TARGETING ADOLESCENT MOTHERS WITH DEPRESSIVE SYMPTOMS FOR EARLY INTERVENTION.

Field, Tiffany ; Pickens, Jeffrey ; Prodromidis, Margarita 等

ABSTRACT

Infants of mothers with depressive symptoms show developmental delays if symptoms persist over the first 6 months of the infant's life, thus highlighting the importance of identifying those mothers for early intervention. In Study 1, mothers with depressive symptoms (n = 160) and mothers without depressive symptoms (n = 100) and their infants were monitored to identify variables from the first 3 months that predict which mothers would still be symptomatic at 6 months. A "dysregulation" profile was noted for the infants of depressed mothers, including lower Brazelton scores, more indeterminate sleep, and elevated norepinepbrine, epinephrine, and dopamine levels at the neonatal period, and greater right frontal EEG activation, lower vagal tone, and negative interactions at the 3- and 6-month periods. A group of maternal variables from the neonatal and 3-month assessments accounted for 51% of the variance in the mothers' continuing depressive symptoms. These variables included greater right frontal EEG activatio n, lower vagal tone, and less positive interactions at 3 months, and elevated norepinephrine, serotonin, and cortisol levels at the neonatal stage. In Study 2, a similar sample of mothers with depressive symptoms (n = 160) and without depressive symptoms (n = 100) was recruited and followed to 3 months. Those symptomatic mothers who had values above (or below) the median (depending on the negative direction) on the predictor variables identified in Study 1 (taken from the first 3 months) were then randomly assigned to an intervention or a control group at 3 months. These groups were then compared with each other, as well as with the group without depressive symptoms, at 6 and 12 months. The intervention, conducted from 3 to 6 months, consisted of free day care for the infants and a rehab program (social, educational, and vocational) plus several mood induction interventions for the mothers, including relaxation therapy, music mood induction, massage therapy, and mother-infant interaction coaching. Although th e mothers who received the intervention continued to have more depressive symptoms than did the nondepressed mothers, their interactions significantly improved and their biochemical values and vagal tone normalized. Their infants also showed more positive interations, better growth, fewer pediatric complications, and normalized biochemical values, and by 12 months their mental and motor scores were better than those of the infants in the control group.

Longitudinal studies on mothers with depressive symptoms have examined various kinds of depression, such as postpartum depression, dysthymia, and major depressive disorder, and various degrees of chronicity (Campbell, Cohn, & Meyers, 1995; Field, Healy, Goldstein, & Guthertz, 1990; Lyons-Ruth, Zoll, Connell, & Grunebaum, 1986; Murray, 1992). Irrespective of the type of depression, mothers' depressive mood states appear to affect infants' development negatively (Beardslee, Bemporad, Keller, & Klerman, 1983; Field, 1984; Orvaschel, 1983; Zuckerman & Beardslee, 1987). Findings indicate that infants and children of depressed mothers are more likely to have problems, including sleep disorders, accidents, growth failure, and psychosomatic complaints. For example, in a prospective study, Radke-Yarrow, Cummings, Kuzynski, and Chapman (1985) noted that clinically depressed mothers displayed very little affection, and their toddlers showed greater sadness, talked less often, and engaged in less exploratory behavior.

The typical paradigm has been to study the face-to-face interactions of depressed mothers and their infants. In this context, depressed mothers have been described as having flat affect or depressed mood, and as being less vocal and less responsive to their infants (Cohn, Campbell, Matias, & Hopkins, 1990; Field, Healy, Goldstein, & Guthertz, 1990). This, in turn, may affect their infants' language development, problem-solving ability, mastery motivation, and social competence.

The type of depression may be less critical than whether the mother has chronic depressive symptoms (Campbell et al., 1995; Sameroff & Seifer, 1983). Most studies on postpartum depression have found that the majority of mothers experience remission in the first few months after childbirth. However, Lyons-Ruth et al. (1986) reported that only 16% experienced remission in the first year, and Field et al. (1990) found that only 20% of the mothers who had depressive symptoms when their infants were 3 months old were symptom-free when they were 6 months old. Further, Field (1992) noted that infants whose mothers' depression went into remission showed significant improvement in their interactional behavior and developmental performance. In contrast, infants whose mothers still had symptoms when they were 6 months old showed developmental delays at 12 months. Because intervention is probably needed most by mothers who have chronic symptoms (Campbell et al., 1995), it is important to find measures that can identify those mothers.

Behavioral remission (e.g., the abatement of such things as slow speech, slouched posture, and a slow gait) is often noted even when a person continues to report depressed feelings. On the other hand, some physiological variables appear to be stable even when behavioral symptoms are not present. For example, Henriques and Davidson (1990) have reported that depressed subjects show greater relative right frontal EEG activation, which remains after they go into remission and no longer show behavioral symptoms, suggesting that this might be a physiological characteristic of chronic depression. Similarly, Field, Fox, Pickens, Nawrocki, and Soutulla (1995) documented greater relative right frontal EEG activation in chronically depressed mothers and their 3-month-old infants. Heart rate is another potential predictor variable, inasmuch as high, stable resting heart rate has been reported for chronically depressed subjects. For example, Field, Pickens, Fox, Nawrocki, and Gonzalez (1995) found lower vagal tone (and h igher heart rate) in mothers with chronic depressive symptoms and their infants. Field (1995) also documented elevated cortisol and catecholamines in mothers with chronic depression.

The purpose of the following two studies was, first, to identify variables from the first 3 months of an infant's life that predict the mother's continuing depression at 6 months (Study 1). Second, in Study 2, those predictor variables were used to identify a sample of high-risk mothers, and the effectiveness of an intervention program was tested.

STUDY 1

Study 1 was designed to identify variables that predict which mothers would show continuing depressive symptoms at 6 months postdelivery. Behavioral, psychophysiological, and biochemical assessments were made at the neonatal, 3-month, and 6-month periods. Variables from the neonatal and 3-month assessments were entered into a regression model to determine their contribution to the variance in the mothers' depressive symptoms at 6 months.

