Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Fam Process. Author manuscript; available in PMC 2006 June 21.
Published in final edited form as:
PMCID: PMC1480626

Relational Factors and Family Treatment Engagement among Low-Income, HIV-Positive African American Mothers


Clinically derived hypotheses regarding treatment engagement of families of low-income, HIV-positive, African American mothers are tested using univariate and multivariate logistic regression models. Predictors are baseline family relational factors (family support, mother’s desire for involvement with family, and family hassles) and mother’s history of substance dependence. The study examines a subsample of 49 mothers enrolled in a clinical trial testing the efficacy of Structural Ecosystems Therapy (SET). SET is a family-based intervention intended to relieve and prevent psychosocial distress associated with HIV/AIDS. Participants in the subsample were randomly assigned to SET and attended at least two therapy sessions. Findings reveal that family relational factors predicted family treatment engagement (family support, p<.004; mother’s desire for involvement with family, p<.008; family hassles, p<.027). Family support predicted family treatment engagement beyond the prediction provided by the other relational factors and the mother’s own treatment engagement (p<.016). History of substance dependence was neither associated with family treatment engagement nor family support. Post hoc analyses revealed that family hassles (p<.003) and mother’s desire for involvement with family (p<.018) were differentially related to family treatment engagement in low- versus high-support families. Implications for clinical practice and future research are discussed.

Family-based interventions hold promise for preventing and relieving the psychosocial distress associated with adaptation to HIV/AIDS, and as such have been the focus of a major research initiative by the National Institute of Mental Health (Pequegnat & Bray, 1997; Pequegnat & Szapocznik, 2000). Such interventions can help the person afflicted with HIV/AIDS and family members adjust to this serious chronic condition. However, as we learned in a family therapy trial for low-income, HIV-positive African American mothers, it is difficult to engage families in treatment. The research literature reflects this point as well (Coatsworth, Santisteban, McBride & Szapocznik, 2001; Liddle & Dakof, 1995; Stanton & Todd, 1979, 1981; Wermuth & Scheidt, 1986). Engaging families in treatment challenged our clinical team and led us to generate hypotheses about what was happening in these family systems that interfered with or facilitated engagement.

The current study tests clinically derived hypotheses regarding family treatment engagement using univariate and multivariate logistic regression models. We begin with a review of the literature on treatment engagement, in family-based interventions, and engagement of African American s and families with HIV/AIDS in treatment and clinical research. We go on to describe our treatment model, Structural Ecosystems Therapy (SET), and the clinical impressions that led to our hypotheses. The purposes of the study are to begin to disentangle family relational processes associated with family treatment engagement, and provide recommendations for clinical practice. Our view is that it is incumbent upon therapists to recognize how intervention strategies interact with family relational processes, leading to either treatment engagement or “resistance,” a self-protective response of individuals and families.

Treatment Engagement in Family-based Interventions

Empirical studies on family engagement in family-based interventions are limited to interventions aimed at a target child. A computer-assisted search of journal articles from the years 1980 to 2002 using AIDSEARCH, MedLine, and PsychINFO (entering the terms engagement, retention, participation, and dropout, crossed with family or family therapy) yielded no empirical studies on family treatment engagement when the target client was an adult. Some of the factors associated with family engagement in child-centered interventions include demographics such as race and socioeconomic status (Bischoff & Sprenkle, 1993), negative history with institutions (Perrino, Coatsworth, Briones, et al., 2001), and client-therapist ethnic differences (Bischoff & Sprenkle, 1993).

Perrino and colleagues (2001) emphasize the importance of the family system in treatment engagement, noting that its role has not been adequately addressed. These researchers found that higher levels of family organization, communication, and shared views predicted parental engagement in an intervention to prevent adolescent substance abuse. Further evidence of the role of family relational factors in treatment engagement comes from studies demonstrating that intervening in family processes can enhance family engagement in treatment for behavior-problem adolescents (Coatsworth et al., 2001; Santisteban, Szapocznik, Perez-Vidal, et al., 1996; Szapocznik, Perez-Vidal, Brickman, et al., 1988). The premise of these strategies is that addressing the family processes that give rise to the presenting symptom can enhance engagement.

Engagement of African American s and Families with HIV/AIDS

The clinical literature on psychotherapy with African American families identifies engagement barriers such as the negative history between African American families and social institutions, and the fear of exposing family secrets (Boyd-Franklin, 1989). Barriers to recruitment and retention of African American subjects in clinical research (Thompson, Neighbors, Munday, & Jackson, 1996) include mistrust of the research establishment because of a history of unethical research practices with African American s (Armstrong, Crum, Rieger, et al., 1999; Harris, Gorelick, Samuels, & Bempong, 1996). The most infamous of these incidents is the Tuskegee study of untreated syphilis, which contributes to the distrust that impedes HIV-education efforts in African American communities (Thomas & Quinn, 1991). Another barrier to research engagement is reluctance in sharing information with professionals (Holder, Turner-Musa, Kimmel, et al., 1998).

