Document characteristics
Figure 1 presents a flow diagram outlining the document search and appraisal process. Throughout the search process, it became evident that relatively few COSAP programs existed that met the review criteria; however, multiple documents were found as sources of evidence to support each program. A total of 32 documents were retained for this review, spanning 7 different COSAP programs implemented in the USA, UK, Spain and Canada. Types of documents retained included outcome evaluation studies (n = 16), grey literature community evaluation reports (n = 9), qualitative studies (n = 3), book chapters (n = 3) and conference presentations (n = 1). Additional file 1 outlines the COSAP programs and related documents that were included in this review.
A wide range of outcomes was reported for these programs. The majority were proximal and measured shortly after program completion. We summarized these to reflect general categories: child behavioural changes (e.g. aggression, conduct), child emotional changes (e.g. anxious, depressive, loneliness, self-esteem), parenting (e.g. parenting skills, parent mental health), relapse prevention (reduction in parental substance use) and family cohesion (e.g. bonding, family communication, time spent together). Very few studies were longitudinal and measured reduction or prevention of child substance use later in life. There were also slight variations in terms of dosage, structural format, content, target population (e.g. parents enrolled in concurrent drug treatment, African-American families) and eligibility for program participation (e.g. abstinence during program, demonstrated program commitment, length of time in recovery).
The COSAP programs included in this review were classified by the research team according to their potential support for the two candidate program theories, based on the program descriptions found in supporting documents. For example, programs that promoted a disease-based conceptualization of addiction, emphasized the important of abstinence and implemented a primarily knowledge-based curriculum were categorized within the family disease model. Programs that used a skills-based curriculum aimed at reducing risk factors for substance use and enhancing protective factors within the family were categorized within the family prevention model. Programs that appeared to draw elements from both models were classified as hybrids (see Additional file 1). It is important to note that the purpose of this review was not to rank or compare programs in terms of relative merit or efficacy. Rather, main findings with respect to how and why these programs achieve outcomes are presented below.
Main findings: demi-regularities
Four demi-regularities were identified in this review as being fundamental in generating positive COSAP program outcomes. These are presented below along with key examples of the contexts, mechanisms and outcomes. Quotations are provided to further illustrate the findings; these represent single examples of multiple instances that were evident in the data.
-
1.
Creating opportunities for positive parent-child interactions
Programs that consistently provided opportunities for positive parent-child interactions were found to produce outcomes of improved family cohesion. Documents from a number of programs (e.g. Safe Haven, Strengthening Families Program (SFP), Moving Parents and Children Together (MPACT), Family Competence Program (FCP)) discussed that the program succeeded in bringing families together for shared time that would not have otherwise been spent. It was frequently noted that providing opportunities for parent-child interactions in an enjoyable and supportive environment (context) led to improvements in family cohesion (outcome).
We identified the mechanism of hopeful enjoyment through which this outcome is achieved. Providing multiple opportunities for positive parent-child interactions (context) during the program was found to foster a sense of joy and pleasure (mechanism) among family members and an increased sense of hope that the family unit could be restored (mechanism). A qualitative evaluation of MPACT program provides a useful example of this process:
I think the sheer fact that we went every week and we didn’t miss a week and we all did it together, just that alone I think … It made us feel good about ourselves ([29], p.3).
Our review documented multiple instances of programs that encouraged families to spend time together in a supportive and non-punitive environment. This allowed parents to develop empathy (mechanism) for their children, and in turn children were allowed a safe space (context) to express themselves to their parent during the program [30]. Having parents and children attend together (context) increased positive interactions (outcomes) and encouraged children to feel loved and appreciated by parents (mechanism) [27].