METHOD

Sample

The sample consisted of 260 mothers (70% of them adolescents) of normal, full-term infants. Approximately 63% were African American, 22% Hispanic, and 15% non-Hispanic White. They were recruited at the hospital postdelivery (mothers of preterm infants and infants with prenatal complications were excluded). The mothers, on average, were 17.8 years old, had 10.3 years of education, and their socioeconomic status was 4.4 on the Hollingshead Index. Adolescent and young adult mothers were recruited because it was thought that they would be more compliant with the intervention if they were still in high school and if the intervention and day-care program were situated in their high school. In addition, it was felt that the intervention might have more impact with younger women by helping them complete high school and begin career training. They were not breast-feeding their infants (as is typical of this population). The mothers were the primary caregivers (based on their reported plan to be the ones spending the greatest number of hours per day with their infants). The refusal rate was extremely low (7 mothers).

The Beck Depression Inventory (BDI) was used to classify the mothers according to depressive symptoms (dysphoria). While not providing a diagnosis of depression, self-report measures such as the BDI are typically used when the symptom of dysphoria is of primary interest (Seifer, 1995). Further, such dimensional measures, using continuous scales, typically offer more information and have better psychometric properties, making them more powerful in statistical tests (Cohen, 1990). Although dysphoric mood is only one of the symptoms in the depression complex, it is the primary symptom and the one thought to affect infants most. As Seifer (1995) has pointed out, another advantage of a single-symptom measure is that "it defines more precisely an individual characteristic that may be implicated in poor developmental outcomes" (p. 421). Thus, mothers with BDI scores greater than 12 (the cutoff for depression in most research protocols) were assigned to the depressed group (n = 160). Mothers with scores less than 9 on the BDI (and matched on socioeconomic status and age) were assigned to the nondepressed group (n = 100). Mothers with scores in the 9-12 range were not included, in order to ensure a large enough difference between the nondepressed and depressed groups. Others have noted that only a small proportion of women with depression seek treatment (Leaf, Bruce, Tischler Freeman, Weisman, & Myers, 1988; O'Hara, Zekoski, Philipps & Wright, 1990), and this held true for the mothers in the present study (not one was in treatment or taking antidepressant medication).

Neonatal Assessment

The mothers were interviewed, the infants were assessed, and mother-infant interactions were observed in the mothers' hospital rooms. Brazelton assessments, sleep/wake behavior, and urine sampling were conducted by a research associate who was blind to the group status of the subjects. Medical records were reviewed for standard birth measures (gestational age, birth weight, birth length, head circumference, and 5-minute Apgar scores). In addition, the ponderal index (ratio of weight to length, providing a measure of intrauterine growth deprivation) and the Obstetric and Postnatal Complications Scales (Littman & Parmelee, 1978) were used.

Two instruments were employed to assess maternal depression: the Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and the Diagnostic Interview Schedule for Children (Costello, Edelbrock, & Costello, 1985). The BDI is a 21-item questionnaire, with each item scored on a 4-point scale indicating the presence and severity of depressed feelings/behaviors/symptoms. It is commonly employed in research on nonclinically depressed samples, and it has reasonably good psychometric properties (Beck, 1976). The Diagnostic Interview Schedule for Children (DISC) addresses specific symptoms as well as their chronology and duration. This standardized diagnostic interview has a step structure that minimizes interviewing time. Reliability and validity of the DISC have been found to be as good as or better than other structured diagnostic interviews (Costello, Edelbrock, & Costello, 1985). The DISC was used to diagnose dysthymia (prevalent in earlier samples; see Field, 1995) and major depression (rarely diagnosed in previous samples of depressed adolescent mothers).

The Maternal Stress Interview (Field, 1980) consists of 52 questions concerning socioeconomic status, marital status, other family members, and home/life variables, such as crowding stress. The alpha coefficient for internal consistency was .82 for this scale.

The Problem Oriented Screening Instrument for Teenagers (POSIT; National Institute on Drug Abuse, 1987) was administered to the mothers to screen for social, emotional, and behavioral problems. This 139-item questionnaire measures various social, familial, physical, and mental stressors. Items are grouped into ten subscales in order to identify specific problem areas: Substance Use/Abuse, Physical Health Status, Mental Health Status, Family Relations, Peer Relations, Educational Status, Vocational Status, Social Skills, Leisure and Recreation, and Aggressive Behavior/Delinquency.

The Interaction Rating Scale (Field, 1980) was utilized to measure mothers' sensitivity to infant cues. Feeding interactions were coded by research assistants who were blind to the mothers' group status. Kappa coefficients for interobserver reliability averaged .82, and the alpha coefficient for internal consistency was .83.

The mothers provided a sample of first morning urine, and the infants' urine was collected on the second day. Urine samples were frozen and sent to Duke University to be assayed for catecholamines (norepinephrine, epinephrine, dopamine), 5-HIAA (serotonin metabolite), and cortisol (see Kuhn, Schanberg, Field, Symanski, Zimmerman, Scafidi & Roberts, 1991). Catecholamine and 5-HIAA assays were conducted by high-pressure liquid chromatography with electrochemical detection, cortisol was determined by radioimmunoassay, and creatinine was assayed calorimetrically (Kuhn et al., 1991).

In other studies, "flat" affect (limited number and variability of facial expressions) and lower activity levels in infants of symptomatic mothers have been noted (Abrams, Field, Scafidi, & Prodromidis, 1995; Field, 1992). Because these infant behaviors might contribute to disturbed interactions independent of the mothers' depressive behavior, the Brazelton Neonatal Behavior Assessment Scale was administered and sleep/wake behavior assessments were conducted. For the sleep-observation sessions (45 minutes), a research assistant continuously coded the infant's sleep/wake states and behaviors (using a laptop computer), yielding the percentage of time that the different sleep states and behaviors occurred (see Field, Schanberg, Scafidi, Bauer, VegaLahr, Garcia, Nystrom, & Kuhn, 1986). An adaptation of Thoman's sleep state criteria was used to define categories (Thoman, 1975). Prior to coding, the research assistant was trained to .90 reliability.

The Brazelton Neonatal Behavior Assessment Scale was administered midway between feedings and following the sleep observations (Brazelton, 1973). This scale consists of 20 neurological reflex items and 27 behavior items constituting 7 factors: habituation, orientation, motor behavior, range of state, regulation of state, autonomic stability, and abnormal reflexes (Lester, Hoffman, & Brazelton, 1985). At all assessment periods, infant growth (weight, length, and head circumference) was measured, because failure to thrive has been noted in some samples.

