Maternal and Child Health Journal

, 10:139

Mother-Infant Interaction, Life Events and Prenatal and Postpartum Depressive Symptoms Among Urban Minority Women in Primary Care

Authors

    • Department of PsychiatryChildren's Hospital of Philadelphia
    • Department of PsychiatryChildren's Hospital of Philadelphia
  • Luis H. Zayas
    • Washington University in St. Louis
  • M. Diane McKee
    • Albert Einstein College of Medicine
Article

DOI: 10.1007/s10995-005-0042-2

Cite this article as:
Boyd, R.C., Zayas, L.H. & McKee, M.D. Matern Child Health J (2006) 10: 139. doi:10.1007/s10995-005-0042-2

Objectives: Prenatal and postpartum depression are significant mental health problems that can have negative effects on mother-infant interactions. We examined the relationships among mother-infant interactions, depressive symptoms, life events, and breastfeeding of low-income urban African American and Hispanic women in primary care settings. Methods: Participants were 89 African American and Hispanic women who were part of a larger mental health intervention study conducted in community health centers. Questionnaire data on depression, as well as negative and positive life events, were collected during pregnancy and at three-months postpartum, while mother-infant interaction observations and breastfeeding practice were only collected at three-months postpartum. Results: The ratings of maternal behavior for ‘depressed’ mothers did not differ from ‘nondepressed’ mothers. Except for gaze aversion behavior, infants' behavior while interacting with their mothers did not differ by maternal depression level. Hierarchical regression analyses revealed that maternal positive life events positively predicted infant interactional summary ratings, while maternal negative life events were inversely associated with maternal interactional summary ratings. Conclusions: To improve services in primary care, perinatal screenings for depression can help identify those women most at risk. When follow-up use of structured diagnostic instruments is not possible or cost-effective, clinician assessment of severity of depression will determine women with clinical levels of depression. Reducing negative life events is beyond the control of women or clinicians but cognitive interventions to help women focus on positive life events can reduce the deleterious effects of depression on mothers and their infants.

KEY WORDS:

perinatal depressionmother–infant interactionethnic minority

INTRODUCTION

Depression is a significant mental health problem in the lives of women and their families, especially those in low-income settings. One recent community-based survey, the National Comorbidity Survey-Replication, showed that 6.6% of women experienced a major depressive disorder in the preceding year and 12.9% during their lifetime (1). It has been estimated that approximately 10–15% of women develop depressive symptoms after childbirth (2, 3); some of these reaching levels of major depression, sometimes requiring hospitalization and medication. Some studies have shown rates of 10–30% for depressive symptoms during pregnancy (4, 5), although other studies with substantial proportions of African American and Hispanic women report higher rates in their samples (6, 7). Because of concern with the effects of maternal perinatal depression on mothers' psychosocial functioning and the interactions with their infants, which bear on children's developmental outcomes, women of childbearing age are an important group to target in both research and intervention (8). It is with this in mind that we report on a longitudinal study that examined perinatal depressive symptoms, life events, breastfeeding, and the interactions with their infants in a group of low-income African American and Hispanic (mostly Puerto Rican and Dominican) women receiving obstetrical services at community health centers in the south Bronx of New York City.

Depression's Impact on Mother's Interaction with Her Infant and Infant Functioning

A particular concern for women during pregnancy and the postpartum period is the potential impact of depressive symptoms on the early interaction with their infants. Depressed mothers have difficulty regulating their sense of irritation and intrusiveness when interacting with their infants; report lower levels of self-efficacy; are more likely to be distracted and preoccupied, more self-focused and less focused on their child; are less likely to match their children's mood states; are more punitive, rejecting and hostile in their childrearing, and are often less nurturant and responsive with their infants than non-depressed mothers (922). Additionally, a depressed mother is more likely to fail to give her infant the appropriate levels of stimulation to facilitate development of emotional expression and regulation (23). In general, depression is related to poor mother-infant interaction which may be the intermediate point between maternal depression and child psychophysiological and behavior difficulties (24).

