Social support and education groups for single mothers: a randomized controlled trial of a community-based program

Methods

This study was approved by the Research Ethics Board of the Faculty of Health Sciences at McMaster University, Hamilton, Ont.

Single mothers with young children were recruited through advertisements in community flyers at various locations (e.g., family physicians' offices, libraries, shelters) in Hamilton, Ont. The advertisements asked mothers if they were “feeling alone” and “parenting on their own.” Interested mothers who contacted the research program coordinator were told about the study and were assessed for eligibility. Eligibile mothers were those who had at least one child 3–9 years old, were able to speak English, had no acute psychiatric crisis (e.g., psychotic or suicidal behaviour) or threat of violence (e.g., stalking by ex-partner) and gave informed written consent to participate in the trial. They were asked to complete questionnaires during a home visit and were then randomly assigned to the intervention or control group. Randomization was done in blocks of 4, with numbers generated by a random-numbers table and sealed in opaque envelopes. It was not possible to conceal the group allocation from the participating mothers.

Mothers assigned to the intervention arm of the trial were asked to participate in a 10-week program of group sessions (1.5 hours per week) that provided social support and education; their children attended an activity-based group program during the sessions. Each support group comprised 6–10 women and 2 leaders, and sessions were held in a community location (e.g., public housing project, church hall). A manual was used to guide the structure, content and implementation of the program as well as the training of the leaders. Content covered 2 thematic areas: child-related (e.g., child development and behaviour, behaviour management, school involvement, child welfare agencies) and maternal (e.g., social isolation, stress and coping, personal care and development, relationships, grief, economic disadvantage). There was no specified order for introducing content; experience with similar groups20^,22^,41 has shown that all content areas are covered over the course of the program. We chose this format because it responds to maternal needs by allowing participants to bring urgent concerns to the group in a timely fashion. Leaders used group processes to create a therapeutic milieu, drew on cognitive behavioural techniques and provided structured group counselling. In our study, we used a pool of 3 leaders; each had previous group training and experience, underwent specific training for this program, participated in weekly supervision and, throughout the study, were videotaped at random leading their groups.

Group sessions were held on weekdays either in the morning (9:30–11:00 am) or in the evening (6:00–7:30 pm). To enhance participation in the group sessions, mothers received weekly phone reminders as well as bus tickets or taxi fare to assist with transportation. In addition, mothers and children attending sessions were given snacks (day groups) or dinner (evening groups).

Mothers allocated to the control arm of the trial were given a standard list of community resources and the option to attend the program of group sessions at the end of follow-up period.

Assessment data were collected from all participants by interviewers working in pairs; at least one of the interviewers was blind to the mothers' group allocation. Data were collected during home visits at baseline (before the intervention period) and at 3 follow-up visits (after the intervention period and at 3 and 6 months). At each visit, mothers and children received gifts of appreciation for their participation (e.g., retail gift certificates, children's books, craft supplies).

A total of 117 mothers were eligible and agreed to participate in the study; 59 were randomly assigned to the intervention arm and 58 to the control arm (Fig. 1). One mother in the control arm was withdrawn from the study immediately after randomization because she did not return calls. The sample size was selected to provide adequate power (1 – β ≥ 0.80) to detect medium standard effects (0.50), allowing for 20% attrition and some clustering of effects within groups.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1: Flow of participants through the trial. R = randomization. *Mothers in the intervention arm attended 10 weekly group sessions (1.5 hours per week) during which they received social support and education; mothers in the control arm were given a standard list of community resources and the option to attend group sessions at the end of the follow-up period.

The concepts and outcome measures used in the study are described in an online appendix (available at www.cmaj.ca/cgi/content/full/173/12/1451/DC1). The outcome measures were maternal well-being (mood,42^,43 self-esteem44 and social support45) and parenting;46 the measures were chosen for their relevance to the program objectives and for the simplicity, brevity, acceptability and general applicability.

We estimated means and variances for selected variables and compared them between the intervention and control groups. In addition, we used linear regression analysis to test differences in outcome measures before and after the program; we adjusted for group differences by including “treated for nerves or nervous condition” and “wages or salaries as source of financial support” as covariates (participants who did not report whether they had been treated for nerves were coded as the modal value [yes]).

To assess the effect of the program over the follow-up period, we used MLwiN47 to model growth trajectories. In growth curve analysis, developmental outcomes are modelled as a function of time to establish individual-level trajectories, showing a starting point (baseline) and change per unit of time or growth (slope) for each person.48 This modelling made it possible to compare differences between the intervention and control groups in their baseline values and the rates of change. Because of the extended follow-up period and nonlinear pattern of growth (decelerating rates of change), we used the square root of time (in months) to model outcomes. The covariates “treated for nerves” and “wages or salaries as source of financial support” were included in the models at baseline.

Results

Between February 2000 and April 2003, the program was offered to 9 groups of single mothers. Among the 116 participants, assessment data were available for 87% (intervention 90% [53/59], control 84% [48/57]) after the intervention period (first follow-up assessment). At the second and third follow-up visits (which occurred at a mean of 13.7 and 20.2 months after baseline respectively), assessment data were available from 79% (intervention 86%, control 72%) and 72% (intervention 85%, control 58%) of the mothers respectively (Fig. 1).

Table 1 shows the baseline characteristics of the participants. Differences between the intervention and control groups were generally small except for reports of being treated for nerves in the last 6 months and wages or salaries as a source of financial support.