HIV/AIDS adds a layer of complexity to treatment engagement of African American families. On the one hand, crises associated with HIV/AIDS can mobilize the family support network (Boyd-Franklin, Aleman, Steiner, et al., 1995) and therefore be expected to increase family engagement in therapy. On the other hand, secrecy-related barriers (Boyd-Franklin, 1989) are compounded by the stigma of HIV/AIDS, and therefore may impede engagement. Moreover, in view of the multiple challenging circumstances facing poor urban families, the chronic processes of HIV/AIDS are less likely to trigger family responsiveness. In fact, a study of low-income, African American recent mothers found that HIV-positive mothers, like their uninfected counterparts, rated money problems as their top concern, above HIV/AIDS-related concerns (Smith, Feaster, Prado, et al., 2001).

Among family relational factors, only family support has been addressed, albeit obliquely, in relation to engagement of African American s. Holder and colleagues (1998) found that number of household members and proximity of extended family predicted recruitment in a longitudinal family health study. These authors posit that additional household members and nearby relatives may have given the recruited families a sense of greater support and thus kept them more effectively integrated in community institutions.

Structural Ecosystems Therapy (SET)

The data in this study are from a clinical trial comparing the efficacy of SET (Mitrani, Szapocznik, & Robinson-Batista, 2000) to a person-centered approach and community control for reducing distress in low-income, HIV-positive African American mothers. SET is a family-ecological intervention aimed at improving the mother’s well-being through the strengthening and repair of her social support network, with a central emphasis on the family. Although the initial contact is with the mother, SET emphasizes the importance of participation by family and members of other formal and informal support systems. Therapy is conducted in the client’s community to facilitate participation.

SET therapists encourage their clients to invite family members to sessions, and attempt to work through barriers to participation. With the mother’s consent and giving special attention to not revealing confidential information, SET therapists make considerable efforts toward engaging family members via telephone calls and face-to-face encounters in the client’s natural environment. The rationale typically presented to clients is that the family’s participation serves as a source of support to the mother, and/or facilitates resolution of family conflicts.

Despite these efforts, it was difficult to draw adult family members into formal therapy sessions. Of 49 cases in which the mother was engaged in SET (defined as attending two or more sessions), 22 (45%) did not meet our criteria for engagement of family members (two or more sessions attended by an adult member of the formal family system).

Regarding the operationalization of “family engagement,” we emphasize that SET recognizes the legitimacy of extended kinship networks that include persons not traditionally recognized as family, and targets these relationships for treatment. In addition, the relationships between mothers and their children are an important focus of treatment. However, in the current study we limit our focus to engagement of adult members within the formal family system. There are two reasons for setting this parameter: (a) children are excluded in an effort to limit the study to engagement of family members who were likely to be present by their own volition, which might not be the case with children; and (b) only members of the mother’s nuclear (parents/stepparents, siblings, significant other) and extended family (grandparents, uncles/aunts, cousins, in-laws) are included for purposes of maintaining uniformity between this study and the existing literature on engagement, thus allowing for comparisons.1 Likewise, the criterion of two sessions is established to match the criterion used in earlier studies conducted by our research team (Coatsworth et al., 2001; Prado, Szapocznik, Mitrani, et al., 2002; Santisteban et al., 1996).

Clinical Impressions on Family Treatment Engagement Factors

The mothers commonly presented logistical explanations (e.g., busy schedules, transportation problems, unexpected circumstances) for the absence of family members at sessions. Further in-therapy exploration of barriers revealed that beyond these practical reasons, family engagement in treatment was influenced by a complex set of factors having to do with family relationships (Mitrani et al., 2000). Relational barriers to family treatment engagement appear to have sprung from three sources which are listed in order of their apparent influence: (a) unavailable family members, (b) the mothers’ desire to exclude family, and (c) no sense of urgency to work on family problems.

The availability of family members seemed to be a major factor in determining whether or not family participated in therapy. Family unavailability was often because of estrangement or weak bonds with the mother. That is, unavailability for attending therapy seemed to reflect a larger unavailability to lend support. This weak bonding was related either to a family system that was generally fragmented and disengaged, or alienation of the mother within an otherwise intact family system. These latter cases appeared to be related to the mother’s impaired functioning, often involving a history of addiction.

