Program factors related to women's substance abuse treatment retention and other outcomes: A review and critique

3.1.1. Critique 

Although all nine studies contribute to the knowledge base, they also all suffer some design weaknesses. Grosenick and Hatmaker (2000a) and Nelson-Zlupko et al. (1996) surveyed women's perceptions instead of actually measuring the actual behaviors targeted. Ravndal and Vaglum (1994) included only six women in the mixed-sex group and four in the women-only group, making generalization difficult. The women-only and mixed-sex samples in the study of Zilberman, Tavares, Andrade, and El-Guebaly (2003) were collected in different periods (the former sample, 1997–1998; the latter sample, 1986–1996), as were those in the study of Zankowski (1987; women only, 1986; mixed sex, 1984); thus, the impact of selection bias (e.g., change of subject profiles over time) cannot be ruled out.

The remaining four quantitative studies did not control for multiple factors associated with treatment outcome that were operating within their studies. Therefore, it is difficult to assess which factor(s) contributed most or even at all to the outcomes. For example, the single-sex program in the study of Gutierres and Todd (1997) not only involved a single sex but also provided child care and culturally sensitive practices, which were unavailable for women in the coed program. The single-sex program in the study of Roberts and Nishimoto (1996) provided day treatment whereas the mixed-sex program provided a traditional outpatient program. Although Dahlgren and Willander (1989) used a randomized controlled trial, their women-only group lasted for 8 months whereas their control group lasted for only 5 months. Because LOS proved to be significantly related to outcome for both groups, the longer treatment for the women-only group makes it difficult to determine the pure effect of the single-sex factor (Hodgins, El-Guebaly, & Addington, 1997).

Therefore, the hypothesis that a single-sex program may have more positive treatment outcomes for women seems to be supported by the research findings, with a ratio of nine versus three studies. However, a critical review suggests significant methodological problems with most of these studies and more empirical studies, with better methods needed before we can confidently claim such a relationship.

In addition, four other design and procedure issues are worthy of attention. The first issue involves controlling “substance abuse choice” in evaluating treatment program effectiveness. Zilberman, Tavares, Andrade, and El-Guebaly (2003) found that only alcohol-dependent women, not drug-dependent women, in the women-only program had significantly higher retention than women in the mixed-sex group. Their findings may reconcile the inconsistent results between the study of Dahlgren and Willander (1989) and that of Bride (2001). The Dahlgren and Willander study (findings suggested better outcome/retention in the women-only group) included only “women in early phases of alcohol dependence” (p. 499) whereas the Bride study (which found no outcome/retention difference between the women-only and mixed-sex groups) had 72% of the sample identify cocaine as their primary drug. Drug-dependent women may be a distinctly different group as compared with alcohol-dependent women (Zilberman, Tavares, & Andrade, 2003) and, therefore, require different treatment strategies.

The second issue concerns standardizing outcome variables and time intervals for follow-up measurements. In this regard, McLellan et al. (1996) suggested three outcome domains: “reduction of alcohol and drug use,” “improvement in personal and social function,” and “reduction in public health and safety threats” (p. 56). The outcome variables and the times that they were measured varied considerably in the studies reviewed. Some studies included LOS only; others included abstinence and relapses at the 1- and 2-year follow-ups, depression and self-esteem at the 6-month follow-up, and so forth.

For alcohol treatment effectiveness, Kadden and Litt (2004) suggested drinking frequency, intensity, and consequences as universal treatment outcome indicators but not psychosocial or biologic measurements. From the perspective of study replication and cross-study comparison, an overlap of outcome variables across various studies would be both necessary and feasible.

A third issue is whether the single-sex element should be applied to both clients and staff or just to clients. Although 82.4% of the women in the Grosenick and Hatmaker (2000b) study perceived it important to have female staff, 38% believed that having male staff as male figures for their children was also important, to show that “not all men are bad” and to offer male perspectives on issues such as anger and healthy relationships. Finally, Blume and Zilberman (2004) suggested that single- or mixed-sex group choice should be based on an individual client's needs; the combination of women-only group therapy and mixed-sex group therapy may be suggested; and a client's needs may change throughout a treatment process and, thus, the treatment strategies may need to be reassessed.

3.2.1. Residential treatment and retention/other outcomes 

Five studies (one qualitative and four quantitative) were included in this category, and all those reviewed showed that residential treatment had a more positive impact on women's treatment outcome than the alternative (usually outpatient care). One qualitative study (Grosenick & Hatmaker, 2000a) offered a theoretical basis for this conclusion. It found that clients consistently reported that the residential nature of treatment would help them decrease substance abuse and prepare them for independent living after discharge. Most of these clients had received outpatient treatment previously and had relapsed. Perhaps because women are more likely than men to lack resources and to experience stress and burdens from their families (Yaffe, Jenson, & Howard, 1995), a residential program might be more necessary to provide women with a more stabilized structure and necessary support to help them concentrate on recovery (Grosenick & Hatmaker, 2000a).