Assessments at Three and Six Months

The same measures were administered at these two periods. The order of the assessments was as follows: the maternal interview and EEG recordings occurred concurrently, next was the mother-infant face-to-face interaction, followed by saliva samples and growth measures.

The maternal assessment consisted of the same scales used at the neonatal period, including the BDI, the DISC, and the Maternal Stress Interview, repeated at 3 and 6 months. In addition, the mothers' perceptions of their infants' vulnerability were assessed using the Vulnerable Child Scale (VCS; Perrin, West, & Culley, 1989). The VCS, which is based upon Forsyth and Caney's Child Vulnerability Scale (1985), consists of 16 items centering on concern about the child's health (adapted here to be appropriate for infants). Examples include: "This infant seems to be more sickly than other infants" and "This infant seems to have as much energy as other infants of the same age." Each item is rated on a four-point Likert scale, with lower scores being optimal. The original scale was standardized on 320 mothers of 3 1/2 year old children, and both acceptable reliability and concurrent validity were reported. Perrin et al. (1989) also reported excellent test-retest reliability (r = .96), as well as acceptable internal reliability (Cronbach's alpha = .75). VCS scores range from 16 to 64, with higher scores in the present study indicating the perception of greater vulnerability.

Just prior to the mother-infant interaction session, EEG was recorded for mothers and infants. For the infant EEG, the infant was placed on the mother's lap and bubbles were blown to maintain the infant's attention. For the maternal EEG, the mother simply reclined with eyes closed for three minutes.

EEG was recorded using a stretchable lycra cap that was positioned on the subject's head using anatomical landmarks (Bloom & Anneveldt, 1982; Field, Fox, Pickens, Nawrocki, & Soutullo, 1995). Electrode gel was injected into the electrodes at the following sites: F3, F4, P3, P4, and Cz (used as the reference), and impedances were brought below 5,000 ohms. Additional electrodes were placed at the outer canthus of each eye to obtain the subject's EOG (electrooculogram), which was used to facilitate artifact scoring.

The signal was passed through a Grass Model 12 Neurodata Acquisition System with amplifiers set as follows: low-frequency filter, 1 Hz; high-frequency filter, 100 Hz; amplification, 20,000. The line frequency filter was on for all channels. The output from the amplifiers was directed to a Dell 325D PC fitted with an Analog Devices RTI-815 A/D board. The signal was sampled at a rate of 512 Hz and streamed to hard disk using data acquisition software (Snapstream v. 3.21, HEM Data Corp.).

EEG data were analyzed using an EEG analysis software package (EEG Analysis System v. 5.3, James M. Long, 1987-1990). The first step of this process involved the manual elimination of data that were unusable due to artifact (eye movements, muscle activity, or technical difficulties). The remaining artifact-free data were then spectrally analyzed using discrete Fourier transforms to yield power data for specific frequency bands. The average number of seconds of artifact-free data available was 222.1 and 198.5 for the mothers with depressive symptoms and without symptoms, respectively, and 120.1 and 125.9 for their respective infants. The infant EEG data were analyzed from 1 to 12 Hz in 1 Hz bins. For mothers, the alpha band analyzed was 8 to 12 Hz. Frontal alpha laterality ratios (FALR) were computed by dividing the difference between right and left frontal alpha powers by the sum of these powers. A score of zero represents hemispheric symmetry, a negative score represents greater relative right frontal activ ation, and a positive score represents greater relative left frontal activation. Data analyses were also conducted on the natural log power data for both hemispheres in the frontal and parietal regions.

The mothers and infants were then videotaped in a 3-minute face-to-face interaction. The mother was simply asked to "pretend you are playing with your infant at home" (Field, 1980). The infant was placed in an infant seat on a table approximately fifteen inches from the face of the mother, who was seated at the table. Two video cameras and a split-screen generator enabled simultaneous monitoring of the mother's face and torso and the infant's upper body. In addition, heart rate was simultaneously recorded (Field, Pickens, Fox, Nawrocki, & Gonzalez, 1995). EKG electrodes were placed on the right scapula, the lower left lateral costal margin, and the lower left vertebral region just above the waistline of the infants and mothers. Heart rate was then converted to vagal tone using a vagal tone monitor designed by Porges (1985).

Immediately prior to and 20 minutes following the interaction sessions, saliva samples were taken from the mothers and infants by placing a dental swab (dipped in lemonade crystals) along the gumline to be subsequently assayed for cortisol levels. This permitted a determination of the differential stressful effects of these interactions on the depressed and nondepressed mother-infant dyads (Field, Healy, Goldstein, Perry, Bendell, Schanberg, Zimmerman, & Kuhn, 1988). Urine cortisol, the serotonin metabolite 5-HIAA, and catecholamines (norepinephrine, epinephrine, dopamine) were assaysed from first morning urine samples brought to the lab by the mothers (although 24-hour urine samples would have been more reliable, the pilot study indicated compliance problems). These assays were included because mothers with depressive symptoms and their infants were found to differ from control groups on these measures in earlier studies (Field, 1995).

The videotapes were subsequently coded according to three levels of engagement: negative, neutral, and positive (Field, Healy, & LeBlanc, 1989). The coding was completed on a laptop computer whose software yielded a data matrix showing the percentage of time that the mothers and infants were in these states, as well as the percentage of time that they shared these states. The videotapes were also rated using the Interaction Rating Scale (Field, 1980), which includes the following behavior: (a) infant's state, physical activity, head orientation, gaze behavior, facial expressions, vocalizations, and fussiness; and (b) mother's state, physical activity, head orientation, gaze behavior, facial expressions, contingent responsivity, and game playing. These are scored on a 3-point Likert scale and then averaged to obtain a summary rating for both mother and infant. Field (1980) reported interrater reliabilities ranging from .81 to .96 (mean = .88). Interrater reliabilities were determined on the basis of replicati on coding of one-third of the videotapes and calculated using Kappa coefficients, correcting for chance agreement. Heart rate was analyzed off-line using a vagal tone monitor, such that the average heart rate and vagal tone could be estimated for each interaction situation.