There is extensive literature that documents the untoward developmental and behavioral effects that maternal depression has on infants. Dawson and colleagues (25) found that infants of depressed mothers show atypical frontal lobe electrical activity while interacting with the depressed mother and a non-depressed adult. These infants are more likely to have feeding problems (26, 27). Children of depressed mothers are at high risk for exhibiting developmental delays, neurochemical imbalances, and externalizing and internalizing behavior problems (2830). These children have a greater likelihood of having an insecure attachment to their depressed mothers than children of non-depressed mothers (13, 28, 31, 32). Infants of depressed mothers exhibit a range of behavioral and state disturbances, such as high levels of withdrawal, irritability, less exploring by mouth, as well as poor motor, cognitive and developmental progress. They are also less engaged and responsive in their interactions compared to infants of non-depressed parents (11, 27, 3336).

Life Events

Both negative and positive life events have been examined in relation to depression and can be a source of stress. Nevertheless, more research has focused on negative life events and its association with depression. In particular, research shows that negative life events are correlated with anxiety and depression among pregnant women in middle and late pregnancy, with high-risk pregnancies, and with negative feelings about pregnancy (3740). Negative life events serve as a risk factor for postpartum depression (41). Low income status and residing in urban settings that are characterized by poor housing, crowding, and crime are but some of the contextual and chronic life events that minority families contend with that can have profound impacts on parental functioning (42).

On the other hand, there is a paucity of similar research on positive life events and perinatal depression. One study demonstrated that depressed postpartum women reported fewer positive events than non-depressed postpartum women (43). Fewer positive events have been associated with a higher negative and critical affective style for depressed mothers during interactions with their children (44). Among adults, there is some empirical support suggesting that positive life events serve as a protective factor for depression (45, 46); however, the association between positive life events and depression appears to be complex and not well understood.

In this paper, we present findings on perinatal depression and mothers' interaction with their infants. We expect that interactions of depressed mothers will be impaired as compared to those of non-depressed mothers, as is consistent with the research literature. This paper also examines the association between mother-infant interaction and prenatal and postnatal depression and life events. We expect that higher depression levels and negative life events will be associated with lower mother-infant interactional ratings while more positive life events will be associated with higher ratings. We also explore how mother-infant interactions are associated with sociodemographics and breastfeeding. This study is unique because few studies have directly examined risk and protective factors in mother-infant interactions among depressed and non-depressed African American and Latino mothers. Additionally, the study was conducted in the “natural” environment of urban community health centers, where many women of low socioeconomic status (SES) receive their obstetric care and where the first line of mental health assessment and treatment typically takes place.

METHOD

Research Design

The study, approved by human subject boards of Fordham University, Albert Einstein College of Medicine, and Montefiore Medical Center, entailed identifying women through prenatal care team records who met inclusion criteria and approaching them early in the third trimester of pregnancy. Eligible women were 18 years or older, of African American or Hispanic background, and experiencing a low risk pregnancy. When approached about participating, prospective subjects were told that the study would require four to six months of participation, and a random assignment to one of two interventions if they were found to have elevated levels of depressive symptoms or to a comparison group if they had few or no symptoms of depression. From the potential sample of women who initiated prenatal care at the health centers during this period (n = 1056), 549 (52%) were not eligible to participate. Of the 507 eligible women, 187 (37%) could not be contacted. Of the 320 women who were eligible and able to be contacted, 187 (58%) participated in the prenatal assessment. One hundred ten women (60%) participated in the three-month postpartum assessment.

The two interventions that participants could be randomly assigned to were (1) an enhanced psychosocial intervention or (2) the health centers' usual psychosocial services (or treatment-as-usual, TAU). Those women with little or no depressive symptoms were not offered any intervention although they could utilize other clinic social services. The enhanced psychosocial intervention was comprised of three components delivered by graduate social work students: an empirically tested eight session cognitive-behavioral depression (CBT) prevention “course” (47); a four-module psychoeducational course on child development that emphasized sensitive, responsive mothering; and ongoing social support building sessions. TAU services in the community health centers typically consisted of psychological and social services provided by case aides, master-degreed social workers, graduate students, or doctoral level psychologists. Typically, these services included case management, advocacy, and supportive therapy. After the initial data collection point, women completed questionnaires at three-months postpartum when they exited the study. In the third month, mother-infant interaction was videotaped. Analyses of the effects of the two interventions on maternal functioning revealed no difference: both groups showed reduction in depressive symptoms. In regard to mother-infant interactions, the only difference detected was that infants in TAU condition displayed higher physical activity than infants in the enhanced psychosocial intervention (48). As a result of these findings, the participants were not categorized according to the intervention they received for this study.