Two quantitative studies found that residential treatment had a more positive impact on women's treatment outcomes than did less-intensive forms of care. Messina et al. (2000) compared clients randomly assigned to a longer-term inpatient program (10 months of inpatient treatment plus 2 months of outpatient treatment) with those assigned to an abbreviated inpatient program (6 months of inpatient treatment plus 6 months of outpatient treatment). They found that at the 19-month postdischarge follow-up, women in the longer-term program had lower criminal activity and higher employment rates than those in the abbreviated program. Sowers et al. (2002) compared women attending a women's residential program with those attending a day treatment program and found that the residential clients had better outcomes in abstinence, criminal activity, and employment at the “posttest and follow-up” (the specific period not reported, possibly at the end of treatment or shortly after program completion).

Two other quantitative studies suggested the greater impact of residential treatment on women's successful completion of continuing treatment. Scott-Lennox et al. (2000) found that women who had received prior residential care from the same agency as their outpatient treatment were more likely to complete the outpatient treatment. Coughey et al. (1998) found that clients who had received residential treatment were more likely to complete the aftercare program than those who had not received residential treatment. The aftercare program case managers stated that women with unstable conditions or inadequate prior drug treatment were most likely to drop out and that those women should receive “first care” prior to aftercare.

3.2.2. Treatment intensity, retention, and other outcomes 

The review findings indicate that treatment intensity increases retention (four studies) but has only limited impact on treatment outcomes (one study). This means, for example, that a woman may be more likely to complete a treatment if she attends treatment more than once a week but this may not make much difference in her treatment outcomes (AOD use, employment, and criminal behavior), assuming she completes the treatment. It may be that the more intensively a person attends a program, the more likely she will identify with the program, its staff, and other members; develop a sense of belongingness; and, therefore, be less likely to drop out of the program. However, once the program is completed, personal and environmental relapse factors may operate in much the same way, regardless of the intensity of the treatment experience.

Three of the four studies (i.e., Comfort & Kaltenbach, 1999, Roberts & Nishimoto, 1996, Strantz & Welch, 1995) indicate that the dropout rate may not differ much between regular and intensive outpatient programs or among regular, intensive, and residential programs for the first 3 months after admission but that both the intensive outpatient and residential programs have significantly better retention than the regular outpatient program after 3 months. One of the four studies (Wald, 1992) found that clients who received intensive outpatient treatment (≥6 hours of counseling per week) were more likely to complete treatment than those assigned to the less-intensive treatment (51% vs. 19%, p < .0001).

3.3.1. Critique 

Only two of the four studies (Metsch et al., 2001; Stevens & Patton, 1998) investigated follow-up outcomes in addition to LOS. Stevens and Patton measured AOD use, employment, child custody, incidence of arrest, and aftercare involvement. Metsch et al. measured AOD use. Both follow-ups were done at 6 months after discharge; however, Stevens and Patton did not provide p values and it is uncertain if the differences are statistically significant. Furthermore, the data provided by Hughes et al. (1995) indicated that although women in the with-child care program had a significantly higher LOS (M = 300.4 days) than those in the no-child care program (M = 101.9 days; p < .05), the standard deviation was much greater for the former program (SD = 242.3 days) than for the latter program (SD = 93.7 days). This may mean that the with-child care program did not benefit all mothers and thus may enhance retention for only certain groups of mothers. In this regard, a CSAT-sponsored focus group of women (CSAT, 2001) revealed that not all women preferred to have their children join them in residential treatment. Some women felt that being a consistent parent was too demanding and thus wanted to initiate their recovery alone to build a foundation and then gradually involve their children into their recovery program. Unfortunately, these possibilities could not be verified through examination of the other three studies in our review (Metsch et al., 2001; Stevens & Patton, 1998; Wobbie et al., 1997); none of them provided standard deviations for their mean LOS.

In summary, these studies offer consistent findings on this issue: the state of knowledge regarding child care and treatment outcomes can still be improved. Future studies should provide complete statistical analyses, investigate whether certain types of mothers are more likely to benefit from residential programs, and analyze multiple factors and interaction effects in assessing the effect of the child care service. Second, studies should measure not only LOS but also other outcomes such as AOD use, employment, child custody, and criminal behavior. Third, studies should use the same time intervals for follow-up measurements.