RESULTS

Depressed and nondepressed groups were compared using multivariate analysis of variance. Where MANOVAs were significant, ANOVAs were conducted and post hoc comparisons were made using Bonferroni t tests. In addition, because of attrition across the 6-month period, analyses were conducted to determine whether the retained and attrition samples differed.

First, however, analyses were conducted on maternal background variables and infant gender to determine whether these needed to be entered as covariates in the MANOVAs. As can be seen in Table 1, the depressed and nondepressed groups did not differ on the demographic variables, including mother's age, education, socioeconomic status, and ethnic distribution, nor did they differ on background stress. As expected, the depressed group scored significantly higher on the Beck Depression Inventory, had a significantly greater incidence of dysthymia as measured by the DISC, and had higher risk scores on the POSIT.

Neonatal Variables

A MANOVA performed on the birth measures for the infants, including gestational age, birth weight, birth length, head circumference, ponderal index, 5-minute Apgar scores, and obstetric complications, was not significant (see Table 2). However, the infants in the depressed group had significantly more postnatal complications. A MANOVA performed on the Neonatal Behavior Assessment Scale scores was significant, and post hoc ANOVAs revealed that the infants of mothers with depressive symptoms had significantly lower habituation and orientation scores and higher depression scores (see Table 3). MANOVAs performed on the sleep/wake behaviors and Interaction Rating Scale scores were also significant, and post hoc ANOVAs indicated that the infants of depressed mothers experienced significantly less frequent active sleep, more frequent indeterminate sleep, less drowsy behavior, and more jerks/startles (see Table 4). In addition, both the depressed mothers and their infants received significantly lower scores in terms of interaction behaviors.

MANOVAs performed on the mothers' and infants' biochemical data were significant. Univariate ANOVAs revealed that norepinephrine, epinephrine, dopamine, and serotonin (5-HIAA) levels were significantly higher in mothers with depressive symptoms (see Table 5). ANOVAs for the infants' biochemical data revealed that the norepinephrine, epinephrine, and dopamine levels of the infants of mothers with depressive symptoms were significantly higher.

Three-Month and Six-Month Variables

MANOVAs conducted on the maternal interview variables at both the 3- and 6-months periods were significant. Post hoc ANOVAs indicated that the mothers in the depressed group continued to have higher Beck Depression Inventory scores and a greater incidence of dysthymia. Further, at 3 months, they considered their infants more vulnerable (see Table 6). On the mother-infant interaction variables, the depressed mothers and their infants not only received inferior Interaction Rating Scale scores at both time periods, but also showed less frequent positive and/or more frequent negative behaviors during their interactions.

A MANOVA performed on the infant measures was significant, and univariate ANOVAs revealed that the vagal tone of the infants of depressed mothers was significantly lower at both 3 and 6 months. Growth measures also differentiated the infants of depressed mothers, including significantly shorter length and smaller head circumference at 3 months and significantly lower weight and smaller head circumference, as well as more pediatric complications, at 6 months.

Regression Analysis

Correlation and regression analyses were then conducted on the data of the depressed group to determine which maternal variables predicted continued depressive symptoms at 6 months. A correlation analysis of all maternal variables from the neonatal and 3-month periods and Beck Depression Inventory scores at 6 months yielded significant correlations only between BDI score at 6 months and a few 3-month variables, including greater right frontal EEG, less positive interaction behavior, and low vagal tone, as well as elevated norepinephrine, serotonin, and cortisol levels at the neonatal stage. These potential "predictor" variables were entered into a stepwise regression analysis with 6-month BDI scores (continuing depression) as the outcome measure and earlier BDI scores (neonatal and 3 month) as covariates. The results of the multiple regression revealed that the predictor variables entered the equation in the following order: (1) right frontal EEC accounted for 31% of the variance (p [less than] .001); (2) el evated serotonin (5HIAA) accounted for an additional 13% of the variance (p [less than] .01); and (3) elevated cortisol, less positive interaction behavior, elevated norepinephrine, and low vagal tone added 7% to the variance (p [less than] .05). Together, these explained 51% of the variance in continuing depressive symptoms (6-month BDI scores greater than 12). A discriminant function analysis revealed that 73% of the cases were correctly classified by this group of variables: 71% classified as having depressive symptoms had depressive symptoms, and 75% of the mothers without symptoms were correctly classified as not having symptoms (overall p [less than].01).

STUDY 2

A review of the literature revealed that few studies have investigated interventions with depressed mothers. Typically, these interventions involved brief interaction coaching or mood induction (Malphurs et al., 1996; O'Hara et al., 1990; Pickens & Field, 1995). These are not likely to have effects that are as long term as would be the case with a more comprehensive intervention, particularly for low-income adolescent mothers.

In Study 2, a second sample was recruited during the mothers' hospital stay after childbirth. Mothers with and without depressive symptoms were assessed over the first three months. At three months, mothers who were at risk for continuing depressive symptoms were identified using the predictor variables from Study 1. They were then randomly assigned to a comprehensive intervention program or to a control group. At 6 and 12 months, these groups were compared with each other and with the nondepressed group.

METHOD

Sample

A sample similar to that of Study 1 was recruited during the neonatal period (n = 160 mothers with depressive symptoms and n = 100 mothers without symptoms). The same neonatal and 3-month assessments were conducted (see Study 1), and dyads whose values were above the median (or below, depending on the negative direction) on the predictor variables (derived from the regression analysis in Study 1) were randomly assigned to an intervention group or to a control group. The demographics of the two groups are shown in Table 7. As can be seen by comparing Tables 1 and 7, the demographics of the Study 1 and Study 2 samples did not differ.

Assessments

The neonatal and 3-month assessments included the same measures elaborated in Study 1. Longer-term outcomes of the intervention were of interest; thus, 12-month assessments were also conducted. Further, infant development was evaluated at 12 months using the Bayley Scales of Infant Development (Bayley, 1969) and the Early Social Communication Scales (Seibert & Hogan, 1982).

The Bayley Scales of Infant Development include a Mental scale and a Motor scale. These standardized tests of infant development measure a variety of sensorimotor and temperament functions. The Early Social Communication Scales were developed to measure interactive behaviors of older infants. The tester presents a series of toys and engages the child in interactive games. Interactive behaviors are categorized by developmental level, from simple to symbolic, and by communicative function, from self-attention to attention to someone else.