Another issue was the mother’s own wishes regarding her family’s participation. In some cases, even mothers who were connected to supportive families actively excluded them from therapy. This barrier seemed to stem from the mother’s desire to maintain boundaries between herself and her family, or to contain family problems in the context of volatile or fragile relationships.

Insufficient motivation to address family problems in therapy also seemed to restrain family engagement. For inclusion in the larger study, mothers were required to report at least one family problem (e.g., conflicts, communication problems, fear of HIV-disclosure, drugs/alcohol, child custody or other child-related problems, or problems with a romantic partner). However, once in therapy, some mothers claimed to have no problems with family members, i.e., family hassles, that they wished to address.

The current study tests three hypotheses based upon these clinical observations: Hypothesis 1. Family relational factors (family support, mother’s desire for involvement with family, and family hassles) will predict family engagement in treatment; Hypothesis 2. History of substance dependence will be associated with less family support and lower family treatment engagement; and Hypothesis 3. Family support will predict family treatment engagement beyond the effect of other predictor variables. Post hoc analyses explore whether or not family hassles and mother’s desire for involvement with family are differentially related to family treatment engagement in low- versus high-support families.



Participants in the current study are 49 low-income, HIV-positive African American mothers who were among 209 enrolled in a longitudinal study on the efficacy of SET. To be eligible for participation in the larger study, women must have met seven criteria at baseline: (a) HIV-positive, (b) second-generation African American , (c) at least 18 years of age, (d) had at least one child, (e) not homeless, (f) reported no use of illicit drugs during the previous 6 months, and (g) identified at least two interpersonal problems, one of which was family-related. In addition, at the beginning of the study, a criterion was established to exclude mothers who did not have a minimum of 200 CD4 cells at admission. As the efficacy of protease inhibitors became known, the criterion was reduced to 50 CD4 cells at admission.2 All exclusion criteria were established to maximize the likelihood of retaining participants in the study.

Participants were recruited during the period of time between fall 1996 and spring 1999, and came from community-based agencies that provide health and/or social services to HIV-infected individuals. Among these agencies was a large urban medical center and its affiliated clinics. After determining study eligibility, the mother and the recruiter arranged for the baseline assessment session. After explaining the study and prior to starting the assessment, the assessor read and obtained informed consent.

Characteristics of the Current Sample

The subsample in the current study consists of those mothers who were successfully engaged in SET. That is, they were randomly assigned to the SET condition and attended two or more sessions. Unless otherwise indicated, all data presented are from this subsample of 49 mothers. With the exception of session attendance data, the mothers reported all information at the baseline assessment.

Mean age of the sample was 34.76 years (SD=8.07). The reported time since HIV diagnosis ranged from 6 months to 16 years (median = 6 years, 5 months). Modal level of education was less than completion of high school. Most of the mothers were unemployed (80%) and receiving some form of public assistance (86%). Median annual income was $6,384.

As per inclusion criteria, all mothers had at least one child (median number of children=3), with 78% having two or more children. Fourteen percent had at least one child that was perinatally infected with HIV. In addition, each mother was asked to identify family members who lived with her, who were available to offer instrumental or emotional support, or with whom she had the most disagreements. Family members thus identified included children (33 cases, or 67%),3 spouses or partners (18 cases, or 37%), siblings (12 cases, or 24%), parents (12 cases, or 24%), grandchildren (2 cases, or 4%), stepchildren (2 cases, or 4%), ex-boyfriend or spouse (1 case, or 2%), grandparents (1 case, or 2%), and other extended family (19 cases, or 39%).

The mothers reported their marital status as either (a) not married and not living with a partner (45%), (b) not married but living with a partner (23%), (c) divorced (18%), (d) married and living with a partner (10%), (e) married but living apart (2%), or (f) widowed (2%).

Treatment Parameters

Therapy sessions were offered at whichever location was convenient to the mother, her family, or other members of her extended network. The vast majority of sessions were conducted in the mother’s home. Sessions were routinely videotaped for clinical supervision and treatment adherence rating unless a session participant objected or circumstances precluded taping. The minimum attendance required by the intervention was eight sessions. The average number of session hours was 10.39 (SD = 7.07). The average session length was 63.07 minutes (SD = 32.89). The median and modal length of session was 60 minutes.


The measures used in this study were selected from the larger battery administered to the mother and her family. The measures section is divided into two subsections: (a) instruments and (b) variables (that describes how instruments were used in measuring each variable).

