The relationship between maternal HIV status and child depressive symptoms: Do maternal depressive symptoms play a role?

BEHAVIORTHERAPY29, 409-422, 1998

The Relationship Between Maternal HIV Status and Child Depressive Symptoms: Do Maternal Depressive Symptoms Play a Role? HEATHER BIGGAR REX FOREHAND

University of Georgia FAMILY HEALTH PROJECT RESEARCH GROUP This study examined whether maternal depressive symptoms serve as a mediator, moderator, or both, between maternal HIV status (absence vs. presence of HIV) and child depressive symptoms. Participants were 224 noninfected children, ages 6 to 11, and their mothers, 38% of whom were HIV-infected. Initial analyses indicated that HIV-infected mothers and their children reported more depressive symptoms than noninfected mothers and their children. The primary analyses suggested that maternal depressive symptoms play a moderating, but not a mediating, role as the direction of the relationship between maternal depressive symptoms and child depressive symptoms differed between HIV-infected and noninfected groups. Explanations for the findings are offered and implications for prevention and intervention programs are considered.

The prevalence of HIV/AIDS in women is and will continue to be a societal problem of major proportion. The Centers for Disease Control (CDC, 1996) reported that, as of December 1996, 85,500 women had been diagnosed with AIDS. This number represents a dramatic increase (from 6% to 15%) over the past decade in the percent of total AIDS cases represented by women (CDC). It is expected that the proportion of female AIDS cases will continue to increase (Cameron, 1994). Approximately 55% of women with AIDS Heather Biggar, Department of Psychology; Rex Forehand, Institute for Behavioral Research. Members of the Family Health Project Research Group include Rex Forehand (PI, University of Georgia), Lisa Armistead (Co-PI, now at Georgia State University), Edward Morse (Co-PI, Tulane University), and Patricia Simon (Co-PI, LSU Medical Center). Appreciation is expressed to the Centers for Disease Control and Prevention and the University of Georgia's Institute for Behavioral Research for support. In particular, the support of Dr. Janet Moore and Dr. Kim Miller of the CDC is acknowledged. Correspondence concerning this article should be addressed to Rex Forehand, IBR, Barrow Hall, University of Georgia, Athens, GA 30602. 409 0005-7894/98/0409-042251.00/0 Copyright 1998 by Associationfor Advancementof BehaviorTherapy All rights of reproduction in any form reserved.

410

BIGGAR ET AL.

are African American (CDC), and the majority of these women hail from economically poor urban communities. Women who are African American and living in poverty experience more acute and chronic stressors than most other women, including illness and imprisonment of loved ones, changes in income, and unique parenting stresses (Belle, 1984; Forehand & Kotchick, 1996; McLanahan, 1983). These stressors are associated with depressive symptoms (see Barbee, 1992, and Brown, 1990, for reviews). Thus, many HIV-infected women must cope not only with seriously compromised health, but also with the stress that accompanies living in poverty and the resulting depressive symptoms. Not only are the vast majority of HIV-infected women in their childbearing years (CDC, 1996), but over one-half of them currently have children (Chu & Diaz, 1993; Michaels & Levine, 1992; Schable et al., 1995). The impact of parental HIV/AIDS on children may range from rejection and discrimination by others to compromised parenting and, eventually, to orphanhood. In fact, it has been estimated that by the year 2000, as many as 125,000 children in the United States will have been affected and orphaned by maternal HIV (Michaels & Levine). Despite this staggering figure, there is a lack of empirical research focusing on these women and their children (Armistead & Forehand, 1995; Cameron, 1994). Because they often are not infected themselves, the children of infected women are often overlooked by health care workers and researchers (Taylor-Brown, 1991). Consequently, limited knowledge is available regarding the social, psychological, economic, and environmental factors affecting women who are HIV-infected and their noninfected children. One consequence that does seem clear is that, as a result of the physical and psychological sequelae of HIV infection and the stressors present in their communities of residence, children of women who are HIVinfected are likely to be at high risk for emotional problems (Armistead & Forehand; Levine, 1995). For example, two recent studies indicated that child depressive symptoms are more prevalent in children of HIV-infected mothers than in children of noninfected mothers (Forehand et al., 1998; Forsyth, Damour, Naglin, & Adnopoz, 1996). In research with noninfected samples, there is a compelling literature indicating a relationship between maternal depressive symptoms and child depressive symptoms (for reviews, see Beardslee, Bemporad, Keller, & Klerman, 1983; Downey & Coyne, 1990; Forehand, McCombs, & Brody, 1987). However, this literature is based on research primarily with middle-class Caucasian samples. African American women and their children have been ignored, as have women who are HIV-infected. In the current study, we test whether depressive symptoms of African American mothers mediate, moderate, or serve both functions in the relationship between their HIV status and their child's depressive symptoms. According to Baron and Kenny's (1986) definition of mediation, a mediator is a variable that "represents the generative mechanism through which the focal independent variable is able to influence the dependent variable of