Participants

There were 89 women who participated with their infants in the videotaped interaction at three months postpartum (i.e., 48% of the women who participated in the prenatal assessment and 81% of the women who participated in the three month postpartum assessment). Approximately 56% (n = 50) of the women self-identified as African American, 42% (n = 37) as Hispanic, and 2% (n = 2) did not identify their race/ethnicity. The Hispanic population included 28 Puerto Ricans, 4 Dominicans, 3 Central and South Americans, and 2 Mexicans. Eighty-two percent of the women had grown up in the United States. The participants' ages ranged from 18 to 41 years with a mean age of 24.6 (SD = 5.7) and their highest educational level ranged from 8 to 17 years with a mean level of 12.3 (SD = 2.1). Sixty percent (n = 53) of the participants were single, 15% (n = 13) were married and living with their husbands, 20% (n = 18) were co-habitating with a partner, 3% (n = 3) were separated or divorced, and 2% did not report marital status (n = 2). Women who participated in the videotaped interaction were more likely to be African American (X2(1) = 13.28, p < 0.01) than women who participated in the prenatal assessment. However, there were no significant differences in age, educational level, marital status, prenatal depression score, and intervention condition (i.e., CBT, TAU, or non-depressed comparison), between women who were videotaped and those who were not.

Measures

Depression

We used the Beck Depression Inventory, Second Edition [BDI-II; (49)], a 21-item self-report inventory derived from clinical observations about the attitudes and symptoms displayed by depressed patients. The BDI-II assesses the intensity of depression and covers such categories as mood, pessimism, guilt, sense of failure, suicidal thoughts, and fatigue. The BDI-II has shown an internal consistency of 0.93 for a group of non-patients and 0.92 for outpatients, and a test-retest correlation of 0.93 for outpatients over the course of a week. Internal consistency in this study was 0.89. We used a cut-off score of 14 to identify depression in our sample as suggested in the BDI-II manual.

Life Events

We used 67 of 82 items of the Revised Life Events Questionnaire (50, 51) to identify negative and positive life events. The Revised LEQ contains items commonly used in life event inventories plus nine additional items of special relevance to female respondents established by Norbeck (50). The modified LEQ asks respondents to indicate the events they have experienced during the past year. Examples of these life events are: major personal illness, troubles at work with employer or co-workers, difficulty finding housing, infidelity, being robbed, taking a trip, gaining new friends, acquiring a pet, beginning school, moving to a new town, etc. Once an event is identified in the questionnaire, respondents rate the valence (either positive or negative) and impact of the events on their lives on a 4-point Likert scale. A negative impact score is derived from summing up the impact scores for negative events and a positive impact score by summing up the impact scores of the positive events (52). Norbeck (50) reported test-retest reliability for the Revised LEQ after one week to be 0.78 to 0.83. In our study, we used the total number of negative life events as the negative life events score and the total number of positive life events as the positive life events score.
Table I.

Comparisons of Mother-Infant Interaction Ratings

Interaction Ratings

Non-depressed

Depressed

Total

Mean

SD

Mean

SD

Mean

SD

Mother

 State

2.88

0.33

2.83

0.39

2.87

0.34

 Activity

2.39

0.67

2.39

0.72

2.39

0.68

 Orientation

2.58

0.70

2.74

0.45

2.62

0.64

 Gaze

2.86

0.39

2.83

0.39

2.85

0.39

 Silence

1.85

0.83

1.83

0.89

1.84

0.84

 Expressions

2.46

0.73

2.30

0.93

2.41

0.78

 Vocalizations

2.19

0.73

2.13

0.76

2.17

0.73

 Infantized

2.42

0.70

2.39

0.72

2.41

0.70

 Contingent response

2.42

0.70

2.39

0.72

2.41

0.70

  Game playing

2.64

0.58

2.52

0.73

2.61

0.62

  Summary

2.45

0.37

2.43

0.36

2.45

0.36

n

59

23

82

Infant

 State

2.84

0.45

2.92

0.28

2.87

0.41

 Activity

2.78

0.46

2.71

0.55

2.76

0.49

 Orientation

2.26

0.76

2.13

0.80

2.22

0.77

 Gaze

 1.88*

0.82

 1.42*

0.78

1.74

0.83

 Expressions

2.50

0.71

2.29

0.62

2.44

0.69

 Vocalizations

1.69

0.75

1.42

0.50

1.61

0.70

 Fussiness

2.52

0.68

2.42

0.65

2.49

0.67

 Summary

2.35

0.41

2.20

0.36

2.31

0.39

n

58

24

82

*p < 0.05 using MANOVA.