Intervention

Brief interventions, such as interaction coaching, have been tried with depressed mothers (Maphurs, Larrain, Field, Pickens, Pelaez-Nogueras, Yando, & Bendell, 1996; Pickens & Field, 1995). However, comprehensive programs, such as the one used in this study, have rarely been tried with depressed mother-infant dyads.

It was thought that an educational/vocational experience might attenuate the mothers' depression. Further, substitute caregiving would both give the mothers the opportunity to pursue educational or vocational interests and reduce the infants' exposure to their mothers' depression. Thus, a 3-month social/educational/vocational rehab program was arranged, and free day care in a model infant nursery in a local public vocational high school was provided. Although this kind of intervention had not been tried with depressed mothers, a similar intensive intervention program was successfully used with nondepressed teenage mothers (Field, Widmayer, Ignatoff, & Stronger, 1982). Mood induction activities were also part of the program, including music mood induction, relaxation therapy, massage therapy, and infant massage. In addition, interaction coaching was provided to help the mothers and infants improve their interactions.

Vocational high school. The mothers attended a vocational high school in the morning and participated in social and vocational rehab activities and aerobics in the afternoon, while their infants received all-day care. In addition, the mothers spent approximately one hour per day in the nursery, helping to take care of their infants. The mothers also received relaxation therapy, massage therapy, and music mood induction, and their infants received massage therapy. The mothers and infants together were given interaction coaching. Attendance averaged 80% across mothers, with very low variability.

Mood induction. The mood induction intervention targeted two commonly noted parenting problems of depressed mothers: depressed mood state and insensitivity to their infants' emotional cues. Depressed mood state is a problem because the infants do not receive feedback for their behaviors (contingent responsivity), and the depressed behavior also serves as a model for the infants' own behavior (Field, 1986). The mother's insensitivity to her infant's emotional cues probably derives from her being preoccupied (Ingram & Smith, 1984). The interventions designed here focused on altering the mothers' mood state, thereby increasing sensitivity to their infants' cues.

Music mood induction was reported to be effective with 100% of subjects in a review of mood induction techniques (Clark, 1983). In contrast, mood induction cards were effective with only 68%, and greater effects were reported for despondency and happiness using the music mood induction. In the intervention implemented here, the mothers listened to a half hour of different rock music selections for four weeks (two sessions per week).

Relaxation therapy is another effective technique for reducing depression (Agras, 1983). In one study, relaxation therapy was as effective as psychotherapy and pharmacotherapy in reducing depression-related anxiety (McClean & Hakistian, 1979). In another study, only the relaxation therapy group continued to show reduced anxiety levels (Reynolds & Coats, 1986). Relaxation therapy has also been noted to slow heart rate, alter EEG, and diminish cortisol production (Cooper, Jaffe, Lamprey, Botha, Shives, Baker, & Seftel, 1985; Platania-Solazzo, Field, Blank, Seligman, Kuhn, Schanberg, & Saab, 1992). Here, progressive muscle relaxation and visual imagery were used. Tapes were prepared by a relaxation therapist; gentle Far Eastern music was combined with descriptions of positive visual images and instructions on progressive muscle relaxation. These sessions were held two times per week for four weeks.

Massage therapy has been found to reduce depression and anxiety, as well as levels of stress hormones (norepinephrine and cortisol), in hospitalized, depressed adolescents (Field, Morrow, Valdeon, Larson, Kuhn, & Schanberg, 1992). In addition, their sleep patterns and clinical condition improved. In the intervention implemented in the present study, the mothers were given a 20-minute Swedish massage two times per week for four weeks.

Infant massage has become increasingly popular because it not only relaxes the infant and alleviates difficult problems for parents, such as colic and disturbed infant sleep, but also because it enhances parental sensitivity to infant cues and the parent-child relationship (Auckett, 1981; Field, Grizzle, Scafidi, Abrams, & Richardson, 1996; Schneider, 1982). For the 15-minute massage sessions, the infant was placed in a prone position for 5 minutes of stroking, followed by 5 minutes in a supine position for cycling of the limbs (flexion and extension), then 5 minutes in the prone position during which time the infant was stroked again (Field et al., 1986). The infants received massage two times per week for four weeks.

Interaction coaching. Field (1977) found that a technique called maternal imitation of infant behavior contributed to mothers becoming more sensitive to their infants' cues of being underaroused or overaroused. The infants, in turn, became more attentive and responsive than they typically were during spontaneous interactions. In another attention-getting technique, mothers engaged in more game playing and their infants' affect improved (Clark & Seifer, 1983; Pickens & Field, 1995). Thus, for the interaction coaching in this study, the mothers were given instructions on imitation (they were asked to imitate their infants' behaviors) and attention getting (they were asked to keep their infants' attention). The intent was to help them be both more sensitive to their infants' cues (in the case of imitating their infants) and more active (in the case of getting their infants' attention).

RESULTS

Neonatal and Three Months

MANOVAs and ANOVAs comparing the depressed and nondepressed groups on the birth measures and neonatal assessments (Table 8) and 3-month variables (Table 9) led to results similar to those of Study 1, with the depressed group having inferior scores. At 3 months, mothers in the depressed group who had low interaction scores (below the median), greater right frontal EEG, and low vagal tone (below the median) at 3 months and elevated norepinephrine, serotonin, and cortisol levels (above the median) at the neonatal stage were randomly assigned to an intervention group or a control group. These groups were then compared with each other and with the nondepressed group at 6 and 12 months.

Six Months

Maternal depression scores. An ANOVA and post hoc Bonferroni t test indicated that, following the 3-month-long intervention (from 3 to 6 months), the intervention group had a lower mean BDI score than did the control group. However, the score was still higher than that for the nondepressed group (see Table 10). Analyses revealed a similar pattern for the incidence of DISC dysthymia. No significant differences across groups were noted for background stress.

Interaction ratings. MANOVAs on the interaction ratings and percentage of time positive, neutral, and negative behaviors occurred were significant, as were most of the ANOVAs. Post hoc Bonferroni t tests indicated that the mothers and infants in the control group had lower interaction ratings and engaged in more frequent negative and less frequent positive behaviors than did the mothers and infants in the intervention group, who had scores that were comparable to those of the nondepressed group.

