A Meta-analytic Study of Couple Interventions During the Transition to Parenthood
Abstract
The present meta-analysis integrates results of 21 controlled couple-focused interventions with expectant and new parents. The interventions had, on average, small effects on couple communication (d = .28 standard deviation units) and psychological well-being (d = .21), as well as very small effects on couple adjustment (d = .09). Stronger effects emerged if the intervention included more than five sessions, included an antenatal and postnatal component, and was led by professionals rather than semiprofessionals.
The transition to parenthood is one of the most challenging family transitions (Huston & Holmes, 2004). During this transition, many parents show a decline in couple adjustment and positive couple communication (Cowan & Cowan, 2000; Doss, Rhoades, Stanley, & Markman, 2009). In addition, psychological distress is quite common in young parents. For example, studies show that up to 80% of new mothers experience mild symptoms of depression in the first weeks after giving birth, and 10 to 30% of new mothers develop clinical depression (Miller, Pallant, & Negri, 2006). Psychological distress may also have a negative effect on parenting behavior, such as sensitivity, investment in the child, overall quality of parenting, and cooperation between parents (Foster, Garber, & Durlak, 2008).
Starting in the 1970s, interventions with expectant and new parents have been developed and evaluated (Liebenberg, 1973). The goal of couple-focused interventions is to strengthen marital relationships by preparing couples for the difficulties inherent during the transition to parenthood and promoting relationship skills (e.g., communication, conflict management), mutual support, and realistic expectations about parenthood. Two narrative reviews of couple-focused interventions in the transition to parenthood reported some positive effects on couple adjustment, couple communication, and psychological well-being, but not all of the included studies were successful (Cowan & Cowan, 1995; Petch & Halford, 2008). Available meta-analytic studies on couple education in general have found significant improvements in couple adjustment and satisfaction of d = .15 to d = .54 standard deviation units (Giblin, Sprenkle, & Sheehan, 1985; Hawkins, Blanchard, Baldwin, & Fawcett, 2008). Unfortunately, these authors did not report separate results for included studies with expectant and new parents. Moreover, they failed to include further outcome variables, such as parental mental health and parenting, which may also be affected by these interventions. Thus, the goal of this study was to meta-analyze the effects of interventions that were targeted at preventing a decline of couple adjustment during the transition to parenthood on a broader range of outcome variables.
We focused on interventions that start during pregnancy or in the first six months after birth because (a) the most recent narrative review on early interventions used the same inclusion criterion (Petch & Halford, 2008), (b) prevention should ideally start before problems develop, (c) couple adjustment before birth predicts postnatal adjustment (Karney & Bradbury, 1997), (d) there are many new challenges in the transition to parenthood (Cowan & Cowan, 2000), (e) sudden deterioration of couple adjustment have been observed following birth (Doss et al., 2009), and (f) there is a high risk for postnatal depression in the first months after giving birth to a child (Miller et al., 2006).
In our first research question, we considered the strength of average effects of early interventions on couple adjustment, couple stability, and couple communication (e.g., amount of conflict, problem-solving). We also analyzed the effects on parenting, parental stress, and psychological well-being, as the intervention studies often included these additional outcome measures. On the basis of the available narrative reviews, we expected to find significant positive intervention effects on couple adjustment and couple communication. In addition, because couple adjustment may affect psychological well-being (Fincham, Beach, Harold, & Osborne, 1997), and because the participation in interventions may be a source of positive feelings, we expected to find a positive effect on psychological well-being. As positive couple relations may also promote effective parenting (e.g., coparenting) and reduce parental stress, we also expected to find positive intervention effects on these outcomes. Because of the lack of available studies, we could not include child development as a further outcome variable.
Moderating Effects of Study Characteristics
The second research question focused on moderator effects of study characteristics on the effect sizes. Available narrative reviews of interventions with expectant and new parents could not test whether study characteristics such as the length of the intervention or the qualification of the interventionist would moderate the effect size (Cowan & Cowan, 1995; Petch & Halford, 2008). This information is, however, very important for selecting and designing interventions that may produce above-average effects and for identifying couples that may benefit most from these interventions. We selected variables that were assessed in the included studies and that have been found to moderate the effect sizes in previous meta-analyses on interventions with couples or expectant and new parents.