Another useful illustration of this demi-regularity is an example where desired program outcomes were not achieved. The Focus on Families (FOF) evaluation indicated that the program did not achieve desired outcomes for family cohesion [31–33]. In this case, program structure was such that children did not attend all sessions with parents, suggesting a lack of sufficient opportunities for positive parent-child interactions (context). We hypothesize that the mechanism of hopeful enjoyment did not have sufficient opportunity to fire in this case. Further, older children actually reported negative effects of parental involvement, suggesting that attempts by parents to increase parent-child interaction time were not only lacking in enjoyment but were in fact met with rejection. The authors of that paper hypothesized that older children who were accustomed to lack of supervision (context) perceived increased family time as an unwelcome intrusion [32]. This provides further evidence that the mechanism of hopeful enjoyment needs to be triggered in order for positive outcomes in family cohesion to occur, in the context of child age and prior experience with parental supervision. Programs that facilitate positive parent-child interactions can help families achieve a restored connection when it is developmentally appropriate for them to do so, as is more likely to be the case for younger children. For older children who are at an individuation developmental stage, attempts at eliciting hopeful enjoyment of family interactions may misfire and fail to yield positive outcomes. The accumulation of CMO evidence found in support of this demi-regularity suggests that positive parent-child interactions occurring within a safe and supportive environment, which can be actively fostered by the COSAP program, are instrumental in yielding increased family cohesion.
-
2.
Supportive peer-to-peer relationships
Environments that fostered supportive peer relationships among child participants and among parents were noted across many programs as being instrumental in achieving positive child psychosocial outcomes and to a lesser extent positive parenting outcomes. Evidence was found to support this process in a couple of ways. First, supportive peer-to-peer relationships between the child participants elicited mechanisms of trust and safety within the group as well as validation of experience. Improvements in child psychosocial functioning were consistently reported in these cases (e.g. [29, 34, 35]). Social isolation (context) is common among children living with parental substance use, and the mere fact of being placed in a supportive group of their peers (context) allowed for feelings of safety (mechanism) to emerge and enabled the sharing of experiences (outcome). For example, a qualitative evaluation of the Betty Ford Children’s Program, which was described as purposefully grouping children according to similar age (context), demonstrates this finding:
I have a lot of, you know, really close friends but they … can’t relate to my situation … you come here and you meet friends who are just like you ([36], p. 389)
Conversely, negative peer-bonding between child participants was noted in one COSAP program evaluation [37], further suggesting that the trust and safety mechanism is instrumental in achieving positive child behavioural outcomes. This finding would suggest that attention to participant characteristics such as similar age and lived experience, when forming program peer groups, is warranted as it will facilitate positive peer-bonding and will more likely trigger perceptions of trust and safety.
Second, we noted that parent participants who were placed in supportive groups with other peers also exhibited positive outcomes through a mechanism of validation of experience. Parents struggling with parenting at the same time as recovering from substance abuse (context) were reported to have found the group dynamics and peer relationships fostered with other parents during the COSAP programs to be beneficial. The process of validation (mechanism) for these parents can be described as the normalization of experience and sharing of mutual struggles among supportive peers. Further, at least one program evaluation discussed the possibility that the strong bonds formed between participants were a motivator to continue attending sessions (outcome) [38]. It is possible that the creation of supportive peer relationships was a contributing factor to engagement and program commitment, also leading to improvements in parenting skills and child psychosocial outcomes.
Both adults and children appeared to benefit greatly from meeting others and making friends, specifically with people who lived in similar circumstances. For many, this seemed to bring mutual understanding as families’ experiences were normalized and they realized that they were not alone with their struggles. ([29], p. 4)
Thus, according to this demi-regularity, programs that address the social isolation common to substance using families through strategies such as appropriate peer grouping are more likely to trigger mechanisms of validation and trust.
-
3.
(Addiction) Knowledge is power
Programs that specifically emphasized knowledge about addiction and education around the impact that substance abuse has on children and families were found to yield improvements in parenting and child psychosocial outcomes. The following key mechanisms were identified within this demi-regularity: parental recognition and responsibility and children relinquishing responsibility for parental addiction.
In the Betty Ford program, for example, knowledge was described as “opening the door for them” ([36], p.390) and that simply knowing the truth about their parent’s addiction was helpful. Further, the provision of knowledge (context) allowed children to realize that they were not responsible or at fault for their parent’s addiction (mechanism) [29]. Shame and secrecy (context) are hallmarks of family addiction, according to the family disease model. The mechanism of relinquishing responsibility is triggered when children are provided with information about parental addiction that had been previously withheld or downplayed (context). The process of relinquishing responsibility appears to be a key element of the knowledge provision demi-regularity and may set the stage for subsequent improvements in child emotional and behavioural outcomes to occur, either directly through this demi-regularity or through one of the others identified in this review.