Instruments. Feetham Family Functioning Survey (Roberts & Feetham, 1982). This instrument assesses family functioning by asking respondents to rate four dimensions for each of 36 items regarding their family relationships. A sample item is: “The amount of discussion with your relatives regarding your concerns and problems.” The four dimensions are: 1) “How much is there now?”; 2) “How much should there be?”; 3) “How important is this to you?”; and 4) “How much would you like there to be?” Each dimension is rated on a 7-point Likert scale ranging from “little” to “much.”

Social Support Questionnaire Short Form (Sarason, Sarason, Shearin, & Pierce, 1987). This instrument asks respondents to list, and specify their relationship to (e.g., sister, friend, spouse, etc.) every individual on whom they can rely for six aspects of support (e.g., dependability, emotional support, acceptance). Respondents also rate their satisfaction with their social support network on each of the six aspects of support using a 6-point Likert scale ranging from “very satisfied” to “very dissatisfied.”

Hassles Scale (Delongis, Folkman, &Lazarus, 1988). The Hassles Scale assesses problems occurring in the past month, asking respondents to rate the extent to which each item is a hassle to them. The original Hassles Scale was modified in the larger study for relevance to our population. This modified version of the Hassles Scale (Smith et al., 2001) contains 64 items. Item domains include family, finances, house chores, work, and education.

Brief Symptom Inventory: Global Severity Index (Derogatis, 1993). This measure consists of 53 items that assess the respondent’s psychological symptoms within the past seven days. The overall composite scale includes the following dimensions: (a) depression, (b) anxiety, (c) paranoid ideation, (d) psychoticism, (e) somatization, (f) interpersonal sensitivity, (g) hostility, (h) phobic anxiety, and (i) obsessive-compulsive behavior. Each item is rated on a 5-point Likert scale ranging from “not at all” to “extremely.” A sample item is “How much were you distressed by feeling others are to blame for most of your troubles.”

Structured Clinical Interview for DSMIII-R Diagnosis Non-Patient Version for HIV-Infected Persons (SCID-NP-HIV; Spitzer, Williams, Gibbon, & First, 1988). The SCIDNP-HIV is widely used to assess current and lifetime psychotic disorders as well as disorder of mood, psychoactive substance use, anxiety, and adjustment. Three research associates familiar with DSMIII-R criteria administered the interview, and achieved an interrater reliability with each other (kappa) of .81. The SCID-NPHIV was also used to determine study eligibility. Mothers who met criteria for current substance abuse or dependence were excluded from the study.


Family treatment engagement (dependent variable). Family treatment engagement was coded as a dichotomous variable (family engaged vs. family not engaged). Mothers who had an adult family member present at two or more therapy sessions were categorized as family engaged. As noted, for the current study, “family” was restricted to parents/stepparents, siblings, significant other, grandparents, uncles/aunts, cousins, and in-laws.

Family support (independent variable). Family support was measured using the total count of family members (excluding children4) listed in the Social Support Questionnaire Short Form (Sarason et al., 1987) and six items from the Feetham Family Fucntioning Survey (Roberts & Feetham, 1982). The items used from the Feetham Family Functioning Survey are those having to do with emotional or instrumental support received from family members (excluding children). The ratings on these six items for the dimension of “How much is there now?” are used in this scale. Cronbach’s alpha for the family support scale is .82.

Family hassles (independent variable). This variable was measured using seven items from the Hassles Scale (Delongis, et al., 1988) and two items from the Feetham Family Functioning Survey (Roberts & Feetham, 1982). The items used from the Hassles Scale are those related to hassles with family members (excluding children). To avoid confounding with measures of distress, we used the count of hassles rather than their impact. The items used from the Feetham Family Functioning Survey are those related to disagreements with family members (excluding children). The ratings on these two items for the dimension of “How much is there now?” are used in this scale.

Mother’s desire for involvement with family (independent variable). This factor was measured using ratings on the “How much would you like there to be?” dimension from 14 items of the Feetham Family Functioning Survey (Roberts & Feetham, 1982). Items used are those having to do with support from and affiliation with family members (excluding children). Cronbach’s alpha for this scale is .84.

Distress (independent variable). Distress was measured using the Global Severity Index (Derogatis, 1993) of the Brief Symptom Inventory. Cronbach’s alpha for the Global Severity Index in this study is .96.

History of substance dependence (independent variable). Substance dependence included dependence on either alcohol or drugs. Participants who met the criterion, as determined by the SCID-NP-HIV, were categorized as having a history of substance dependence.


Preliminary Analyses

A series of ANOVAs was conducted to investigate whether the two groups, family engaged (n=27) vs. family not engaged (n=22), differed on factors not included in the hypotheses, but possibly related to family treatment engagement. These factors were age, education, income, time since HIV diagnosis, number of sessions attended by the mother, mother’s distress, and therapist assignment. The results revealed that only the number of mother’s sessions discriminated between the two groups, with mothers in the family engaged group having attended significantly more sessions (p<.0001). Number of mother’s sessions was therefore used as a control variable in testing whether family support predicted family treatment engagement beyond the effect of the other predictors (Hypothesis 3).