Table 2 shows the levels of functioning at each of the follow-up assessments. At baseline, there were only small differences between the 2 groups. Results before and after the intervention period were examined using regression analysis of functioning at the first follow-up assessment (n = 101; mean 7.0 [standard deviation 3.5] months from baseline to the first follow-up visit). At this follow-up assessment, mothers in the intervention group showed significant improvements in scores for mood (β = –6.55 [standard error (SE) 2.27]; p < 0.01; standardized effect = 0.55) and self-esteem (β = –1.48 [SE 0.72]; p < 0.05; standardized effect = 0.29) compared with the mothers in the control group. Loss to follow-up at this point was relatively small (15 [13%]) and showed only weak, nonsignificant associations with the participants' baseline functioning scores and sociodemographic characteristics.

The estimated growth curves showing changes in functioning over time and the differences between the intervention and control groups at baseline and at the third follow-up assessment (mean 20.2 months after baseline) are presented in an online table (available at www.cmaj.ca/cgi/content/full/173/12/1451/DC2). Both the intervention and the control groups showed improvements in all 4 measures of functioning, with no statistically significant differences between them. For example, at baseline, the adjusted mean mood scores in the control and intervention groups were 18.38 and 15.44 respectively, for a difference of 2.94. By the third follow-up visit, these estimates were 12.22 and 9.92 respectively, for a difference of 2.30. The rate of change in mood from baseline to the third follow-up visit was –1.45 in the control group and –1.30 in the intervention group, for a difference of 0.15. Fig. 2 illustrates the growth curve models for the 4 measures of functioning.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2: Growth curves for changes in depressive mood (A), self-esteem (B), social support (C) and parenting (D). CES-D = Center for Epidemiologic Studies Depression Scale. For details about the scales and score ranges see the online appendix at www.cmaj.ca/cgi/content/full/173/12/1451/DC1.

Sample losses at the third follow-up assessment differed between the 2 groups (intervention 15% [9/59] v. control 42% [24/57]). We investigated this difference using logistic regression analysis and modelled nonresponse at the third follow-up assessment as a binary outcome (1 = response, 0 = nonresponse). The predictors included in the model were the sociodemographic characteristics of the participants and their baseline functioning scores. The only predictor significant at p < 0.05 was group allocation (intervention v. control).

Interpretation

We evaluated a community-based program in which vulnerable single mothers of young children were randomly assigned to participate in group sessions offering social support and education or to receive a standard list of community resources. The participants were especially vulnerable in terms of income (household income fell below the poverty line49 in two-thirds of these families) and mood.

Over the short term (from baseline to the first follow-up assessment at the end of the intervention period), we found that the program significantly improved mood and increased self-esteem among the single mothers but had no reliable effect on levels of social support and parenting. Over the longer term, growth curve analyses indicated that program effects on mood and self-esteem attenuated, with no significant differences between the intervention and control groups at 18 months of follow-up. In general, the growth curve analyses showed improved outcomes among all of the participants, irrespective of their group allocation. This finding may indicate a generalized response among the mothers consistent with a natural descent from peak periods of stress and conflict observed at enlistment.

Our study had limitations. First, the participants came from a population of low-income single mothers living in an urban setting who responded to community advertisements. This method limits our inferences to similar families who would agree to participate in such a trial. Second, attrition at the final follow-up assessment was 28% overall, with greater losses in the control group than in the intervention group (42% v. 15%). Our analysis suggests that differences between groups in numbers lost to follow-up is an unlikely explanation for the absence of long-term program effects; however, without actual assessments, we cannot rule out this bias. Third, the measures selected to evaluate program effects may have been insensitive to change. However, all measures exhibited change over time, and some program effects (positive changes in depressed mood and self-esteem) were detected at program completion. The selection of some of the measures or methods of data collection may also have been flawed, and alternate measures or methods may be superior to describe program effects. Fourth, poor compliance or attendance at group sessions may have influenced the results. However, overall, only 8 (14%) of the 59 mothers in the intervention group attended less than half of the group sessions, which suggests that compliance was generally good. Fifth, there may have been variation in the fidelity of program delivery by the group leaders, although this was not seen to be a problem during weekly group supervision sessions. Finally, our study evaluated the effectiveness of the program in real conditions (as opposed to an efficacy study using ideal conditions). We made no specific attempt to prevent mothers in the intervention group from receiving additional interventions (cointerventions) that may have influenced the study outcomes, nor did we try to prevent mothers in the control group from receiving a similar intervention program (contamination). Our broad tracking of service use among the participants identified no appreciable differences between the intervention and control groups; therefore, it is unlikely that service use outside of the program influenced the results. Similarly, participation of the mothers' children in the activity groups may have augmented the intervention effects, but this is unlikely since child attendance varied greatly across the families.

In summary, the results from our randomized controlled trial indicate that participation in this community-based program of group sessions offering social support and education to single mothers accelerated improvements in depressive mood and self-esteem during the intervention phase but had no significant differential effect on levels of social support or parenting. In the longer term, significant differences in outcomes between the intervention and control groups were not evident. Since many single mothers have to cope with many acute and chronic stresses in their lives, it is possible that this intervention is insufficient to bring about enduring long-term effects.50 Our analysis may have underestimated some of the positive short-term effects of the program because of delays in collecting data owing to lost contact (e.g., moving, telephone disconnected) and the need to trace families. Nevertheless, it appears that focused, time-limited, group-based support programs, on their own, have limited potential to improve the quality of life of low-income single mothers over the longer term. Alternative approaches, such as more intensive nonfinancial programs and increased financial support, should be rigourously evaluated.