Intensity of the intervention.
Given the many new demands and stressors in the transition to parenthood, a sufficient number of sessions may be needed for promoting positive change. In fact, the meta-analysis of Giblin et al. (1985) found longer interventions to have stronger effects. In addition, Hawkins, Blanchard, et al. (2008) found that moderate-dosage programs produced larger effect sizes than low-dosage programs did.
Onset of the intervention.
Because couple adjustment before the birth of a child is an important predictor of later adjustment (Karney & Bradbury, 1997), we expected that interventions with a prenatal component would be more effective than interventions that start after birth.
Qualification of the interventionist.
A higher qualification of the interventionist may be associated with stronger intervention effects, because clinical skills and persuasion power may be needed for being effective. Such an association has been shown for parenting education (Pinquart & Teubert, in press).
Target person.
No clear hypothesis can be stated on whether women may benefit more from interventions than men. Hawkins, Blanchard, et al. (2008) found similar intervention effects on both genders, but they did not focus on the transition to parenthood, which poses more stressors for mothers than for fathers (Bianchi, Milkie, Sayer, & Robinson, 2000).
Self-report versus observation.
Some meta-analyses found that effect sizes for observational measures of communication were significantly larger than for self-report measures (Butler & Wampler, 1999). This may indicate that attitudinal change lags behind behavioral change or that couples are more likely to show the trained behavior in the laboratory than during their daily interactions.
Randomization.
On the one hand, nonrandomized studies may be more likely to assign highly distressed couples to the intervention condition (because they would not be willing to participate in the control condition; Bryan, 2002). In this case, nonrandomized studies may overestimate intervention effects and show stronger effects than randomized studies. On the other hand, it might be more difficult to affect the outcome variables of couples with severe problems, which would lead to the reverse result. The meta-analysis of Hawkins, Blanchard, et al. (2008) reported lower effects in nonrandomized studies, and we wanted to replicate this finding in our meta-analysis with new and expectant parents.
Publication status.
As insignificant results may be less likely to be published (the file drawer problem; Lipsey & Wilson, 2001), unpublished studies may report smaller effects than published studies. Hawkins, Blanchard, et al. (2008) found support for this suggestion, but only in one of eight comparisons.
Additional intervention goals.
As some interventions added a parenting education component, we expected these interventions to show stronger effects on parenting than pure couple-focused interventions.
Method
Sample
Studies that had been published or were in press by the end of 2009 were identified by systematic search in electronic databases (PsychInfo, Medline, Psyndex, ISI Web of Knowledge) and cross-referencing. Search terms used were intervention or prevention or psycho-education or home visiting and transition to parenthood or parents. We included intervention studies in the present meta-analysis if (a) they used a control group receiving no intervention or only a minimal intervention, (b) they had a couple relationship component, such as improving communication skills or increasing support between new mothers and fathers, (c) they started during pregnancy or in the first six months after a childbirth, and (d) effect sizes were reported or could be computed from the available information. In addition, if more than one publication was available on an individual intervention study, we included these papers in our analysis, but only if they reported results on different outcome variables. Duplicate results were, however, omitted. If interim results during the intervention and final results at the end of intervention were reported, we coded only the final results.
We excluded trials of programs that were not targeted at improving the couple relationship (e.g., programs exclusively focused on parenting or child development) or did not provide sufficient information for computing effect sizes. We did not exclude nonrandomized studies because (a) only a small number of randomized studies were available and (b) we wanted to test whether the effect sizes would differ due to randomization.
The search identified 27 reports. Three reports had to be excluded from the current meta-analysis because they did not provide enough information for computing effect sizes or duplicated results that had already been reported in an included paper.
Coding of Studies
We entered the following variables: year of publication; the number of participants in the intervention and in the control condition; sample characteristics (mean age, percentage of participating mothers); randomization (1 = yes, 0 = no), publication status (1 = published, 0 = unpublished), intervention goals (pure couple relationship promotion, couple relationship promotion + parenting education); kind of prevention (1 = universal, 2 = selective); number of sessions; onset of intervention (1 = prenatal, 2 = postnatal, 3 = both); professional training of the person delivering the intervention (1 = professionals, such as psychologists, social workers, and family therapists, 0 = semiprofessionals/lay persons); the measurement of the outcome variables; change in outcome variables; and time interval between the end of interventions and the follow-up (if available).