Family Relational Variables and History of Substance Dependence as Independent Predictors (Hypotheses 1 and 2)

The correlation between family support and history of substance dependence was not significant (r = -.157, p>.39), thus refuting the first part of Hypothesis 2 (history of substance dependence will be associated with less family support).

We tested Hypothesis 1 (family relational variables will predict family treatment engagement) and the second part of Hypothesis 2 (history of substance dependence will predict lower family treatment engagement) by entering each predictor into a univariate logistic regression model. All available observations were used in these models, resulting in a different sample size for history of substance dependence (family engaged n= 19; family not engaged n = 12) because of missing data. This approach uses all available information (Little & Rubin, 1987). The univariate logistic regression analyses (Table 1) revealed that each of the three family relational factors significantly predicted family treatment engagement (family support, p< .004; mother’s desire for involvement with family, p< .008; family hassles, p<.027), thus confirming Hypothesis 1. History of substance dependence did not predict family treatment engagement, thus refuting this aspect of Hypothesis 2.

Table 1
Family Relational Factors and Substance Abuse History as Predictors of Family Treatment Engagement: Results of Univariate Logistic Regression Models

We calculated the odds ratio, a standard measure of effect size in a logistic regression, for each of these effects to illustrate their relative magnitude. The odds ratios were calculated by comparing .5 standard deviations above to .5 standard deviations below the mean. The likelihood of family member engagement increased by a factor of 2.61 for mothers who reported more family support, 2.33 for mothers who reported more desire for involvement with family, and 2.00 for mothers who reported more family hassles.

Incremental Prediction from Family Support (Hypothesis 3)

A hierarchical logistic regression analysis was used to test Hypothesis 3 (family support will predict family treatment engagement beyond the effect of the other predictors). For this analysis, all variables that were significant (p<.05), in the single predictor logistic analyses, were entered into the hierarchical logistic regression model (Table 2), which tests whether an individual predictor has a significant effect independently of other predictors. Checks for multicollinearity among predictors indicated that this would not be a concern (Menard, 1995).

Table 2
Family Relational Factors as Predictors of Family Treatment Engagement: Results of the Hierarchical Logistic Regression Model

Variables were entered sequentially, in reverse order of their conceptualized importance in predicting family treatment engagement. The variables were entered in the following order: (1) number of mother’s sessions (control variable), (2) family hassles, (3) mother’s desire for involvement with family, and (4) family support.

The final model was highly significant (p<.0001), with number of mother’s sessions (p<.008) and family support (p<.016) significantly predicting family member engagement. The model correctly classified 87.8% of the cases. Family support predicted family treatment engagement after controlling for the other variables, thus confirming Hypothesis 3.

For interpretational purposes, we examined the estimated coefficients after entering each variable. The first variable entered, the number of sessions attended by the mother, was highly significant (p<.0001). Mothers who participated in more therapy sessions were 9.53 times more likely to have family members engaged. After entering family hassles, the number of sessions attended by the mother remained significant. That is, the number of family hassles reported by the mother did not appear to mediate the effects of the number of sessions attended by the mother. The significant effect of family hassles apparent in the univariate analysis was not seen in the hierarchical model, indicating that the number of family hassles was related to the number of sessions that the mother attended (r2=.38, p<.02). Thus, any effect that family hassles had on family treatment engagement was likely mediated by the effect of family hassles on the number of sessions the mother attended.

Mother’s desire for involvement with family was a strong predictor of family treatment engagement after controlling for number of mother’s sessions and family hassles, increasing its predictive power by a factor of almost three from the univariate estimate. Here, mothers whose desire for involvement with family was .5 standard deviations above the mean were almost 6 times more likely to have a family member engaged than, mothers whose desire for involvement with family involvement was .5 standard deviations below the mean.

After entering family support into the hierarchical logistic regression model, mother’s desire for involvement with family was no longer significant. Thus, the effect of mother’s desire for involvement with family on family treatment engagement was likely mediated by the amount of family support. Minimally, this implies that there was a tendency for mothers who desired involvement with family to have high family support and those who did not desire family involvement to have little or no family support.