MATERNALHIV STATUSAND CHILDDEPRESSIVESYMPTOMS

411

interest" (p. 1173). As applied to the current study, we suggest that maternal depressive symptoms may be a mediating variable through which maternal HIV infection influences child depressive symptoms. However, an absence of statistical mediation does not mean that maternal depressive symptoms do not play a role in the relationship between maternal HIV infection and child depressive symptoms; an alternative explanation is that maternal depressive symptoms serve as a moderator. A moderator is a qualitative or quantitative variable that "affects the direction and/or strength of the relation between an independent or predictor variable and a dependent or criterion variable" (Baron & Kenny, p. 1174). In other words, a moderator functions as a second independent variable that interacts with the focal independent variable to affect its relationship with the dependent variable. We examined the potential role of maternal depressive symptoms as both a mediator and as a moderator.

Method Overview The data presented in the current investigation are from The Family Health Project, an ongoing, longitudinal investigation in New Orleans, Louisiana, which was designed to investigate the psychological and sociological correlates of HIV infection in inner-city, African American families. Specifically, the project examined two groups of women: one group of individuals identified by a physician as HIV-infected; a second group of individuals selfidentified as not HIV-infected (subsequently referred to as noninfected). 1 All of the women had at least one noninfected child who was between the ages of 6 and 11 at the beginning of the project and who served as the target child for the study. The present investigation is limited to data from the first assessment. (See Family Health Project Research Group, in press, for further details.)

Participants Participants included two groups. The first group consisted of 85 women who were HIV-infected and one of their noninfected, 6- to ll-year-old children. The second group included 139 noninfected women and one of their noninfected, 6- to ll-year-old children. All participants were African American and were recruited from the inner-city area of New Orleans. Demographic characteristics of the two groups and tests of differences are presented in Table 1. With the exception of mothers' age, the two groups did not differ on any of the demographic variables. Participants resided in a high-crime, inner-city environment. Approximately two-thirds of the sample lived in government housing developments. Approximately 50% of the women were the only adult in the household. i In research with similar populations, the percentageof women who self-identifyas noninfected, but who are actually HIV-infected,is less than 1% (J. Moore, personal communication, February 22, 1996).

412

BIGGAR ET AL.

TABLE 1 SAMPLE CHARACTERISTICS OF MATERNAL HIV-INFECI'ED AND NONINFECTED GROUPS

Groups NonN°ninfectedb HIV infected Chi M SD t Test % % Square

Groups HIVa Characteristics

M

SD

Mother Age (years) 31.25 5 . 2 5 33.08 5.67 Education % Less than high school % High school or GED % High school + vocational training % Some college Monthly income $629 419 $700 490 % Married Child Age (years) % Female

8.68

1.71

8.69

1.66

2.44* 4.43 35 40

48 30

8 17

10 12

13

19

1.57

59

51

1.27

1.15

.04

a n = 85. b n = 139. * p < .05.