Infant measures. A MANOVA on the infant measures at 6-month outcome was significant. Individual ANOVAs and post hoc Bonferroni t tests indicated that the vagal tone of the intervention group approximated that of the nondepressed group and was significantly higher than that of the control group. Infant weight for the intervention group was higher than that for the control group, but lower than that for the nondepressed group. In terms of pediatric complications, the score for the intervention group was also more optimal than that for the control group and not significantly different from the score for the nondepressed group.

Biochemical variables. A MANOVA on the mothers' biochemical variables at 6 months was significant. ANOVAs and post hoc Bonferroni t tests indicated that the intervention group had significantly lower epinephrine, dopamine, serotonin, and cortisol levels than did the control group, approximating the levels of the nondepressed group (see Table 11). This suggests that their biochemistry had normalized.

The MANOVA on the infants' biochemical variables was also significant. ANOVAs and post hoc Bonferroni t tests indicated that the intervention group's norepinephrine, epinephrine, dopamine, and cortisol levels were higher than those of the control group. However, the biochemical levels of the intervention group approximated those of the control group, again suggesting normalization.

Twelve Months

Maternal depression. An ANOVA and post hoc Bonferroni t tests revealed that the intervention group continued to experience significantly greater depressive symptoms (i.e., higher BDI score) than did the nondepressed group (see Table 12). However, the BDI score for the intervention group was significantly lower than that of the control group (see Table 12).

Infant development. A MANOVA on the infant development scores was significant. Subsequent ANOVAs and post hoc Bonferroni t tests indicated that the intervention group scored higher on the Bayley Mental and Motor scales than did the control group, with scores approximating those of the nondepressed group. In addition, on the Early Social Communication Scales, the intervention group had higher Responding and Initiating scores than did the control group, and a higher Responding score than even the nondepressed group.

Physical measures. A MANOVA on the infants' physical measures was significant. Post hoc ANOVAs and Bonferroni t tests revealed that the intervention group had greater length and fewer pediatric complications than did the control group, with values approximating those of the nondepressed group.

DISCUSSION

Most of the mothers with depressive symptoms in Study 1 continued to have symptoms over the first 6 months (about 13% went into remission). Continued symptoms over the first 6 months has been associated with developmental and growth delays for infants starting at around one year (Field, 1992). This highlights the importance of identifying variables that predict which mothers will experience continuing symptoms. In addition to identifying predictor variables, Study 1 documented a "dysregulation" profile, starting as early as the neonatal period, for infants of symptomatic mothers. The data from Study 2 showed the effectiveness of an intervention program for these adolescent mothers and their infants.

The depressed mothers' negative interactional behaviors, as well as their continuing symptoms, across the first 6 months were consistent with findings from earlier research (see Field, 1992, 1995, for reviews of studies conducted by Cohn et al., 1990; Murray, 1992; and O'Hara, Rehm, & Campbell, 1983). In addition, several physiological and biochemical differences were noted; for example, the symptomatic mothers had elevated levels of catecholamines (norepinephrine and epinephrine), which often accompany depression, along with elevated dopamine and serotonin metabolites (5-HIAA). Also seen in adults with symptoms of chronic depression, 90% of the symptomatic mothers had greater right frontal EEG activation (see Henriques & Davidson, 1990). The regression equation revealed that these variables (right frontal EEG, lower vagal tone, and less positive interaction behavior at 3 months, and elevated norepinephrine, serotonin, and cortisol at the neonatal period) were responsible for 51% of the variance in the mothe rs' continuing symptoms at 6 months. Greater right frontal EEG alone accounted for 31% of the variance, with elevated serotonin adding another 13%. The discriminant function analysis indicated that the mothers' continuing depressive symptoms were predicted by these variables with 73% accuracy. These variables were then used to identify mothers who need early intervention.

The dysregulation profile of the infants of symptomatic mothers might also be useful for targeting intervention dyads. The infants had inferior Brazelton scores, not unlike those reported in earlier studies (Abrams et al., 1995; Zuckerman & Beardslee, 1987). Specifically, their inferior habituation and orientation scores suggested high stimulus thresholds and less ability to attend to stimulation, and their Lester "depression" scores were consistent with their frequently noted flat affect and low activity levels. Their immature sleep patterns (e.g., greater indeterminate sleep and more jerks/startles) were also consistent with physiological dysregulation, as were their elevated norepinephrine, epinephrine, and dopamine levels (which mirrored their mothers' elevated levels). The absence of cortisol elevations among both the mothers and infants was unexpected, because elevated levels have been noted in previous samples (Field, 1995). However, cortisol may "reequilibrize" with continuing depression, and more se vere depression ratings may be associated with lower cortisol levels, as if cortisol conforms to an inverted U function. At 3 months, greater right frontal EEG activation and lower vagal tone suggested both autonomic and central nervous system involvement.

The less positive interaction behaviors were not surprising, because they have been reported by many investigators, although not typically as early as the neonatal period, as found here. While infant growth delays by one year have been reported in a previous sample (Field, 1992), the delays noted in the present study were seen as early as 3 months and involved head circumference and length. At 6 months, growth delays were noted for head circumference and weight.

Such growth delays highlight the need for early intervention. Because the intervention in this study was multifaceted (the mothers participated in a social/educational/vocational rehab program and several mood-induction interventions, with day care provided for their infants), it is not possible to determine which components were most important, or whether all were necessary for successful outcomes.

Interestingly, despite the intervention and its several positive outcomes, participating mothers continued to have elevated Beck Depression Inventory scores (although they were not as high as those for the mothers in the control group). Nonetheless, mother and infant interaction behaviors became more positive and, for both, several physiological and biochemical measures normalized (i.e., became more like the nondepressed group). Normalization entailed a decrease in epinephrine, dopamine, and serotonin for the mothers and, for the infants, an increase in norepinephrine, epinephrine, dopamine, and cortisol. Although the meaning of the increased levels for the infants is unclear, given the lack of norms for neurotransmitters in infants, the fact that their levels approximated those of the infants of nondepressed mothers is encouraging. In addition, the increase was comparable to that noted in the catecholamine levels of preterm infants who received early intervention (Kuhn et al., 1991).

