. In the baseline model, only foreign-born mothers who have resided in Australia between 10 to 14 years are more depressed than natives. However, this mental health disadvantage disappears once other correlates of maternal depression are modeled statistically. 22

24 Table 4 About Here The story is more complicated in the United States, where immigrant mothers who have been in the country for 5 or more years significantly less depressed than native-born mothers. In all three specifications, immigrant mothers who have been in the United States for 5 to 9 years are the least likely to be depressed compared to citizen mothers, followed by those with a duration in the country of 10 to 14 years and then 15 years or more. Maternal depression among recent immigrants is similar to that of native mothers. These differentials suggest that the most recent immigrants face greater acculturative stress than mothers who have had more time to learn the ropes in their new country, but their higher depression may also reflect stresses associated with limited access to social and medical benefits during the five year moratorium, which does not apply to longer-term immigrants. Yet, for immigrant mothers who have lived in the United States beyond the five-year cut point, there is evidence that the protective aspects of immigrant status fade with time. These results are consistent with the growing literature on the erosion of immigrant health with longer exposure to US dietary and behavioral practices. English proficiency operates in the same way as it did in Table 3 for both Australia and the United States. Because first time mothers may be less well prepared for the challenges associated with childbirth, and mothers with multiple births may have had several in their origin country, we reestimate the analyses for first births which represent about 40 percent of the birth cohort in each country. Table 5, which reports these estimates, reveals trivial nativity differences in the likelihood of depression in Australia, and reaffirms how the lack of English proficiency is associated with an elevated risk of maternal depression. Yet, for the United States, the results are quite different. Compared with the results based on all births, the clear relationship between duration and depressive symptoms is weaker; however, mothers who arrived in the United States 23

25 5 to 9 years prior to birth are still the least likely to be depressed. The attenuation of the duration effect for the United States suggests that for immigrant mothers, the first birth in the host country is particularly stressful. Therefore, by excluding mothers who had a previous birth (either in the United States or abroad), there is less differentiation by duration. That recent immigrants who bore their first child in the United States are less depressed than natives reaffirms how nativity operates as a protective factor even for new arrivals. However, for first time US mothers, lack of English proficiency does not protect against depressive symptoms a possible reflection of thinner social support networks for new arrivals. 5. Conclusions Given the different admission regimes and welfare benefits available to immigrants in the two Anglophone nations compared, it is perhaps not surprising that maternal depression varies by nativity status in the United States, but not in Australia. US immigrant mothers have significantly lower depression scores than their citizen counterparts, and the protective effect of foreign birth is robust across several specifications that model maternal, child, and household characteristics. By comparison, foreign-born Australian mothers are not protected from depressive symptoms following the birth of a child, and in fact, appear to be at higher risk of depression following childbirth. Nevertheless, lack of English proficiency does influence the risk of maternal depression in both countries, albeit in opposite ways. That Australian immigrant mothers who do not speak English are at significantly higher risk of depression than English-fluent moms is consistent with the linguistic homogeneity of the nation, despite the fact that nearly one in four residents are foreign-born. Stated differently, because English proficiency is a criterion for admission, mothers 24

26 who do not speak well are at a considerable disadvantage in procuring needed social and emotional support following childbirth. Although language can be a formidable barrier in the United States as well, the existence of large co-ethnic social networks where home country languages proliferate, can facilitate new mothers procurement of social supports that reduce stress following childbirth. Our data does not allow us to evaluate this claim, but it is consistent with the acculturation literature and worthy of further empirical scrutiny. As a measure of acculturation, length of residence in the host country does not conform to the predictions of straight-line assimilation in either country. In the United States, recent immigrants experienced the highest risk of depression and were similar to natives, but there is also evidence that the nativity protective effect against maternal depression fades after about a decade. Length of residence did not differentiate risks of maternal depression in Australia, unlike English proficiency. The distinct contexts of reception represented by our comparison of Australia and the United States proved instructive for understanding the association between nativity and maternal wellbeing following childbirth. In Australia, English proficiency is a required skill and mothers who are not fluent are at an increased risk for maternal depression. Conceivably these mothers face high levels of social isolation compared with immigrant mothers who are able to communicate in English. The contrasting results in the United States suggest that immigrant mothers are more likely to be socially connected even without English proficiency compared to their Australian counterparts, perhaps an artifact of family reunification as the most common mode of immigration. Based on our results, it appears that acculturation is very much a risk factor for maternal depression in the United States, much like a wide variety of other health 25

27 outcomes; yet it is protective against maternal depression in Australia, at least in terms of English proficiency. It is difficult to disentangle further what aspects of immigrant social policy drive the salience of nativity as a correlate of maternal depression. Australian immigrant women qualify for most benefits immediately and all benefits after two years of residence, which may partly explain why length of residence is not a major determinant of maternal depression. In the United States, it appears that recent immigrants (less than five years of residence) do not enjoy the protective effect of foreign-born status on maternal depression, as their risks are similar to native-born mothers. It is plausible that the lack of access to social benefits and healthcare, coupled with acculturative stress, exacerbates depression risk for recent immigrants relative to immigrants who have been in the country for longer periods. With increasing restrictions for immigrants at the state-level, it is important to consider the potential consequences for the wellbeing of mothers and thus their families of creating more hostile context of reception policies. Like many prior studies, ours features several limitations that warrant discussion and further research. First, the two datasets do not use identical measures of maternal depression, and we are unable to conduct a direct comparison of the prevalence of maternal depression in the United States and Australia. It would be useful for future comparative work if these surveys could employ the same measurement scales. That said, the LSAC and ECLS-B are remarkably similar in many ways, which allowed for creation of identical or very similar measures for all major theoretical constructs. Second, although both the LSAC and ECLS-B are longitudinal studies, we are limited in our ability to exploit the temporal ordering in the data because both studies interview mothers about one year after the birth. Finally, mental health history is 26