When other adults were present, they were most frequently the mother, sister, or male partner of the woman. Several additional characteristics of the HIV-infected sample deserve mention. The primary route of HIV infection was heterosexual transmission (80% of the sample). O f the infected mothers, 11% had an HIV-infected child (who was not the target child). The primary antiretroviral therapy being utilized to slow the spread of the virus was Zidovudine (used by 62% of the sample). In addition, 16% of the HIV-infected women reported receiving current treatment by a mental health professional, and 2% of these women reported that their target child was currently being treated by a mental health professional. Aside from maternal (18 to 45 years old) and child (biological child, 6 to 11 years old, residing with mother, noninfected) requirements for participation, one additional and important criterion was included for eligibility of HIV-infected mothers: They were required to have CD4 counts of less than 600 in order to participate. CD4 lymphocytes are immune cells that protect the body from infection and disease. HIV acts by infecting and killing those very cells whose function it is to protect the body, thus leaving the individual unable to protect herself. Thus, the fewer CD4 ceils an individual has, the more susceptible to infections and cancer she is (i.e., the more advanced is her illness). By requiring each woman to have a CD4 count of less than 600, we insured that some immunosuppression was occurring; however, we did

MATERNAL HIV STATUS AND CHILD DEPRESSIVE SYMPTOMS

413

not specify a particular stage of illness for participation. As of 1993, CDC specified that individuals having a CD4 count of less that 200 can be categorized as individuals with AIDS. Typically, however, clinical evidence (e.g., Kaposi's sarcoma) of infection-regardless of bloodstream analysesalso can determine diagnosis of AIDS. Between HIV infection without symptoms and HIV infection cum AIDS exists a period (labeled as symptomatic) during which nonspecific symptoms (e.g., fever, swollen lymph nodes, skin rashes) emerge. Hence, the current HIV-infected sample was classified into three groups: asymptomatic (49%), symptomatic (23%), and AIDS (28%). The current sample was drawn from a larger sample of 100 HIV-infected women and their children and 149 noninfected women and their children, all of whom participated in the initial phase of the Family Health Project. Twenty-five mother-child dyads were excluded from the current analyses because child responses on the Child Depression Inventory were deemed invalid (e.g., child responded randomly or did not respond to items). This occurred primarily with younger children (ages 6 and 7).

Measures In order to utilize only measures that were culturally sensitive and otherwise appropriate for the target population, a number of steps were taken. First, based on the available literature and related theory, an outline of the constructs to be evaluated was created. This outline was revised and expanded based on a focus group conducted with African American women who were HIV-infected. Second, measures of the constructs were selected or created and pilot tested with a group of African American women who matched the demographic characteristics of the study sample and had children in the same age range. These women were not known to be HIV-infected. Third, two more focus groups, both consisting of HIV-infected, African American women residing in New Orleans, were conducted in order to further refine constructs and improve the sensitivity of the measures. Fourth, a meeting of selected members of the Family Health Project and a grant consultant with expertise in projects focusing on African American families was held in order to integrate information obtained during focus groups, piloting, and previous projects with African American families. Based on this meeting, measures were further revised and then piloted again with a number of African American women (not known to be HIV-infected) and their children. Finally, all measures were field tested with HIV-infected, African American women who resided in New Orleans, Louisiana. Demographic information and health status. Demographic information was obtained from mother and child report. For HIV-infected mothers, information regarding health status (e.g., CD4 counts, number of opportunistic diseases) for classification into CDC stages (asymptomatic, symptomatic, AIDS) was obtained via medical chart review. Brief Symptom Inventory (BSI). The BSI (Derogatis & Spencer, 1982) is a 53-item inventory that was developed as a global measure of psychological

414

B1GGAR ET AL.