Further, parents who attend these programs are also provided with knowledge about how their alcohol or drug use has impacted their children and the family unit. The parental recognition and responsibility mechanism is triggered under these circumstances, whereby parents are able to recognize the impact of their behaviours and take responsibility for how it has affected their children. For example, evaluations of the Celebrating Families! program and MPACT program both documented instances where parents gained new understanding of the impact that alcoholism has on the family (mechanism) [39], a realization of not playing the appropriate role as a parent (mechanism) [29] and “the shock that some of the adults conveyed as they began to take in the effects of their lifestyle on their children” ([38], p.18). These programs reported positive outcomes with respect to parenting, such as improved positive parenting and parenting efficacy, by using strategies of knowledge provision to challenge the shame, secrecy and lack of recognition inherent to substance-involved families (context) [38].
Engaging hard to reach or marginalized families
For certain programs where the participating families were recruited from particularly marginalized populations (e.g. poverty, cultural minority), engagement emerged as an important intermediary factor that was necessary in order for outcomes to be achieved. Interestingly, the process of engagement is not explicit within the family prevention model nor the family disease model, perhaps because engagement is assumed to occur once recruitment is established or that engagement is equated with program attendance. As such, engagement did not initially emerge within either candidate program theory. However, as data abstraction progressed, we noted that only those programs classified in this review as aligning with the family prevention model were attuned to this issue. Information on recruitment best-practices exists within family-based intervention literature (e.g. [40]); however, engagement is rarely distinguished from attendance. For the purpose of the present realist review, we conceptualize engagement more broadly than mere program attendance; it refers also to acceptance and uptake of materials. Given the nature of COSAP intervention, engagement is not limited to the client-staff dyad or therapeutic alliance. COSAP programs are group delivery format where clients must engage with each other, the program content and with the program staff. Other realist reviews have identified engagement as an important feature of program success (e.g. [18]). In the present realist review, two instances were identified where successful program engagement yielded positive outcomes: (a) responsiveness to client socio-economic needs and (b) matching to client lived experience. These are discussed below.
-
a)
Responsiveness to client socio-economic status (SES) needs. Programs that are responsive to the SES realities and needs of their clients (context) will encourage program engagement (outcome) by fostering a sense of trust (mechanism) among families who are typically marginalized (context). For example, SFP and the Safe Haven program both went to extensive lengths to encourage and incentivize participants, such as providing meals, transportation, childcare, basic necessities and vouchers redeemable for family activities. Families participating in these programs were contextually characterized as low income, low education, having unstable housing, child welfare involvement and, unsurprisingly, often mistrustful of service providers. The key mechanism here is the sense of trust and acceptance that is developed on the part of the client in response to these staff/program efforts, as evidenced here: “Basic material supports provide a message to needy families that the staff really care about them” ([41], p. 260). In the case of the Safe Haven program, this process was described thusly, as a result of basic necessity provisions:
The Safe Haven staff began to know and understand the unique circumstances of each of the participating families. This seemed to increase staff empathy for the families. The families, in turn, reported to the process evaluator that they felt the staff “cared about them” ([42], p. 46).
In the example above, the program’s responsiveness to client SES needs (context) affected both the staff’s ability to engage with the families and vice versa (outcome), through a process of trust building (mechanism).
-
b)
Matching to client lived experience. Programs that took appropriate steps to match staff and client lived experience of family addiction and/or cultural background (context) were more successful in engaging clients (outcome) by fostering trust and personal identification with the program materials (mechanism). This was true of programs such as Safe Haven where extensive efforts to make the program and staff culturally consistent with an African-American worldview led to increases in client acceptance and engagement: “They put it in a way Black people can understand” ([43], p. 233). This program also specifically recruited staff who were themselves also in recovery from substance use (context). Clients noted this shared life experience as being beneficial: “They share of their experiences… this helps” ([43], p. 233). The ability of clients to identify with program content and program staff facilitated trust (mechanism) in program and led to increased engagement (outcome). We hypothesize that once engagement is established, other program outcomes will be more easily facilitated via the demi-regularities explored above, creating a series of CMO chains.