Family Hassles and Mother’s Desire for Involvement with Family in Low- Versus High-Support Families (Post Hoc Analyses)

Because family support was confirmed as a powerful predictor of family treatment engagement, we were interested in whether it had an impact on the relationship between family treatment engagement and the other family relational variables. Chi-square analyses were conducted to determine whether the influence of family hassles and mother’s desire for involvement with family depended on the amount of family support. These analyses examined the odds ratio of family treatment engagement differentially by family support. To examine these interactions, continuous variables (family support, family hassles, and mother’s desire for family involvement) were categorized into dichotomous variables (“low” and “high”) using a median split.

Family treatment engagement and family hassles in low- vs. high-support families. The family support by family hassles interaction was significant. For mothers with low family support, family hassles predicted family treatment engagement, χ2(1) = 9.67, p < .003, such that mothers with high family hassles were 9.67 times more likely to have family treatment engagement than mothers with low family hassles. Thus, for women with low family support (Figure 1), having low family hassles was a barrier to engaging the family into therapy, resulting in a family engagement rate of only 7.7%, in contrast to a 70% engagement rate when family hassles were high. High-support families engaged at similar rates regardless of family hassles.

Figure 1
Family Engagement and Family Hassles in Low vs. High Support Families

Family treatment engagement and mother’s desire for involvement with family in low- vs. high-support families. The family support by mother’s desire for involvement with family interaction also proved to be significant. For mothers in high-support families, desire for involvement with family was a significant predictor of family treatment engagement, χ2(1) = 5.74, p< .018, such that mothers with a high desire for involvement with family were 5.74 times more likely to have their families engaged in treatment. For these mothers, family treatment engagement was 88.2% when they had high desire for involvement with family, but only 44% when they had low desire for involvement with family (Figure 2). For mothers with low family support, the desire for involvement with family did not predict family treatment engagement, with engagement rates of 33% irrespective of desire for involvement with family.

Figure 2
Family Engagement and Mother's Desire for Involvement with Family in Low vs. High Support Families


The purpose of this study was to examine the role of family relational factors in treatment engagement of adult family members of low-income, HIV-positive African American mothers. We demonstrated that the family relational factors of support for the mother, the mother’s desire for involvement with her family, and the mother’s hassles with family members all predicted engagement. In addition, multivariate modeling showed that after the effect of the other relational factors and the mother’s level of engagement were taken into account, family support still added significantly to the prediction of family treatment engagement. The finding regarding family support suggests that such support generalized to participation in therapy that was aimed at helping a family member. Thus, the existence of supportive bonds was more powerful than relational barriers to family treatment engagement.

Given its key role in family treatment engagement, future studies should address factors associated with family support for HIV-positive mothers, and work toward the development of treatment strategies for enhancing support and reducing isolation. One important issue in this regard is stigma related to HIV/AIDS within the family. Some of the mothers described being forced to use separate eating utensils at family gatherings, and other forms of segregation.

Exploring the interaction between family support and the other relational factors shed additional light on their association with family treatment engagement. We found that in high-support families, the mother’s desire for involvement with family was highly predictive of family treatment engagement, whereas in lowsupport families, family engagement in treatment was low regardless of the mother’s desire for involvement with her family. This finding suggests that mothers with supportive families could choose whether or not to include their families in treatment, whereas mothers with less supportive families did not have this option. It seems that in supportive families it is particularly important to work through the mother’s own reluctance to include family members, perhaps by contracting to help her maintain or strengthen boundaries within the family.

The presence of family hassles was associated with increased family treatment engagement in low-support families, but not in high-support families. It is possible that in high-support families, the mother’s in-therapy disavowal of family problems was a protective effort to prevent family confrontations. The relationship between family support and conflict resolution styles should be addressed in future studies.

In low-support families, it seems that problematic relationships may have stimulated family participation in treatment, particularly by engendering the mother’s own commitment to treatment. Thus, in low-support families, which are the most challenging for family treatment engagement, a particularly effective engagement strategy for both the mother and her family may be to highlight the value of family therapy for addressing family problems.

The finding that the pressure of family hassles was related to engagement in low-support families is noteworthy given that African American s are believed to eschew revealing private family matters in psychotherapy. If, as it appears, these families were motivated to participate in therapy to work on their family hassles, then they showed a seemingly uncharacteristic willingness to air family business. It is possible that there is less reluctance to reveal family problems in low-support families because family members do not feel as protective of the family, or because they are motivated to work on family problems as a means of strengthening bonds.

There were two aspects of our intervention that may also have facilitated the willingness to address family hassles. The first is SET’s emphasis on building therapeutic alliances based on compatible relational goals, and joining techniques that respect the existing family structure. Secondly, all of the therapists were themselves African American women. It is possible that the combination of ethnic matching and careful attention to alliance building helped counteract concerns about intrusiveness and judgmental attitudes from therapists. Future studies might address the impact of these factors on treatment engagement.