By 12 months (6 months after the end of the 3-month-long intervention), the BDI score of the mothers in the intervention group remained higher than that of the mothers in the nondepressed group, but their infants were not showing the growth or developmental delays noted in an earlier study (Field, 1992). Their infants, compared with those in the control group, had fewer pediatric complications, greater length, and superior mental, motor, and social communication scores, with values approximating those of the nondepressed group. This suggests that the positive effects of the intervention were persisting for the infants. Further, despite the fact that the BDI score remained higher than normal for the mothers in the intervention group, their behavior had improved and their biochemistry suggested less stress and depression.

Assessing these mothers during pregnancy and studying the buffering effects of fathers may help disentangle prenatal and genetic factors. In addition, further studies are needed on the dysregulation noted in the infants. Finally, because the intervention was so comprehensive, it is not clear which components are critical to achieve these effects. Nonetheless, the data suggest that, at the very least, adolescent mothers who are not receiving treatment and who continue to experience depressive symptoms can be identified, and those mothers and their infants can be offered a fairly cost-effective intervention that significantly attenuates the infants' delays in growth and development.

The authors thank the infants and parents who participated in this study and the research assistants who helped with data collection. This research was supported by an NIMH Research Scientist Award (#MH00331) and an NIMH Research Grant (#MH46586) to Tiffany Field.

Tiffany Field, Margarita Prodromidis, and Julie Malphurs, Touch Research Institutes, University of Miami School of Medicine and Nova Southeastern University.

Jeffrey Pickens, James Madison University.

Nathan Fox, University of Maryland.

Debra Bendell, Fremont Kaiser Permanente.

Regina Yando, Harvard Medical School.

Saul Schanberg and Cynthia Kuhn, Duke University Medical School.

Reprint requests to Tiffany Field, Touch Research Institutes, University of Miami School of Medicine, Department of Pediatrics (D-820), P.O. Box 016820, Miami, Florida 33101.

REFERENCES

Abrams, S. M., Field, T., Scafidi, F., & Prodromidis, M. (1995). Newborns of depressed mothers. Infant Mental Health Journal, 16, 231-237.

Agras, W. S. (1983). Medical uses of relaxation training. Paper presented at Grand Rounds, University of Wisconsin, Department of Psychiatry, Madison, WI.

Auckett, A. D. (1981). Baby massage. New York: Newmarket Press. Bayley, N. (1969). Bayley Scales of Infant Development. New York: The Psychological Corporation.

Beardslee, W. R., Bemporad, J., Keller, M. B., & Kierman, G. L. (1983). Children of parents with major affective disorder: A review. American Journal of Psychiatiy, 140, 825-832.

Beck, A. T. (1976). Cognitive therapy and emotional disorders. New York: International Universities Press.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. E., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Bloom, J. L., & Anneveldt, M. (1982). An electrode cap tested. Electroencephalography and Clinical Neurophysiology, 54, 591-594.

Brazelton, T. B. (1973). Neonatal Behavioral Assessment Scale. London: Spastics International Medical Publications.

Campbell, S. B., Cohn, J. F., & Meyers, T. (1995). Depression in first-time mothers: Mother-infant interaction and depression chronicity. Developmental Psychology, 31, 349-357.

Clark, D. M. (1983). On the induction of depressed mood in the laboratory: Evaluation and comparison of the Velten and musical procedures. Advances in Behavior Research and Therapy, 5, 27-49.

Clark, G. N., & Seifer, R. (1983). Facilitating mother-infant communication: A treatment model for high-risk and developmentally delayed infants. Infant Mental Health Journal, 4, 67-82.

Cohen, J. (1990). Things I have learned (so far). American Psychologist, 45, 1304-1312.

Cohn, J. F., Campbell, S. B., Matias, R., & Hopkins, J. (1990). Face-to-face interactions of postpartum depressed and nondepressed mother-infant pairs at two months. Developmental Psychology, 26, 15-23.

Cooper, R., Jaffe, B. I., Lamprey, J. M., Botha, L. Shives, R., Baker, S., & Seftel, H. C. (1985). Hormonal and biochemical responses to TM. Postgraduate Medical Journal, 61, 301-304.

Costello, E. J., Edelbrock, C. S., & Costello, A. J. (1985). Validity of the NIMH Diagnostic Interview Schedule for Children: A comparison between psychiatric and pediatric referrals. Journal of Abnormal Child Psychology, 13, 579-595.

Field, T. (1977). Effects of early separation, interactive deficits and experimental manipulations on infant-mother face-to-face interaction. Child Development, 48, 763-771.

Field, T. (1980). Interactions of preterm and term infants with their lower-and middle-class teenage and adult mothers. In T. Field, S. Goldberg, D. Stern, & A. Sostek (Eds.), High-risk infants and children: Adult and peer interactions. New York: Academic Press.

Field, T. (1984). Early interactions between infants and their postpartum depressed mothers. Infant Behavior and Development, 7, 517-522.

Field, T. (1986). Interventions for premature infants. Journal of Pediatrics, 109, 183-191.

Field T. (1992). Infants of depressed mothers. Development and Psychopathology, 4, 49-66.

Field, T. (1995). Infants of depressed mothers. Infant Behavior and Development, 18, 1-13.

Field, T., Fox. N., Pickens, J., Nawrocki, T., & Soutullo, D. (1995). Right frontal EEG activation in 3- to 6-month-old infants of "depressed" mothers. Developmental Psychology, 31, 358-363.

Field, T., Grizzle, N., Scafidi, F., Abrams, S., & Richardson, S. (1996). Massage therapy for infants of depressed mothers. Infant Behavior and Development, 19, 109-114.

Field, T., Healy, B., Goldstein, S., & Guthertz, M. (1990). Behavior state matching and synchrony in mother-infant interactions of nondepressed versus depressed dyads. Developmental Psychology, 26, 7-14.

Field, T., Healy, B., Goldstein, S., Perry, S., Bendell, D., Schanberg, S., Zimmerman, E. A., & Kuhn, C. (1988). Infants of depressed mothers show "depressed" behavior even with nondepressed adults. Child Development, 59, 1569-1579.