symptomatology. Adequate reliability, with test-retest coefficients ranging from .68 to .91, and validity data have been presented (Derogatis, Rickels, & Rock, 1976; Kremer, Atkinson, & Ignelzi, 1981). For the purposes of the present study, only the Depression subscale was administered. The internal consistency (.85) and test-retest reliability (.84) of this scale have been shown to be adequate and to have adequate discriminant and convergent validity (see Derogatis & Spencer). Each item was rated on a 4-point Likert scale ranging from 0 (not at all) to 3 (extremely). This scale represented a modification of the standard BSI, on which individuals rate the items on a 5-point Likert scale. Additional modifications included minor word and format changes to increase simplicity of verbal administration and comprehensibility. A mean score of the Depression subscale, ranging from 0 to 3, was utilized in analyses. The alpha coefficient for the current sample was .84. Children's Depression Inventory (CDI). The CDI (Kovacs, 1981) is the most widely used self-report measure of children's depressive mood (Kazdin, 1981). The CDI consists of 27 items, each of which allows the child to select among alternatives on a 3-point scale reflecting degree of particular symptoms. Standardization data are available for children ranging in age from 7 to 17, and the scale has been utilized with children as young as age 6 (Politano, Nelson, Evans, Sorenson, & Zeman, 1986). Adequate reliability and validity data have been reported (Kazdin; Kazdin, French, Uris, EsveldtDawson, & Sherick, 1983). For example, test-retest reliability has been cited at .72, and the CDI correlates significantly with global ratings of depression and Diagnostic and Statistical Manual of Mental Disorders (DSM-III; American Psychiatric Association, 1980) diagnoses. The CDI was modified slightly for the purposes of the present study in that instructions and items were reworded for verbal administration (e.g., "I am sad many times" changed to "You are sad many times"). The alpha coefficient for the current sample was .79.

Procedures Participants in the HIV-infected group were recruited across 2 years from two sources: (a) the primary public HIV clinic in the city of New Orleans (93%); and (b) the private practices of physicians treating HIV-infected females (7%). Ninety-five percent of women invited to participate did participate. A stratified random sample of noninfected women and their children was drawn from five public schools serving the zip code areas covering the catchment areas from which the HIV-infected sample resided. The noninfected sample was stratified based on school attended, gender of child, and age of child, and was drawn in two equal waves spanning 2 school years. In each of the two waves, letters briefly describing the study and inviting participation were sent home to 30 African American mothers randomly selected by school personnel at each school. Ninety-four percent of the mothers receiving letters indicated an interest in participating. The first 15

MATERNAL HIV STATUS AND CHILD DEPRESSIVE SYMPTOMS

415

women at each school in each of the two waves of data collection to indicate an interest in participating constituted the sample. Once each woman and her child agreed to participate, a data collection session was scheduled. In order to hold constant the impact that the child's attendance at school may have had on the interaction patterns between mother and child, all sessions for both groups were conducted while the child's school was in session (i.e., not during summer or winter breaks). Each noninfected woman and her child was assessed at the child's school, while each HIV-infected woman and her child was assessed in a medical setting. Two assessments occurred, separated by a period from 2 days to 2 weeks. In the first assessment, conducted by medical sociologists and medical sociologist graduate students, sociodemographic data were collected. In the second assessment, conducted by clinical psychologists and clinical psychology graduate students, data pertaining to mother and child psychosocial adjustment, including the BSI and CDI, were collected. The mother and child were administered the measures separately in each assessment. All measures were administered orally, with pictorial cue cards depicting response categories utilized for the BSI and CDI to facilitate data collection. Participants received $50 per day in compensation for their time and effort at the completion of each assessment. Details on interviewer training are available in the report of the Family Health Project Research Group (in press). Results

Primary Analyses Table 2 presents means and standard deviations for the measures of child and mother depressive symptoms for the HIV-infected and noninfected groups. Both children and mothers in the infected group reported higher levels of depressive symptoms than those in the noninfected group. For the CDI, 13% and 4% of the children of infected and noninfected mothers, respectively, were in the clinical range [(score > 19; X 2 (1) = 6.94, p < .01)]. As

TABLE 2 MEAN SCORES FOR CHILD AND MOTHER DEPRESSIVE SYMPTOMS IN THE HIV-INFECrED AND NONINFECTED GROUPS HIV-infected Measure Child Depression Inventory Brief Symptom Inventory Depression Subscale

Noninfected

t Test a

M

SD

M

SD

(df = 222)

10.96

7.67

8.47

5.40

2.63*

.65

.66

.41

.54

2.87*

a p value set at .025 based on Bonferroni correction for conducting two t tests. * p < .05.