Computation and Testing of Effect Sizes
Calculations for the meta-analysis were performed in five steps, using random-effects models and iterative maximum likelihood estimations (Lipsey & Wilson, 2001).
- 1
We computed effect sizes d for each study as the difference in the posttreatment measure between the intervention condition and control condition divided by the pooled standard deviation (SD). Similar effect sizes were computed for follow-ups. Variables were coded so that scores of d > 0 indicate improvement. Outliers that were more than 2 SD from the mean effect size were recoded to the value at 2 SD (Lipsey & Wilson, 2001).
- 2
Effect size estimates were adjusted for bias due to differences in pretests between intervention and control condition and due to overestimation of the population effect size.
- 3
Weighted mean effect sizes were computed. The significance of the mean effect size was tested by dividing the weighted mean effect size by the estimated standard error of the mean. Then, confidence intervals (CI) that include 95% of the effects were computed.
- 4
An analog to the analysis of variance and weighted regression analysis were applied for testing whether the effect sizes would differ between conditions, such as randomized versus nonrandomized studies. A significant Q score indicates that the size of the effects differs between studies.
- 5
To interpret the practical significance of results, we used Cohen's criteria (Cohen, 1992) and the Binomial Effect Size Display (BESD; Lipsey & Wilson, 2001). The BESD is based on a median split of treatment and control group: If the numbers of participants of the intervention group and control group are of equal size and the participants with above- and below-average improvements are identified with a median split, then the percentage of participants with above-median improvements in the intervention condition and in the control condition is estimated.
Measures
Couple adjustment.
This outcome was assessed with the Marital Adjustment Scale (Spanier, 1976; six studies), the Marital Adjustment Test (Locke & Wallace, 1959; three studies), and other measures (6 studies).
Couple communication.
Ten different measures were used to assess partner communication, such as questionnaires (eight studies; e.g., the Partner Awareness Scale by Christensen & Schenk, 1991) and coding of observational data (two studies, four samples).
Relationship stability.
It was assessed by the frequencies of divorces or separations (three studies) and with the Marital Instability Index (Booth, Johnson, & Edwards, 1983; one study).
Parenting.
The Nursing Child Assessment Teaching Scale (Barnard, 1978; one study) and six other measures were used for assessment of parenting.
Parental stress.
This variable was assessed with the Parenting Stress Index (Abidin, 1983; one study) and related scales (two studies).
Psychological well-being.
Parental psychological well-being was measured with the State Trait Anxiety Inventory (Spielberger et al., 1983; two studies), the Symptom Checklist SCL-90 (Derogatis, Lipman, & Covi, 1973; two studies), and other scales (eight studies).
Results
The included 24 papers describe the results of 21 different interventions. The papers by Hawkins, Lovejoy, Holmes, Blanchard, and Fawcett (2008) and Lovejoy (2004), those by Bryan (1995, 2000, 2002), as well as those by Cowan and Cowan (1987) and Schulz, Cowan, and Cowan (2006) provided data from the same intervention but different outcome variables, and the paper by Hawkins, Fawcett, Carroll, and Gilliland (2006) described effects of two different interventions. Outcome data were available for 1,230 parents from the intervention condition and 1,109 from the control condition. Fourteen studies reported data from mothers and fathers, six studies from mothers only, and one study from fathers only. All available studies focused on heterosexual couples. Eighteen interventions provided universal prevention of couple adjustment problems, and three studies provided selective prevention for parents at risk. In addition, seven interventions were delivered before birth, seven after birth, and eight included both a before-birth and an after-birth component. The mean number of sessions was 11.4 (SD = 14.8, range 1–82). Sixteen studies used an experimental design (randomization), and five studies were quasi-experimental. An overview of the included studies is given in Table 1. The parents were, on average, 26.8 years old (SD = 2.7); 79% were married, 80% had a high school degree, and 23% were members of ethnic minorities. About 98% of the mothers were expecting or had given birth to their first child.