Finally, substance dependence history, which we expected to be inversely related to family treatment engagement and family support, was not associated with either. This seems to speak to the resilience of the families in regard to wear and tear on relationships because of addiction. One possible scenario is that substance dependence led to circumstances that served to maintain family bonds, such as joint caregiving of children. It is also possible that family support toward the mother was related to HIV/AIDS, such that concern for the mother’s health and/or sympathy for her situation overrode past injuries. This is an important area for future study and can clarify the impact of substance dependence on family relationships, and suggest strategies for maintaining bonds.


A limitation of the current study is that it relied upon the mother’s self-report for relational variables. Thus, we have the view of one person regarding family functioning variables such as support and hassles. Future studies should use reports from multiple family members or more objective indicators of family relational factors, such as observational data.

Another limitation is the restriction on the definition of “family” to exclude members of the extended kinship network. As noted, this restriction was imposed to maximize comparability of this study to the existing literature on treatment engagement. Future studies might include a broader operationalization of family to examine whether these findings apply to the larger extended kinship network.

Finally, another limitation of the current findings is the constrained variability on history of substance dependence. Of the 31 mothers in the current sample for whom we have substance use history data, only 8 did not have a history of substance dependence. Of those eight, five had a history of substance abuse. We investigated whether the lack of correlation between history of substance dependence and family support would hold up when we looked at the data set from the entire clinical trial, and found support for the earlier finding (r = .002, p> .98). Of 196 mothers in the larger data set with substance use history data, 63 had no history of substance dependence, and only 10 of those 63 had a history of substance abuse. Although this test bolsters our confidence in the finding regarding a lack of relationship between substance dependence and family support, further study is warranted.

The impetus for this study was the observation that involving family members in family therapy is a difficult proposition. This study begins to disentangle relational mechanisms involved in family treatment engagement by testing hypotheses based upon clinical experience. The findings suggest studies that can further tease apart these complex processes, and clinical approaches for engendering treatment engagement. While some of our clinical hunches were supported, others were not, demonstrating both the value and danger of relying on clinical presumption. We believe that the iterative process demonstrated in this study is a valuable paradigm for advancing family interventions.


We express our appreciation to the families who participated in this project. This research was funded by a grant from the National Institute of Mental Health (grant 1-R37 MH-55796) to José Szapocznik, Principal Investigator.

1In a critical review of the research on family therapy dropout studies, which appeared in this journal, Bishoff and Sprenkle (1993) decried the lack of common operational definitions which make such studies difficult to compare. We have therefore paid special attention to defining our engagement variable in a manner that maximizes comparabiity to other engagement studies.

2We found no significant correlation between reported CD4 count and family engagement (p>.8).

3Although all of the mothers had children, not all had substantive contact with their children.

4Items regarding children were excluded from all of the relational variables in order to mirror the procedure of excluding children from the count of family who attended sessions.