Mother–Infant Interaction

In the third month of their babies' lives, mothers were videotaped interacting with their infants. Mothers were brought to the university campus and ushered to a room where they could sit on a cushioned quilt on the floor or a chair surrounded by colorful small toys. Mothers were instructed to interact with the infant in the same manner they would at home and given 20 min to do so. While our plan was to code only a three-minute segment of the videotaped interaction, we chose to extend the recordings in order to give mothers copies of their tapes.

Expert coding was conducted using the Interaction Rating Scales [IRS; (53, 54)]. The IRS has two sets of Likert-type categorical scales: one for the infant's and one for the mother's behavior. The infant ratings include state (drowsy or alert), physical activity, head orientation, gaze aversion behavior, facial expressions, fussiness and vocalizations. Maternal interactional ratings include state, physical activity, head orientation, gaze aversion behavior, silence during infant gaze aversion, facial expression, vocalizations, infantized behaviors (i.e., imitative behaviors such as cooing), contingent responsivity, and game playing. Each behavior is rated on a 3-point scale with 1 being the worst behavior (e.g., infant predominantly drowsy; mother never imitative of infant or no simplified behaviors) and 3 being optimal (e.g., infant not fussing, rare gaze aversion; mother constantly looks at infant, contingent talking and sensitive pacing of vocalization). Scores on the items are then averaged for mother and infant.

The videotapes were sent to Dr. Tiffany Field, the developer of the IRS, for expert coding. Coders were blind to the mother's depression level and group condition. The coders were trained to 0.90 inter-rater reliability criteria on both the infant and maternal interactional rating items before they coded the videotapes. Each video was fast-forwarded to the fifth minute of the videotape and the next consecutive three minutes were used for coding the interaction. This allowed time for mothers to settle in with their child to the videotaping and we felt it would give us a better sample of mothers' behaviors once adjusted to the setting. When extended feeding of infants occurred, coders fast-forwarded to a point in the tape when the baby was not being fed and resumed interaction coding at that point to complete the three minutes.

Breastfeeding Practice

To assess breastfeeding behaviors at 3-months postpartum, mothers were asked if they were breastfeeding. A mother could respond by indicating that she was breastfeeding her infant exclusively, breastfeeding and bottlefeeding, or bottlefeeding her infant exclusively. If she reported having started and ceased breastfeeding, she was asked when this occurred.

RESULTS

Comparisons of Mother–Infant Interactions

To examine the impact of maternal depressive symptomatology on mother-infant interactions, the sample was divided into a ‘depressed’ group (n = 24) and a ‘non-depressed’ group (n = 53) based on their three-month BDI-II scores. A score of 14 and above identified the ‘depressed ‘ group. Table I displays means and standard deviations of the maternal and infant interaction ratings for the two groups and the total sample. Multivariate analyses of variance (MANOVA) were conducted separately for maternal and infant ratings. Findings showed that the ratings of maternal interactional behavior were similar for both the ‘depressed’ and ‘non-depressed’ groups. Similarly, infant ratings did not differ across groups, except that infants of ‘depressed’ mothers exhibited more gaze aversion than infants of ‘non-depressed’ mothers (F(1,80) = 5.59, p < 0.05). These results were also found when covarying intervention status.