416

B1GGAR ET AL.

the Likert rating scale for the BSI was changed for the current project from a 5-point to a 4-point scale, it was not possible to determine the percent of mothers in the clinical range. Correlations were computed among demographic variables (mother's age, education, and marital status; child's age and gender) and child and maternal depressive symptoms. One significant correlation emerged: Child age was associated with the CDI (r = -.15, p < .05). As a result, child age was entered initially into all regression analyses predicting CDI to control for that effect. As maternal age differed significantly between the HIV-infected and noninfected groups, it was entered initially into all regression analyses. The first procedure examined maternal depressive symptoms as a mediator between maternal HIV status (absence vs. presence of HIV) and child depressive symptoms. According to Baron and Kenny (1986), three conditions are required to demonstrate mediation: (a) the independent variable (maternal HIV status) must relate to the mediator (maternal depressive symptoms); (b) the independent variable (maternal HIV status) must relate to the dependent variable (child depressive symptoms); and (c) the mediator (maternal depressive symptoms) must relate to the dependent variable (child depressive symptoms). If these three conditions are met, then in order to show a mediational effect, the strength of the relationship between the independent variable and the dependent variable must decrease when the mediator is added to the equation. The closer to zero this independent-dependent association is after adding the mediator, the stronger the mediation effect. Using the procedures outlined above, we utilized hierarchical regression analyses to test for mediation. The results of the regression analyses are presented numerically in Table 3 and presented pictorially in Figure 1. We first tested for a relationship between maternal HIV status and the "mediator;' maternal depressive symptoms (Step 1). A significant relationship was found (see top panel of Table 3). Next, we tested for a relationship between presence versus absence of HIV and child depressive symptoms (Step 2). This relationship also was significant (see middle panel of Table 3). However, the relationship between maternal depressive symptoms and child depressive symptoms (Step 3) was not significant (see bottom panel of Table 3), thereby discounting the hypothesis that maternal depressive symptoms acted as a mediator. We next moved to an examination of maternal depressive symptoms as a moderator. According to Baron and Kenny's (1986) model, an interaction of maternal HIV status and maternal depressive symptoms is required for demonstration of a moderation effect. Again using a hierarchical multiple regression analysis, demographic variables were entered in Block 1, followed by maternal HIV status and maternal depressive symptoms in Block 2, and the interaction of maternal HIV status and maternal depressive symptoms in Block 3. The results of the regression analysis are presented in Table 4. Of primary interest for moderation is that a significant effect did emerge for the interaction term.

MATERNAL HIV STATUS AND CHILD DEPRESSIVE SYMPTOMS

417

TABLE 3 HIERARCHICAL MULTIPLE REGRESSION ANALYSES CONDUCTED TO TEST FOR MEDIATION

Variable R2

Mediation Step 1 (Dependent Variable: Maternal BSI-Depression Subscale) Block 1 Maternal Age Block 2 Maternal HIV status

AR 2

B

SE

-.01

.00

.00 .04

.25*

Mediation Step 2 (Dependent Variable: Child CDI) Block 1 Maternal age Child age Block 2 Maternal HIV status

.08

.20*

.02

.06

Mediation Step 3 (Dependent Variable: Child CDI) Block 1 Maternal age Child age Block 2 Maternal BSI-Depression

-.01

.04

-.01 -.51

.01 .27

-.04 -.13

2.48*

.88

-.01 -.51

.01 .27

-.04 -.13

.34

.71

.03

.04 .19"

.02

.02

.00

* p < .01.