| Topics | Effective NIG | Study Design/ Components | Number of Sessions | Outcomes | |
|---|---|---|---|---|---|
| Black-Olien (1993) | Prevention of marital breakdown after childbirth | 63 m | r, a | 7 | Psychological well-being |
| Bryan (1995, 2000, 2002) | Reflecting couple changes in meaning/identity, roles, and relationship/interaction | 35 c | n, a | 3 | Parenting, couple communication |
| Bolte, Ziegenhain, and Küster (2009) | Relationship skills, parenting training, stress reduction | 37 c | n, a, p | 10 | Psychological well-being |
| Carpenter (1995) | Enhancing couple communication and problem-solving skills | 16 c | r, a , p | 4 | Couple adjustment, communication, relationship stability |
| Cowan and Cowan (1992); Schulz et al. (2006) | Parenting and couple, relationship information, group, support | 23 c | r, a | 24 | Couple adjustment, couple separation |
| Doherty, Erickson, and LaRossa (2006) | Parenting and couple relationship information, group discussion, skill-training | 65 f | r, a, p | 8 | Parenting |
| Fergusson, Grant, Horwood, and Ridder (2006) | Parenting education, supporting parental health, encourage stable partnership | 207 m | r, p | 24 | Partner violence, partnership stability, parenting, psychological well-being, child development |
| Gjerdingen and Center (2002) | Promoting positive couple relationship, household work planning | 69 c | r, a | 2 | Couple adjustment, couple communication, psychological well-being |
| Halford, Petch, and Creedy (2010) | Relationship and parenting skill-training, information and support | 32 c | r, a, p | 6 | Couple adjustment, psychological well-being, parenting |
| Hawkins et al. (2006), Instructor-guided | Couple relationship education | 51 c | r, p | 5 | Couple adjustment, parenting |
| Hawkins et al. (2006), Self-guided | Couple relationship education | 50 c | r, p | 1 (+bibliotherapy) | Couple adjustment, parenting |
| Heinicke, Fineman, Ponce, and Guthrie (2001) | Parenting education, promotion of partner support | 31 m, 27 f | r, a, p | 82 | Partner support, relationship stability, parenting, child development, psychological well-being |
| Kermeen (1995) | Promotion of couple adjustment | 63 c | r, a | 7 | Couple adjustment |
| Liebenberg (1973) | Counselling, clarifying feelings about marriage, coping with stress | 16 m | r, a | 15 | Couple adjustment, parenting |
| Lovejoy (2004); Hawkins, Lovejoy, et al. (2008) | Couple relationship, parenting | 39 c | n, p | 1 (+bibliotherapy) | Couple adjustment, father involvement in parenting |
| Marcenko, Spence, and Samost (1996) | Promotion of positive couple relationship, health education, parenting education | 125 m | r, a, p | 35 | Psychological well-being, support |
| Markman and Kardushin (1986) | Learning to tolerate pain, husband support for pregnant wife | 39 c | n, a | 7 | Couple adjustment, psychological well-being |
| Matthey, Kavanagh, Howie, Barnett, and Charles (2004) | Promotion of couple adjustment | 50 m | r, p | 1 | Partner awareness psychological well-being, social support, parenting competence |
| Midmer, Wilson, and Cummings (1995) | Promotion of mental health, couple adjustment, and parenting | 41 c | r, p | 2 | Couple adjustment, psychological well-being |
| Reichle and Franiek (2008) | Relationship skills and parenting training | 57 c | n, p | 10 | Couple adjustment |
| Shapiro and Gottman (2005) | Promotion of couple adjustment and parenting | 18 c | r, a, p | 2 | Couple adjustment, couple communication, psychological well-being |
- Note: r = randomization; n = nonrandomized (quasi-experimental) trial; a = antenatal sessions(s); p = postnatal session(s); c = couple; m = mothers; f = fathers; NIG = effective number of participants in intervention condition.