Comparisons of Prenatal and Postpartum Ratings of Depression, Life Events, and Breastfeeding

To explore the differences between ‘depressed’ and ‘non-depressed’ mothers on the other variables of interest, a MANOVA was conducted for the continuous variables and a chi-square analysis was conducted for the categorical variable. Table II illustrates the means and standard deviations of the depression and negative and positive life events at prenatal and postpartum for the two groups and the total sample. MANOVAs showed that the ‘depressed’ mothers exhibited higher levels of prenatal (F(1,85) = 25.72, p < 0.01) and postpartum (F(1,85) = 193.95, p < 0.01) depression compared to ‘non-depressed mothers’ as expected. In regards to life events, ‘depressed’ mothers reported more postpartum negative life events than ‘non-depressed’ mothers (F(1,85) = 15.48, p < 0.01). A chi-square analysis was conducted to examine breastfeeding among the two groups. Forty percent of the ‘depressed’ mothers breastfed and 33.9% of the ‘non-depressed’ mothers breastfed, however, this difference was not significant (X2(1) = .29, p>0.05).
Table II.

Comparisons of Maternal Ratings of Depression and Life Events

 

Non-depressed

Depressed

Total

Mean

SD

Mean

SD

Mean

SD

Prenatal

 Depression

12.87*

6.86

 22.04*

9.11

15.40

8.55

 Negative life events

7.71

5.51

10.00

5.80

 8.34

5.65

 Positive life events

7.11

3.56

5.79

2.32

 6.75

3.31

Postpartum

 Depression

 5.81*

3.68

 18.79*

4.39

 9.39

6.99

 Negative life events

 5.38*

4.44

 9.45*

3.98

 5.51

4.67

 Positive life events

6.76

3.88

7.88

3.84

 7.06

3.88

n

63

24

87

*p < .05 using MANOVA.

Regression Analyses

The relationships between maternal-infant interactions and maternal reports of depression symptoms and life events were examined using hierarchical multiple regression. Dummy variables were created for race and ethnicity (0: African American, 1: Latina) and breastfeeding status (0: not breastfeeding, 1: breastfeeding at least partially). Hierarchical multiple regression models were calculated separately for maternal and child average summary ratings. Separate hierarchical regressions were also conducted for maternal report during pregnancy (baseline; see Table III) and at three-months postpartum (concurrent with the collection of the maternal-infant interactional data; see Table IV). The order for both of the regression models was as follows: Step 1—control variable (intervention status); Step 2—maternal sociodemographic variables (race/ethnicity, age, education, marital status and breastfeeding status for postpartum assessment); and Step 3—maternal report of depression and negative and positive life events. Depression was significantly correlated with negative life events (rs = 0.40–0.41, ps < 0.001) at both time points, while negative life events were positively correlated with positive life events at 3 months postpartum (r = 0.27, p < 0.05).
Table III.

Hierarchical Regression Analyses of Prenatal Maternal Variables Predicting Postpartum Mother and Infant Interaction Ratings

 

Mother summary rating

Infant summary rating

B (SE B)

Beta

B (SE B)

Beta

Step 1: Control

 Intervention condition

−0.05 (0.05)

−0.11

0.06 (0.05)

 0.13

R2

0.01

 

0.02

 

F

F(1, 85)  =  1.06, ns

 

F(1,85)  =  1.38, ns

 

Step 2: Demographics

 Race/ethnicity

0.13 (0.08)

 0.17

0.03 (0.09)

 0.04

 Age

0.01 (0.01)

 0.08

−0.00 (0.01)

−0.04

 Marital status

0.05 (0.04)

 0.15

−0.03 (0.05)

−0.08

 Education

0.01 (0.02)

 0.08

0.01 (0.02)

 0.05

R2

0.09

 

0.03

 

F

F(5,81)  =  1.59, ns

 

F(5,81)  =  0.42, ns

 

Step 3: Maternal report of depression and life events

 Depression

0.01 (0.01)

 0.21

0.01 (0.01)

 0.27

 Life events-negative

−0.01 (0.01)

−0.22

−0.00 (0.01)

−0.09

 Life events-positive

0.02 (0.01)

 0.19

0.03 (0.01)

 0.28*

R2

0.14

 

0.11

 

F

F(8,78)  =  1.55, ns

 

F(8,78)  =  1.15, ns

 

*p < .05.

Table IV.