We examined the interaction by conducting a hierarchical regression analysis separately for the HIV-infected and noninfected groups to determine if maternal depressive symptoms related to child depressive symptoms similarly or differently in the two groups. The beta coefficient was not significant for either the noninfected group, F(1, 135) = 2.76 (beta coefficient = .14),

.19"

J Maternal HIV Status

.20",

Maternal Depressive Symptoms

.03,

Child Depressive Symptoms

* p < .01 FIG. 1. Pictorial representation of pathways among maternal HIV status, maternal depressive symptoms, and child depressive symptoms.

418

BIGGAR ET AL. TABLE 4

HIERARCHICAL MULTIPLE REGRESSION ANALYSES CONDUCTED TO TEST FOR MODERATION

Variable R2

Block 1 Maternal age

AR 2

Block 2 Maternal HIV status Maternal BSI-Depression

.06

Block 3 Maternal HIV status X BSI-Depression

.08

.05. * * p <

SE

.02

Child age

*p<

B

-.01 -.51

.01 .27

-.04 -.13

2.49** -.05

.90 .72

.19"* -.01

.04

.02 -2.99*

1.42

-.47*

.01.

or the infected group, F(1, 81) = 1.63 (beta coefficient = -.14), but was in the opposite direction for the two groups. To present pictorially the interaction, we performed a median split on maternal depressive symptoms for the HIV-infected sample and for the noninfected sample. The means for child depressive symptoms for the four groups resulting from the median splits are presented in Figure 2. Based on the regression analyses conducted separately for the infected and noninfected group, and the data presented in Figure 2,

12 O

E C¢3 r/l ra~

"1

10


9

Non-Infected

om

8

Low

High

Maternal D e p r e s s i v e S y m p t o m s FIG. 2. Pictorial representation of child depressive symptoms as a function of maternal HIV status and maternal depressive symptoms.

MATERNAL HIV STATUS AND CHILD DEPRESSIVE SYMPTOMS

419

the significant interaction appears to result from maternal HIV status influencing the direction of the relationship between maternal and child depressive symptoms.

Secondary Analyses To examine whether the level of depressive symptoms and the associations between maternal depressive symptoms and child depressive symptoms differed as a function of stage of maternal HIV infection, we subdivided the HIV-infected sample into those mothers who were asymptomatic and those who were symptomatic or had an AIDS diagnosis• (The symptomatic [n = 16] and AIDS diagnosis In = 21] groups were combined because of the small sample size in each category.) We compared these two groups on level of depressive symptoms by way of t tests; the mother-completed BSI-Depressive subscale (asymptomatic M -- .67, SD = .70; symptomatic/AIDS M = .63, SD --- .62), t (83) = .25, and the child-completed CDI (symptomatic M = 10.31, SD = 6.48; symptomatic/AIDS M = 11.81, SD = 9.01), t (83) -- .86, did not differ across the two groups. We next repeated the regression with maternal HIV status consisting of three groups (1 = noninfected; 2 = HIV-infected, asymptomatic; 3 = HIVinfected, symptomatic/AIDS). The results were similar to those reported earlier. That is, there was no evidence for mediation, but a significant HIV status by maternal depressive symptoms interaction emerged for child depressive symptoms, F(1, 218) = 534, p < .05 (beta coefficient = .14). Separate regressions were conducted for each of the three groups, resulting in a positive but nonsignificant relation between mammal and child depressive symptoms for the noninfected sample, F(1, 135) = 2.76 (beta coefficient = •14); a negative but nonsignificant relation between maternal and child depressive symptoms for the HIV-infected, asymptomatic sample, F(1, 44) -- .10 (beta coefficient = -.04); and again, a negative but nonsignificant relation for the HIV-infected, symptomatic/AIDS sample, F(1, 33) -- 1.15 (beta coefficient = -.19).