According to Cohen's criteria, effect sizes of d = .2 are interpreted as small, of d = .5 as medium, and of d = .8 as large (Cohen, 1992). We first analyzed intervention effects at posttest. Interventions had significant small effects on couple communication and well-being as well as significant very small effects on couple adjustment and parenting (Table 2). According to the BESD, 56.9% of couples in the intervention condition and 43.1% of those in the control condition would show above-median improvement in couple communication. Similarly, 52.3% of the participants in the intervention and 47.7% of the control group members would show above-median improvements of couple adjustment. The small effect on relation stability and the very small effect on parenting stress did not reach the level of statistical significance (Table 2).
| k | d | 95% CI | Z | Q | ||
|---|---|---|---|---|---|---|
| Effects at the end of the intervention | ||||||
| Couple variables | ||||||
| Couple adjustment | 23 | .09 | .00 | .18 | 1.98* | 18.83 |
| Couple communication | 19 | .28 | .04 | .52 | 2.27* | 42.47*** |
| Relationship stability | 4 | .25 | −.17 | .67 | 1.16 | 9.35* |
| Parenting variables | ||||||
| Parenting | 11 | .19 | .05 | .33 | 2.73** | 13.86 |
| Parenting stress | 4 | .13 | −.10 | .37 | 1.11 | 1.71 |
| Psychological well-being | 15 | .21 | .08 | .34 | 3.10** | 23.85* |
| Follow-up effects | ||||||
| Couple adjustment | 11 | .12 | −.06 | .30 | 1.34 | 16.01 |
| Couple communication | 11 | .29 | −.01 | .59 | 1.90 | 39.68*** |
| Relationship stability | 1 | .81 | .09 | 1.53 | 3.49*** | – |
| Parenting variables | ||||||
| Parenting | 4 | .12 | −.07 | .30 | 1.23 | 2.84 |
| Parenting stress | – | |||||
| Parental mental health | 5 | .81 | .20 | 1.41 | 2.62** | 39.95*** |
- Note: k = number of samples. Positive d scores indicate desirable changes (improvement). 95% CI = 95% confidence interval of the effect size; t = test of significance of the effect size; Q = test of homogeneity of the effect size (significant effects indicate heterogeneity).
- * p < .05; **p < .01; ***p < .001.
Eleven studies provided follow-up data, with an average time interval of 12.2 months. Significant effects at follow-up were only found for relationship stability and parental mental health. Both effects were interpreted as large. The observed effect on relationship stability, however, should be handled with care as it is only based on one study. One effect size estimation was slightly larger before applying the correction for outliers (subjective well-being: d = .22 vs. d = .21), but the correction procedure did not change the levels of statistical significance.
In the second research questions, we analyzed moderating effects of study characteristics. We focused on couple adjustment and couple communication because the largest numbers of effect sizes were available for these outcomes. As shown in Table 3, effects on couple adjustment varied with length of intervention, showing larger effects of interventions that included more than five sessions. According to the BESD, 57% of the members of longer interventions would show above-average improvements of couple adjustment as compared to 43% of the control group members.