  • Armstrong TD, Crum LD, Rieger RH, Bennett TA, Edwards LJ. Journal of Applied Social Psychology. 1999;29:552–574.
  • Bischoff RJ, Sprenke DH. Dropping out of marriage and family therapy: A critical review of research. Family Process. 1993;32:353–375. [PubMed]
  • Boyd-Franklin N. Black families in therapy. Guilford Press; New York: 1989.
  • Boyd-Franklin N, Aleman JC, Steiner GL, Drelich EW, Norford BC. Family systems interventions and family therapy. In: Boyd-Franklin N, Steiner GL, Boland MG, editors. Children, families and HIV. Guilford Press; New York: 1995.
  • Coatsworth JD, Santisteban DA, McBride CK, Szapocznik J. Brief family therapy versus community control: Engagement, retention, and an exploration of the moderating role of adolescent symptom severity. Family Process. 2001;40:313–332. [PubMed]
  • Delongis A, Folkman S, Lazarus RS. The impact of daily stress on health and mood: Psychosocial and social resources as mediators. Journal of Personality and Social Psychology. 1988;54:486–495. [PubMed]
  • Derogatis LR. Brief Symptom Inventory: Administration, scoring and procedures manual. National Computer Systems; Minneapolis, MN: 1993.
  • Harris Y, Gorelick PB, Samuels P, Bempong I. Why African Americans may not be participating in clinical trials. Journal of the National Medical Association. 1996;88:630–634. [PMC free article] [PubMed]
  • Holder B, Turner-Musa J, Kimmel PL, Alleyne S, Kobrin S, et al. Engagement of African American families in research on chronic illness: A multisystem recruitment approach. Family Process. 1998;37:127–151. [PubMed]
  • Liddle HA, Dakof GA, Little RJA, Rubin DB. Journal of Marital and Family Therapy. Statistical analysis with missing data. Vol. 21. John Wiley & Sons; New York: 1995. 1987. Efficacy of family therapy for drug abuse: Promising but not definitive; pp. 511–543.
  • Menard S. Applied logistic regression analysis. Sage Publications; Thousand Oaks, CA: 1995.
  • Mitrani VB, Szapocznik J, Robinson-Batista CR. Structural Ecosystems Therapy with HIV-positive African American mothers. In: Pequegnat W, Szapocznik J, editors. Working with families in the era of HIV/AIDS. Sage Publications; Thousand Oaks, CA: 2000.
  • Pequegnat W, Bray JH. Families and HIV/AIDS: Introduction to the special section. Journal of Family Psychology. 1997;11:3–10.
  • Pequegnat W, Szapocznik J. The role of families in preventing and adapting to HIV/AIDS: Issues and answers. In: Pequegnat W, Szapocznik J, editors. Working with families in the era of HIV/AIDS. Sage Publications; Thousand Oaks, CA: 2000.
  • Perrino T, Coatsworth JD, Briones E, Pantin H, Szapocznik J. Initial engagement in parent-centered preventive interventions: A family systems perspective. Journal of Primary Prevention. 2001;22:21–44.
  • Prado G, Szapocznik J, Mitrani VB, Mauer MH, Feaster DJ. Factors influencing engagement into a behavioral intervention designed for HIV-positive mothers. AIDS and Behavior. 2002;6:141–151. [PMC free article] [PubMed]
  • Roberts CS, Feetham SL. Assessing family functioning across three areas of relationships. Nursing Research. 1982;31:231–235. [PubMed]
  • Santisteban DA, Szapocznik J, Perez-Vidal A, Kurtines WM, Murray EJ, LaPerriere A. Efficacy of intervention for engaging youth and families into treatment and some variables that may contribute to differential effectiveness. Journal of Family Psychology. 1996;10:35–44.
  • Sarason IG, Sarason BR, Shearin EN, Pierce GR. A brief messure of social support: Practical and theoretical implications. Journal of Social and Personal Relationships. 1987;4:497–510.
  • Smith L, Feaster DJ, Prado G, Kamin M, Blaney N, Szapocznik J. The psychological functioning of HIV-positive and HIV-African American recent mothers. AIDS and Behavior. 2001;5:219–231.
  • Spitzer, Williams, Gibbon First. 1988
  • Spoth R, Redmond C. Study of engagement barriers in family-focused prevention: Research issues and preliminary results. International Quarterly of Community Health Education. 1994;13:365–388. [PubMed]
  • Stanton MD, Todd TC. Structural family therapy with drug addicts. In: Kaufman E, Kaufman P, editors. Family therapy of drug and alcohol abuse. Gardner Press; New York: 1979.
  • Stanton MD, Todd TC. Engaging "resistant" families in treatment: II. Principles and techniques in recruitment: III. Factors in success and cost-effectiveness. Family Process. 1981;20:261–293. [PubMed]
  • Szapocznik J, Kurtines WM. Breakthroughs in family therapy with drug-abusing adolescents. Springer; New York: 1989.
  • Szapocznik J, Perez-Vidal A, Brickman AL, Foote FH, Santisteban, et al. Engaging adolescent drug abusers and their families in treatment: A strategic structural systems approach. Journal of Consulting and Clinical Psychology. 1988;56:552–557. [PubMed]
  • Szapocznik J, Rio AT, Hervis O, Mitrani VB, Kurtines W, Faraci AM. Assessing change in family functioning as a result of treatment: The Structural Family Systems Rating scale (SFSR) Journal of Marital and Family Therapy. 1991;17:295–310.
  • Szapocznik J, Williams RA. Brief strategic family therapy: Twenty-five years of interplay among theory, research and practice in adolescent behavior problems and drug abuse. Clinical Child and family Psychology Review. 2000;3:117–135. [PMC free article] [PubMed]
  • Thomas SB, Quinn SC. The Tuskeegee Syphilis Study, 1932-1972:Implications for HIV education and AIDS risk education programs in the black community. American Journal of Public Health. 1991;81:1498–1505. [PubMed]
  • Thompson EE, Neighbors HW, Munday C, Jackson JS. Recruitment and retention of African American patients for clinical research: An exploration of response rates in an urban psychiatric hospital. Journal of Consulting and Clinical Psychology. 1996;64:861–867. [PubMed]
  • Wermuth L, Scheidt S. Enlisting support in drug treatment. Family Process. 1986;25:25–33. [PubMed]