Discussion Our findings indicate that HIV-infected women report more depressive symptoms than noninfected women. Similarly, children of HIV-infected women report more depressive symptoms than children of noninfected women. In order to examine how maternal depressive symptomatology may link maternal HIV infection and child depressive symptoms, we examined its role as a mediator and moderator. Our findings suggest that maternal depressive symptoms do not serve as a mediator; that is, they do not serve as the generative mechanism through which maternal HIV status influences child depressive symptoms. In particular, we failed to find an association between maternal depressive symptoms and child depressive symptoms in this sample. In contrast, our findings sug-

420

BIGGAR ET AL.

gest that maternal depressive symptoms play a moderating role between maternal HIV status and child depressive symptoms. We believe that both the failure to find a mediating relationship between maternal depressive symptoms and child depressive symptoms for the overall sample and the support found for a moderating relationship originate from the same source: The relationship between maternal depressive symptoms and child depressive symptoms is opposite in the HIV-infected and noninfected groups. However, as the relationship between maternal depressive symptoms and child depressive symptoms was not significant in regressions performed separately with either the noninfected group or the infected group, it is not possible to further interpret the interaction. We conducted secondary analyses to determine if associations between maternal and child depressive symptoms would differ between the asymptomatic and symptomatic/AIDS diagnosis stages of infection. The two groups did not differ on level of depressive symptoms reported by mother or child, and the relationship between maternal and child depressive symptoms was similar in the two groups (i.e., in the negative direction but nonsignificant). These findings suggest that it is the diagnosis of HIV, not the physical symptoms associated with the disease, which account for the findings that emerged in this study. However, it is important to note that any conclusions based on the secondary analyses must be viewed as tentative because of the relatively small sample sizes in the HIV-infected/asymptomatic and the symptomatic/ AIDS diagnosis groups. There are several additional limitations of the present study that should be noted. First, the data are cross-sectional and, as a consequence, causeand-effect inferences cannot be made. Second, this study examined whether presence versus absence of one illness, HIV, related to maternal and child depressive symptoms. Thus, the findings should not be generalized to other illnesses, in part because of the unique stigmatization associated with HIV (Armistead & Forehand, 1995). Third, the noninfected mother-child dyads were assessed in a school setting, whereas the HIV-infected mothers and their children were assessed in a medical setting. The difference in the settings should be considered when interpreting the findings. Fourth, assessment of depressive symptoms by self-report measures alone is not sufficient for diagnosis of clinical depression. Fifth, the women in the control group selfidentified as not being HIV-infected. It is possible that some of these women were infected, which could have influenced the findings. Several clinical implications are evident from the findings. First, as suggested by the moderation effect, the direction of the association between maternal and child depressive symptoms differs between noninfected and infected samples. Thus, it is important not to draw conclusions about HIVinfected samples based on findings from noninfected samples. Second, our findings provide substantial evidence that maternal HIV infection is associated with both mother and child depressive symptoms, indicating that prevention and intervention efforts must consider both mother and child symp-

MATERNAL HIV STATUSAND CHILD DEPRESSIVE SYMPTOMS

421

tomatology. For example, as maternal depressive symptoms were not significantly related to child depressive symptoms in the maternal HIV-infected group, addressing only the mother's depressive symptoms likely will not be sufficient to change child depressive symptoms. Thus, prevention and intervention programs designed for families with maternal HIV infection will need to address the depressive symptoms of multiple family members.