Hierarchical Regression Analyses for Postpartum Maternal Variables of Mother and Infant Interaction Ratings

 

Mother summary rating

Infant summary rating

B (SE B)

Beta

B (SE B)

Beta

Step 1: Control

 Intervention condition

−0.05 (0.05)

−0.11

0.06 (0.05)

 0.13

R2

0.01

 

0.02

 

 Significant F

F(1,83)  =  0.92, ns

 

F(1,83)  =  1.37, ns

 

Step 2: Demographics and breastfeeding

 Race/ethnicity

0.19 (0.08)

  0.25*

0.04 (0.09)

 0.05

 Age

0.00 (0.01)

 0.02

−0.00 (0.01)

−0.05

 Marital status

0.04 (0.04)

 0.11

−0.03 (0.05)

−0.08

 Education

0.01 (0.02)

 0.04

0.00 (0.02)

 0.04

Breastfeeding

0.22 (0.09)

  0.28*

−0.03 (0.10)

 0.04

R2

0.17

 

0.03

 

F

F(6,78)  =  2.67, p < .05

 

F(6,78)  =  0.37, ns

 

Step 3: Maternal report of depression and life events

 Depression

0.00 (0.01)

 0.01

−0.01 (0.01)

−0.17

 Life events-negative

−0.03 (0.01)

 −0.33*

−0.00 (0.01)

 0.00

 Life events-positive

0.02 (0.01)

 0.19

0.03 (0.01)

 0.29*

R2

0.25

 

0.11

 

F

F(9,75)  =  2.82, p < .01

 

F(9,73)  =  0.98, ns

 

*p < .05.

After controlling for intervention status, maternal report during pregnancy of positive life events significantly predicted infant interactional rating (see Table III). That is, more positive life events were predictors of higher infant interaction scores. None of the maternal reports during pregnancy predicted maternal interactional ratings as Table III illustrates.

Table IV shows that at three-months postpartum, race/ethnicity and breastfeeding status were associated with maternal interactional ratings. To examine these findings further, Analyses of Covariance (ANCOVAs) controlling for intervention status were conducted separately for race/ethnicity and breastfeeding status. Results showed that Latinas had a higher average maternal interactional rating than African Americans (F(1,80) = 4.89, p < 0.05). Additionally, women who breastfed had a higher average maternal interaction rating (F(1,80) = 7.55, p < 0.01). As Table IV displays, maternal reports of postpartum negative life events were significantly and inversely related to maternal summary ratings. In regard to infant ratings, maternal reports of postpartum positive life events were associated with higher infant summary ratings. Because of the significant correlations between some of the independent variables, exploratory regression analyses were conducted with variables individually, however, the significance of outcomes was unchanged.

DISCUSSION

This study compared the mother-infant interactions of low-income urban African American and Hispanic women and examined the associations of these interactions with maternal antepartum and postpartum depression and life events. Contrary to the literature, the results show that women with higher symptoms of depression did not differ significantly from those with lower depressive symptoms in the manner in which they interacted with their infants. Similarly, most of the infants ratings did not differ across groups, except that infants with ‘depressed’ mothers were more likely to avert their gaze than infants of ‘non-depressed’ mothers. This particular finding is consistent with empirical evidence.

A possible explanation for the contrary findings is that women in our study did not necessarily meet clinical criteria for depression as we employed a symptom scale rather than a diagnostic measure. Therefore, many more women may have been assigned to the ‘depressed’ groups than may have actually met clinical criteria for minor or major depression. This may have been exaggerated by the cut-point chosen to define our ‘depressed’ group. For research purposes with pregnant samples, it is advantageous to use structured diagnostic interviews, which are considered the gold standard in determining diagnosis, set the BDI-II cut-off higher than is recommended in non-pregnant samples [e.g., raise to  = /> 16; (54, 55)] or remove the somatic items (56). However, for clinical purposes such as those in urban primary care, maintaining the usual cut-off score (i.e., BDI-II score ≥14) allows for greater sensitivity in screening for perinatal depression. While screening may identify more women as having signs of depression (i.e., false positives), it would avoid missing any women who are depressed but are missed (i.e., false negatives). As pregnant and post-partum women are identified, follow-up assessments with structured diagnostic instruments that provide greater specificity can be implemented. As the reality is that such instruments are not always cost-effective in some community health centers, clinical assessments by experienced psychosocial clinicians who are familiar with perinatal depression can be instituted regularly. As mental health symptoms are often overlooked in the context of day-to-day primary care practice, this approach will ensure that few if any perinatal patients—especially disadvantaged, urban minority women—are missed.