Note that our review revealed that cultural adaptation, such as the one described in the Safe Haven program, did not always lead to better outcomes. For example, a number of SFP formats have been culturally adapted for a variety of different ethnic groups in the USA. However, comparisons between generic SFP and culturally adapted versions yielded no improvements in positive outcomes, beyond an increase in retention [44]. Based on the findings above, it is hypothesized that Safe Haven was successful in this regard because the appropriate matching of staff to client lived experience of culture trigged mechanisms of trust and client identification with program materials. Our review did not find evidence of these mechanisms being fired in other culturally adapted programs.
Alignment with candidate program theories
As part of the analysis process, the four demi-regularities described above were examined with respect to their alignment with the candidate program theories. Programs classified within the family disease model were supported with evidence from the “knowledge” and “supportive peer relationships” demi-regularities. This suggests that the provision of knowledge that is specific to family addiction (context) facilitates children in relinquishing the responsibility for their parent’s addiction (mechanism) and enables parents to recognize and take responsibility for the impact of their addiction on their family (mechanism), leading to improved coping and reduced family stress (outcome). Further, social support provided to families within the context of a supportive peer relationship (context) serves to validate the experiences of families living with addiction (mechanism), leading to improved coping and parenting behaviours (outcomes). The family disease model asserts that defining addiction as a disease is fundamental to the process of relieving oneself from the guilt and responsibility for a family member’s addiction [45]. The “knowledge” demi-regularity supports this theoretical assertion. Additionally, the importance of social support and interpersonal bonding are viewed as essential components of Al-Anon and other support groups that exist within the family disease addiction treatment landscape [45]. While COSAP programs extend beyond the scope of a support group, the “peer relationships” demi-regularity accounts for these findings within family disease model programs.
Programs developed from the family prevention model were evidenced with the “positive parent-child interactions”, “supportive peer relationships” and the “engagement” demi-regularities. This suggests that opportunities for positive parent-child interactions within the context of a safe and supportive environment encourage families to seek joy in spending time together (mechanism) and find hope in the restoration of the family unit (mechanism), ultimately leading to improved family cohesion (outcome). Family prevention theory argues that involving parents in the promotion of healthy child functioning will reduce risk factors and enhance strength and protective factors [10]. The importance of social support in coping with family addiction is also noted in some family prevention literature [46]. The fact that both the “positive parent-child interaction” and “peer support” demi-regularities were found to align with the family prevention model adds evidence for this theory.
Engagement was found to be present within programs originating in family prevention only. Despite the fact that engagement has not previously been noted within COSAP program literature, it has been validated elsewhere. Findings from Jackson and colleagues’ [18] realist review of methadone treatment programs emphasized the importance of client engagement, specifically within the contexts of client-centred treatment, attention to client SES conditions and positive therapeutic relationships. From a broader perspective, other health care fields such as nursing have also emphasized the importance of patient engagement. One particularly useful comparison within the nursing literature is the link between treatment preference, patient engagement and health outcomes [47, 48]. Included in treatment preference is the suitability of the treatment to individual lifestyle [47, 49]. Suitability to personal style could be akin to appropriate matching of client lived experience, as was found in the present study. This alignment was not previously included within the family prevention candidate program theory; as such we conclude this theory should be refined in order to account this finding.
It is notable that two programs included in this review were classified as hybrids, as they drew upon elements common to both candidate program theories. Hybrid programs were evidenced from a combination of all demi-regularities to varying degrees, with the exception of engagement. This is an interesting finding in and of itself but also supports the case for using realist methodology in evaluation inquiry. It would suggest that in practice, program implementation is complex. The MPACT program, for example, was reportedly influenced by the SFP model [38] and was then adapted to the UK context. MPACT maintained policy objectives of improving parent-child communication, parenting skills and child wellbeing [38], which is consistent with SFP and other family prevention model programs. However, our review of the MPACT documents revealed a significant emphasis on understanding the impact of parental addiction on children and families, communicating about addiction and empowering children to take responsibility for their own safety and wellbeing, the latter examples being consistent with family disease model program objectives. The research team concluded that MPACT was best classified as a hybrid, as it appeared to successfully integrate elements from both candidate program theories. The implications of the alignment of our findings within the candidate program theories are further discussed in the next section.