| Couple Adjustment | Couple Communication | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| k | d | 95% CI | Z | Q | k | d | 95%-C.I. | Z | Q | |||
| Number of sessions | 7.06* | 1.18 | ||||||||||
| 1–5 | 15 | .01 | −.10 | −.12 | .19 | 8.07 | 14 | .27 | −.02 | .55 | 1.84 | 22.58* |
| ≥6 | 8 | .28 | .12 | .45 | 3.30*** | 3.72 | 5 | .32 | −.15 | .78 | 1.34 | 1.76 |
| Timing of the intervention | 0.56 | 35.12*** | ||||||||||
| Prenatal | 13 | .06 | −.05 | .17 | 1.10 | 10.27 | 9 | .02 | −.10 | .15 | 0.34 | 2.70 |
| Mixed | 6 | .16 | −.08 | .40 | 1.28 | 5.77 | 7 | .75 | .52 | .98 | 6.46*** | 34.28*** |
| Postnatal | 3 | .04 | −.22 | .31 | 0.31 | 0.24 | 3 | −.02 | −.19 | .15 | −0.25 | 0.36 |
| Qualification of the interventionist | 4.13* | 3.84* | ||||||||||
| Paraprofessional | 9 | −.01 | −.16 | .14 | −0.08 | 1.71 | 8 | .06 | −.31 | .42 | 0.30 | 0.16 |
| Professional | 11 | .23 | .08 | .38 | 2.62** | 12.36 | 9 | .56 | .21 | .90 | 3.16** | 20.54** |
| Target person | 4.97 | 1.35 | ||||||||||
| Mothers | 10 | .12 | −.02 | .27 | 1.73 | 6.43 | 10 | .32 | .00 | .64 | 1.97* | 13.04 |
| Fathers | 8 | −.05 | −.21 | .11 | −0.59 | 2.64 | 7 | .33 | −.06 | .72 | 1.66 | 11.54 |
| Mixed samples | 5 | .22 | .03 | .40 | 2.34* | 4.79 | 2 | −.11 | −.81 | .58 | −0.31 | 0.10 |
| Randomization | 0.05 | 1.37 | ||||||||||
| Yes | 17 | .10 | −.01 | .20 | 1.83 | 19.59 | 16 | .34 | .08 | .60 | 2.60** | 24.29 |
| No | 6 | .07 | −.11 | .25 | 0.73 | 2.01 | 3 | −.04 | −.61 | .54 | −0.13 | 0.20 |
| Publication status | 0.15 | 1.38 | ||||||||||
| Unpublished | 8 | .12 | −.04 | .27 | 1.46 | 15.63 | 5 | .04 | −.41 | .50 | 0.19 | 0.40 |
| Published | 15 | .08 | −.04 | .19 | 1.35 | 13.93 | 14 | .36 | .09 | .64 | 2.59** | 23.79* |
| Assessment of dependent variable | 33.60*** | |||||||||||
| Questionnaires | 23 | .09 | .00 | .18 | 1.98* | 18.83 | 15 | .04 | −.07 | .15 | 0.74 | 9.77 |
| Observational data | 0 | 4 | 1.02 | .71 | 1.34 | 6.36*** | 24.71*** | |||||
- Note: * p < .05; **p < .01; ***p < .001.
With regard to the inclusion of prenatal and postnatal components, we found that interventions that included both components had stronger effects on couple communication than interventions that were only delivered during or after pregnancy. In addition, intervention effects on couple adjustment and couple communication varied by the qualification of the interventionists. Our analysis showed that only interventions led by professionals (e.g., clinical psychologists, social workers) had significant effects on couple adjustment and couple communication.
When comparing mothers and fathers, we found similar effect sizes. Similarly, intervention effects did not vary between randomized and nonrandomized studies. In addition, unpublished studies did not differ in their effect size from published studies, thus indicating that our results are probably not influenced by a file drawer problem (Lipsey & Wilson, 2001). Furthermore, stronger effects on couple communication were found in studies with observational measures than in studies with communication questionnaires.
We also tested whether some of the assessed mediators would be correlated, thus indicating the need for multivariate testing. Only one significant association emerged: Randomized trials were more likely to be published (Φ = .39, p < .05), but randomization and publication status did not moderate the size of the observed intervention effects (see Table 3).
Finally, we checked whether interventions that included a parenting component would show stronger improvements of parenting than interventions that had no such component. Because interventions with an additional parenting component lasted longer than other interventions (r = .45, p < .05) we controlled for this confounder. We found that studies with a parenting education component had stronger effects on parenting than other studies (B = .38, β = .81, p < .01). In fact, only the former studies reported an improvement in parenting (k = 9 studies, d = .26, 95% CI = .11 to .42, Z = 3.29, p < .001, vs. k = 2, d = −.04, 95% CI = .32 to .24, Z = −.28, ns).
Discussion
This study extends beyond narrative reports and meta-analyses in that it provides the first meta-analysis on the effects of couple-focused interventions with expecting and new parents. It quantified average intervention effects and identified variables that moderate the effect size. As important findings, the present meta-analysis showed that pure couple-focused interventions do not affect parenting outcomes, and that interventions with a prenatal and postnatal component have stronger effects on couple communication than pure prenatal or postnatal interventions.