A second explanation for this contradictory finding is that for minority women, the impact of a high level of depressive symptoms is not directly manifested in mother-infant interactions. For example, among rural African American families, maternal depression was not linked to family processes, such as mother-child relationship and family routines (57), as had been demonstrated with white families (58, 59). Another study did not find a direct relationship between primary caregiver's mood and parenting in a sample of African American families (60). These examples demonstrate the need to examine maternal and infant interactions among different ethnic groups.

An important finding is that positive life events play a small but important role in mother-infant interaction. The results suggest that positive life events could provide a protective effect for at-risk families in spite of the negative impact of stressful life events. Nevertheless, there is limited research examining positive life events with more emphasis placed on negative life events (61). Our results supported the link between negative life events and lowered maternal interactional ratings during interactions. Although the R2s were significant, they were relatively low, suggesting the complexity of factors that may be involved in mother and infant interactions.

Our findings show that Hispanic women had higher average maternal interactional ratings than African Americans, a finding that is different from those reported in other research (14). This finding may be an artifact of the study sample. That is, a self-selection process may have occurred since more Hispanic mothers dropped out of the study before the videotaping sessions than African Americans, who tended to remain in the study. One possible explanation is that the Hispanic women who remained were generally the higher functioning ones. Our data show that women in greatest need of intervention tended to drop out in greater numbers (62). Two additional points about this ethnic difference in maternal interactional ratings may clarify this finding. First, differential attrition can also be attributed to the apparent pattern of higher residential stability of African Americans than Hispanics in the communities served by the health centers (7). Second, we used a Hispanic population made up largely of Puerto Ricans and Dominicans, which stands in contrast to other studies using the same method that included other Hispanic groups, mainly Cuban (14). Clearly, more research is needed to see if this is indeed an effect of subculture.

Methodological limitations may also account for our findings. As stated earlier, women identified as ‘depressed’ may not have had a clinical level of depression. The depression scores decreased from pregnancy to postpartum for those women identified as ‘depressed’ during pregnancy regardless of intervention (48). Attrition was high, due in part to the length of time women were asked to be in the study as well as the aforementioned issue of residential stability, and of families' moving their health care to other facilities. Although we employed a well-tested coding system for maternal-infant interaction, our videotaping approach was different from the format used in the original interaction rating scale development and subsequent research with it. Thus, this situation may have reduced the saliency of the behaviors typically coded with the scoring system. Future efforts need to include a standard approach to permit sensitivity and comparability with other studies.

Notwithstanding its limitations, the present study points to the need for more research on maternal depression and mother-infant interaction among ethnic and racial minority populations. Not only are they understudied groups in this area, but they often face social, demographic, and psychological risk factors that can negatively affect the mental health of both mother and infant. Studies that examine the natural progression of depression from the prenatal to the postnatal period, and that test interventions to enhance maternal functioning and interaction with offspring are needed. Recent reports (54) show improvements in maternal positive affective involvement and verbalization following interventions for postpartum depression. The primary care setting, with its familiarity to urban families and its unique position to provide frontline mental health prevention programs, holds special promise for addressing the early effects of maternal depression.

In particular, our results have implications for interventions with low-income, pregnant and post-partum women and their infants. Our study suggests that domains such as breastfeeding (63) and life events can be the focus of interventions. Enhancing breastfeeding has the positive effect of improving mother-infant bonding and interaction. Likewise, employing cognitive interventions that focus women on positive events in their lives may reduce the attention on negative events and their consequent impact on maternal mood and interaction with infant. While it is not possible to control negative events in the lives of distressed urban women, attention to controllable, positive activities, such as learning about infant development, interacting sensitively to infant, and increasing social support in the form of friendships and social involvement, may be areas with possible payoffs for intervention and service research.

ACKNOWLEDGMENTS

This research was supported by grant R24MH57936 from the National Institute of Mental Health to Luis H. Zayas. Dr. Boyd's effort on this project was supported by grants from the W. E. B. Du Bois Collective Research Institute, University of Pennsylvania and K01 MH68619 from the National Institute of Mental Health. We are grateful to the women who gave of their time to participate in this study and to be observed with their infants. We thank Drs. Tiffany Field and Maria Hernandez-Reif and their assistants at the University of Miami for providing expert coding of the mother-infant videotapes.

Copyright information

© Springer-Verlag 2005