Field, T., Healy, B., & LeBlanc, W. (1989). Sharing and synchrony of behavior states and heart rate in "nondepressed" versus "depressed" mother-infant interactions. Infant Behavior and Development, 12, 357-376.

Field, T., Morrow, C., Valdeon, C., Larson, S., Kuhn, C., & Schanberg, S. (1992). Massage therapy reduces anxiety in child and adolescent psychiatric patients. Journal of the American Academy of Child and Adolescent Psychiatry, 31(1), 125-131.

Field, T., Pickens, J., Fox, N., Nawrocki, T., & Gonzalez, J. (1995). Vagal tone in infants of depressed mothers. Development and Psychopathology, 7, 227-231.

Field, T., Schanberg, S. M., Scafidi, F., Bauer, C. R., Vega-Lahr, N., Garcia, R., Nystrom, J., & Kuhn, C. M. (1986). Tactile/kinesthetic stimulation effects on preterm neonates. Pediatrics, 77, 654-658.

Field, T., Widmayer, S., Ignatoff, E., & Stronger, S. (1982). Developmental effects of an intervention for preterm infants of teenage mothers. Infant Mental Health Journal, 3, 11-18.

Forsyth, B., & Caney, P. (1985). Long-term implications of problems of feeding and behavior in early infancy: A 3 1/2-year follow-up. Paper presented at a meeting of the Ambulatory Pediatric Association, Washington, DC.

Henriques, J. B., & Davidson, R. J. (1990). Regional brain electrical asymmetries discriminate between previously depressed and healthy control subjects. Journal of Abnormal Psychology, 99, 22-31.

Ingram, R. E., & Smith, T. W. (1984). Depression and internal versus external focus of attention. Cognitive Therapy and Research, 8, 139-152.

Kuhn, C., Schanberg, S., Field, T., Symanski, R., Zimmerman, E., Scafidi, F., & Roberts, J. (1991). Tactile/kinesthetic stimulation effects on sympathetic and adrenocortical function in preterm infants. Journal of Pediatrics, 119, 434-440.

Leaf, P. J., Bruce, M. L., Tischler, G. L., Freeman, D. H., Weisman, M. M., & Myers, J. K. (1988). Factors affecting the utilization of specialty and general medical mental health services. Medical Care, 26, 9-26.

Lester, B. M., Hoffman, J., & Brazelton, T. B. (1985). The rhythmic structure of mother-infant interaction in term and preterm infants. Child Development, 56, 15-27.

Littman, D., & Parmelee, A. (1978). Medical correlates of infant development. Pediatrics, 61, 470-482.

Lyons-Ruth, K., Zoll, D., Connell, D., & Grunebaum, H. (1986). The depressed mother and her one-year-old infant: Environment, interaction, attachment, and infant development. In E. Tronick & T. Field (Eds.), Maternal depression and infant disturbance. San Francisco: Jossey-Bass.

Malphurs, J., Larrain, C. M., Field, T., Pickens, J., Pelaez-Nogueras, M., Yando, R., & Bendell, D. (1996). Altering withdrawn and intrusive interaction behaviors of depressed mothers. Infant Mental Health Journal, 17, 152-160.

McClean, P. D., & Hakistian, R. (1979). Clinical depression: Comparative efficacy of outpatient treatments. Journal of Clinical and Consulting Psychology, 47, 818-836.

Murray, L. (1992). The impact of postnatal depression on infant development. Journal of Child Psychology and Psychiatry, 33, 543-561.

National Institute on Drug Abuse. (1987). The adolescent assessment/referral system. Available from the U.S. Department of Health and Human Services, Alcohol, Drug Abuse and Mental Health Administration.

O'Hara, M., Rehm, L. P., & Campbell, S. B. (1983). Postpartum depression: The role of social network and life stress variables. Journal of Nervous and Mental Disease, 171, 336-341.

O'Hara, M. W., Zekoski, E. M., Philipps, L. H., & Wright, E. J. (1990). A controlled, prospective study of postpartum mood disorders: Comparison of childbearing and non-childbearing women. Journal of Abnormal Psychology, 99, 3-15.

Orvaschel, H. (1983). Maternal depression and child dysfunction: Children at risk. In B. Lahey & A. Kazdin (Eds.), Advances in clinical child psychology. New York: Plenum.

Perrin, E., West, P., & Culley, B. (1989). Is my child normal yet? Correlates of vulnerability. Pediatrics, 83(3), 355-363.

Pickens, J., & Field, T. (1995). Facial expressions and vagal tone in infants of depressed and nondepressed mothers. Early Development and Parenting, 4, 83-89.

Platania-Solazzo, A., Field, T., Blank, J., Seligman, F., Kuhn, C., Schanberg, S., & Saab, P. (1992). Relaxation therapy reduces anxiety in child/adolescent psychiatry patients. Acta Paedopsychiatrica, 55, 115-120.

Porges, S. W. (1985). Method and apparatus for evaluating rhythmic oscillations in aperiodic physiological response systems. United States Patent No. 4, 510, 944 (April 16, 1985).

Radke-Yarrow, M., Cummings, E. M., Kuczynski, L., & Chapman, M. (1985). Patterns of attachment in two- and three-year-olds in normal families and families with parental depression. Child Development, 56, 886-893.

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401.

Reynolds, W., & Coats, K. (1986). A comparison of cognitive-behavioral therapy and relaxation training for treatment of depression in adolescents. Journal of Clinical and Consulting Psychology, 54, 653-660.

Sameroff, A. J., & Seifer, R. (1983). Familial risk and child competence. Child Development, 54, 1254-1268.

Schneider, V. (1982). Infant massage. Toronto: Bantam Books.

Seibert, J., & Hogan, A. (1982). Procedures manual for the Early Social Communication Scales. Unpublished manuscript, University of Miami.

Seifer, R. (1995). Perils and pitfalls of high-risk research. Developmental Psychology, 31, 420-424.

Thoman, E. B. (1975). Early development of sleeping behaviors in infants. In N. T. Ellis (Ed.), Behavior and development in infancy: Human and animal studies. New York: John Wiley and Sons.

Zuckerman, B. S., & Beardslee, W. R. (1987). Maternal depression: A concern for pediatricians. Pediatrics, 79, 110-117.