Given the observed statistically small effect sizes on couple adjustment, couple communication, parenting, and parental psychological well-being at posttest, one may challenge the practical significance of the observed effects. As the large majority of interventions were, however, universal prevention programs and this kind of prevention usually leads to small or very small effects (Burig, 2002), the present results are not surprising. From a statistical point of view the effects might be small; nevertheless the effect size of d = .28 for couple communication means that participants of the intervention condition with above-median improvement outnumber the successful participants of the control condition by 14% (Lipsey & Wilson, 2001). This difference is practically meaningful. Nonetheless, as meta-analyses on general marriage education found larger average effect sizes than the present meta-analysis (Giblin et al., 1985; Hawkins, Blanchard, et al., 2008), our results indicate that more can be done to improve the efficacy of couple interventions during the transition to parenthood.
The moderator analysis provided some suggestions on how to increase intervention effects. First, similar to the work of Giblin et al. (1985) and Hawkins, Blanchard, et al. (2008), we found that a sufficient application rate of intervention sessions is needed for preventing declines in couple adjustment. Thus, couples need time for learning and implementing communication or coping skills. Second, with regard to couple communication, the present meta-analysis was the first to show that a combination of a prenatal and postnatal component could be another way of enhancing intervention effects. As couples have to cope with different challenges during pregnancy and after giving birth to a child (Cowan & Cowan, 2000), interventions that are held during both periods may be best suited for promoting a positive couple relationship.
A third option for enhancing intervention effects is the use of professionals rather than paraprofessionals or lay person as interventionists. The observed stronger effects of interventions conducted by professionals indicate that well-trained interventionists, such as family therapists, are more able to identify couples' needs for change and to develop and implement adequate strategies for addressing these needs. Although the implementation of interventions conducted by paraprofessionals would be more economical, they might be more costly in the long run, as we did not find significant effects of these interventions on couple adjustment. Finally, because only three included studies provided selective prevention for parents at risk, we could not test whether these interventions would have stronger effects than universal programs. The meta-analysis on premarital and marital interventions by Giblin et al. (1985) suggests that this may be the case. If true, a challenge will be to reach those new parents who would benefit most from couple-focused interventions.
Limitations and Conclusions
Several limitations specific to the present meta-analysis have to be mentioned. First, compared to early parenting education interventions, relatively few couple-focused interventions have been conducted. Second, as most studies focused on universal prevention with well-educated parents, intervention effects on couples at risk cannot yet be evaluated. Third, because of the lack of available studies, we could not analyze the effect of couple-focused interventions on child development. Fourth, because of the limited number of available studies, we may not have been able to find some intervention effects, such as statistically significant improvement of couple stability at posttest. Similarly, we were not able to identify very small moderating effects of study characteristics. Fifth, given the average time interval of 12.2 months between the end of the intervention and the follow-up assessment, effects over a longer time span cannot yet be evaluated. Finally, although our meta-analysis focused on immediate transition to parenthood, some authors have suggested that the transition to parenthood extends to 2–3 years postpartum (Cowan & Cowan, 2000).
Despite these limitations, we first conclude that couple-focused interventions produce meaningful, although in a statistical sense, small to very small effects concerning the adjustment of the couple. Second, because of the smaller effects of the present meta-analysis than of previous reviews on marriage and relationship education, more needs to be done to improve the efficacy of couple interventions during the transition to parenthood. In order to ensure larger intervention effects, interventions should include a prenatal and postnatal component, at least six sessions, and be instructed by professionals. Although very short interventions conducted by paraprofessionals would be cheaper to implement, they cannot be recommended as long as there is no sufficient empirical evidence for their effectiveness. Third, as additional effects on parenting were only found when adding a parenting-focused intervention component, interventions are recommended that combine couple- and parenting-focused components. Fourth, given the limited number of available controlled studies, more research is needed on the immediate and long-term effects of couple-focused interventions with expectant and new parents. At the time of this writing, two new intervention studies are being conducted by Pamela Jordan and John Gottman, but information about program effects is not yet available. Finally, we recommend more research on couple-related factors that may moderate the size of intervention effects. Identifying these factors could facilitate matching interventions to the explicit requirements of couples and making the interventions more effective.