References American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. Armistead, L., & Forehand, R. (1995). For whom the bell tolls: Parenting decisions and challenges faced by mothers who are HIV seropositive. Clinical Psychology: Science and Practice, 2, 239-250. Barbee, E. L. (1992). African American women and depression: A review and critique of the literature. Archives of Psychiatric Nursing, 6, 257-265. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychology research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Beardslee, W. R., Bemporad, J., Keller, M. B., & Klerman, G. L. (1983). Children of parents with major affective disorder: A review. American Journal of Psychiatry, 140, 825-832. Belle, D. (1984). Inequality and mental health: Low-income and minority women. In L. Walker (Ed.), Women and mental health policy (pp. 125-150). Thousand Oaks, CA: Sage. Brown, D. R. (1990). Depression among blacks. In D. S. Ruiz (Ed.), Handbook of mental health and mental disorder among Black Americans (pp. 71-93). New York: Greenwood Press. Cameron, T. (1994). Children orphaned by AIDS: Providing homes for a most vulnerable population. AIDS and Public Policy Journal, 9, 29-35. Centers for Disease Control and Prevention. (1996). HIV/AIDS Surveillance Report, 8, 1-39. Chu, S. Y., & Diaz, T. (1993). Living situations of women with AIDS. Journal of Acquired Immune Deficiency Syndrome, 6, 431-432. Derogatis, L. R., Rickles, K., & Rock, A. E (1976). The SCL-90 and the MMPI: A step in the validation of a new self-report scale. British Journal of Psychiatry, 128, 280-289. Derogatis, L. R., & Spencer, P. M. (1982). The Brief Symptom Inventory Administration, Scoring and Procedures Manual. Baltimore: John Hopkins University School of Medicine. Downey, G., & Coyne, J. C. (1990). Children of depressed parents: An integrative review. Psychological Bulletin, 108, 50-76. Family Health Project Research Group. (in press). The Family Health Project: A multidisciplinary longitudinal investigation of children whose mothers are HIV-infected. Clinical Psychology Review. Forehand, R., & Kotchick, B. (1996). Cultural diversity: A wake-up call for parent training. Behavior Therapy, 27, 187-206. Forehand, R., McCombs, A., & Brody, G. H. (1987). The relationship between parental depressive mood states and child functioning. Advances in Behaviour Research and Therapy, 9, 1-20.

Forehand, R., Steele, R., Armistead, L., Morse, E., Simon, P., & Clark, L. (1998). The Family Health Project: Psychosocial adjustment of children whose mothers are HIVinfected. Journal of Consulting and Clinical Psychology, 66, 513-520. Forsyth, B. W. C., Damour, L., Nagler, S., & Adnopoz, J. (1996). The psychological effects of parental human immunodeficiency virus infection on uninfected children. Archives of Pediatric and Adolescent Medicine, 150, 1015-1020.

422

BIGGAR ET AL.

Kazdin, A. E. (1981). Assessment techniques for childhood depression: A critical appraisal. Journal of the American Academy of ChiM Psychiatry, 20, 358-375. Kazdin, A. E., French, N.H., Uris, A. S., Esveldt-Dawson, K., & Sherick, R. B. (1983). Hopelessness, depression, and suicidal intent among psychiatrically disturbed inpatient children. Journal of Consulting and Clinical Psychology, 51,504-510. Kovacs, M. (1981). Rating scales to assess depression in school aged children. Acta Paedopsychiatrica, 46, 305-315. Kremer, E., Atkinson, J. H., & Ignelzi, R. J. (1981). Measurement of pain: Patient preference does not confound pain measurement. Pain, 10, 241-248. Levine, C. (1995). Orphans of the HIV epidemic: Unmet needs in six US cities. AIDS Care, 7, 557-562. McLanahan, S. (1983). Family structure and stress: A longitudinal comparison of two-parent and female-headed families. Journal of Marriage and the Family, 45, 347-357. Michaeis, D., & Levine, C. (1992). Estimates of the number of motherless youth orphaned by AIDS in the United States. Journal of the American Medical Association, 268, 3456-3461. Politano, P. M., Nelson, W. M., Evans, H. E., Sorenson, S. B., & Zeman, D. J. (1986). Factor analytic evaluation of differences between Black and Caucasian emotionally disturbed children on the Children's Depression Inventory. Journal of Psychopathology and Behavioral Assessment, 8, 1-8. Schable, B., Diaz, T., Chu, S. Y., Caldwell, M. B., Conti, L., Alston, O. M., Sorvillo, T., Checko, P. J., Hermann, P., Davidson, A. J., Boyd, D., Fann, S. A., Herr, M., & Frederick M. (1995). Who are the primary caretakers of children born to HIV-infected mothers? Results from a multistate surveillance project. Pediatrics, 95, 511-515. Taylor-Brown, S. (1991). The impact of AIDS on foster care: A family-centered approach to services in the United States. Child Welfare, 70, 193-209. RECEIVED: May 26, 1997 ACCEPTED: January 12, 1998