One double-edged methodological sword is the issue of random assignment of mothers to with-child care versus no-child care groups. On one hand, random assignment can ensure that the two groups share equivalent characteristics and would thus avoid selection bias as in the statement of Wobbie et al. (1997) that the longer LOS for the child-with-the-mother group could be attributable to the possibility that only the less-problematic mothers were allowed to have their child stay with them and that the less-problematic clients normally tend to comply with treatment, resulting in a longer LOS and a higher likelihood of completing the treatment. On the other hand, researchers may need to pay attention to the possible negative impact of the random assignment on women. A woman assigned to a no-child care group may feel disempowered because of her inability to choose to have her child stay with her in treatment, which in turn may negatively impact her recovery. The less-positive outcomes of the no-child care group may not be related to no child care per se but to a mother's feelings of being disempowered by her lack of options (Stevens & Patton, 1998).

3.4.1. Provision for case management 

The qualitative study of Howell and Chasnoff (1999) revealed that treatment providers perceived intensive case management to be critical in helping women recover, particularly after discharge from residential care. Clients, however, perceived case management as both positive and negative. Some appreciated the support whereas others believed that such support may be misused by clients (i.e., enabling). Although this study measured staff's and clients' perceptions regarding the relationship between case management and treatment outcome instead of the actual behaviors, it provides some theoretical foundation for the complexity of this relationship.

Five studies suggested that case management enhances retention or treatment outcomes. The one-group study of pregnant and postpartum substance-abusing women of Lanehart et al. (1996) indicated that intensive case management and aftercare support produced good posttreatment outcomes in the areas of substance abuse, employment, incarceration, and children's birth weight. Evenson et al. (1998) evaluated programs that offered wraparound and intensive case management for women with children. They found that women performed better during the treatment than prior to admission in the areas of substance abuse (e.g., 41% to 94% abstinent), parenting, functional level, and employment. The path analysis results of Laken and Ager (1996) showed a significant association between case management service and treatment retention for pregnant women. McLellan et al. (2003) showed that welfare-recipient women who participated in the CASAWORKS for Families program made significant improvements in the areas of substance abuse and social/family functioning at 6-month follow-up and employment at 12-month follow-up. The CASAWORKS for Families program emphasized interorganizational coordination so that women could receive comprehensive and concurrent services. Jansson et al. (2003) found that, as compared with mothers of drug-exposed infants who received a low-intervention (four or fewer visits within 2 years) case management, mothers who received a high-intervention (five or more outreach service visits) case management were more likely to use postpartum substance treatment, less likely to report recent drug use at 2-year follow-up, and more likely to have child custody rights.

3.4.2. One-stop shopping model 

McMurtrie et al. (1999) stated that comprehensive services provided onsite help women build a trusting relationship with the treatment team, encourage them to use a wide range of services, and prevent dropout. Onsite services also may promote mutual understanding and coordination among various professionals who often have different treatment philosophies and approaches. Both McMurtrie et al.'s nonrandomized three-group design and Volpicelli et al.'s (2000) randomized two-group design suggest a positive impact of the one-stop shopping model on treatment outcomes. McMurtrie et al. found that mothers who participated in the one-stop shopping treatment had a significantly lower percentage of low-birth-weight babies than mothers who attended two other local special prenatal clinics (23% vs. 35% and 47%).

Volpicelli et al. (2000) compared outcomes between a case management outpatient program and a psychosocially enhanced treatment program, in that both programs provided group therapy and onsite child care but the psychosocially enhanced treatment program offered onsite individual counseling and other services whereas the case management outpatient program referred clients offsite to other community sources for such services. The authors found that although clients in both groups made progress, those who attended the one-stop shopping program had higher retention and lower cocaine use at the 12-month follow-up.

To summarize, because there were only two studies reviewed for this issue, more future studies with a randomized controlled design are needed to enhance our confidence in these conclusions. It might also be interesting to compare four groups in a continuum: no case management, regular case management, case management that specifically emphasizes interagency coordination, and case management that provides onsite comprehensive services (e.g., McLellan et al., 2003). In addition, the outcome variables need to be standardized and include not only AOD use but also measures of parenting, depression/anxiety, criminal behavior, and employment.

3.5.1. Critique 

More large-scale studies should be conducted to investigate directly the relationship between a counselor's degree of genuine concern/supportive attitudes and client treatment outcomes. Because only two quantitative studies were identified and reviewed, more randomized two-group studies are needed to verify the function of individual counseling for women. Future studies are also needed to evaluate different types of counseling (individual vs. group) with different styles (supportive vs. confrontational), exploring if a supportive group approach can achieve the same effect as a supportive individual approach does or has a better effect than a confrontational individual approach. Finally, the outcome variables should also be extended to include depression, anxiety, criminal behavior, and employment, in addition to AOD use and retention.