Changes in parenting and chronic interpersonal stress in families having a parent with BD

Families having a parent with BD reported significant pre-post improvements on the difficult child subscale (b = − 0.40, SE = 0.16, p = 0.016), but not on the parental distress subscale (b = − 0.25, SE = 0.20, p = 0.232), the dysfunctional interaction subscale (b = − 0.24, SE = 0.15, p = 0.132), or on the total stress score (b = − 0.32, SE = 0.18, p = 0.079). Reported pre-post changes in chronic interpersonal stress were not significant for families having a parent with BD (b = 0.10, SE = 0.22, p = 0.661).

Across the four assessment points, reported changes on the difficult child subscale followed a quadratic curve (b = 0.12, SE = 0.04, p = 0.008). While improvements were noted until T3, scores on the difficult child subscale significantly increased at T4. Similarly, changes on the total stress score also followed a quadratic curve (b = 0.26, SE = 0.05, p < 0.001); scores improved until T3, but worsened at T4. Finally, changes on the parental distress subscale across time followed a linear curve (b = − 0.12, SE = 0.07, p < 0.001), with continued improvements reported until T4. Linear change on the dysfunctional interaction subscale also improved over the four timepoints, although the effect fell short of conventional levels of statistical significance (b = − 0.28, SE = 0.15, p = 0.063). In sum, despite a small post-intervention effect, improvements in parenting stress continued long after the termination of the intervention (see Fig. 2, panels a through d). In terms of chronic interpersonal stress, no significant changes were reported by families having a parent with BD across the four timepoints (b = 0.07, SE = 0.07, p = 0.319).

Fig. 2

Note: The linear effect of time on all four subscales by intervention group. OBD offspring of parents with bipolar disorder

Between‐group differences between families having a parent with BD and control familiesDifferences in reported levels of parenting and life stress at baseline

There was statistically significant variation in the intercepts of the total stress score (Wald Z = 4.07, p < 0.001), the parental distress subscale (Wald Z = 2.88, p = 0.004), the dysfunctional interaction subscale (Wald Z = 4.47, p < 0.001), and the difficult child subscale (Wald Z = 4.37, p < 0.001). In addition, there was significant variation in the intercept of the chronic interpersonal stress score (Wald Z = 4.11, p < 0.001). These findings indicate that between-subject effects influence scores at T1.

Risk-status significantly predicted intercepts for the total stress score (b = − 1.19, SE = 0.19, p < 0.001), as well as the parental distress (b = − 1.04, SE = 0.16, p < 0.001), dysfunctional interaction (b = − 0.91, SE = 0.22, p < 0.001), and difficult child subscales (b = − 0.99, SE = 0.20, p < 0.001). Significant intercept effects indicated that families having a parent with BD, as expected, reported higher levels of parenting stress than control families at T1. Additionally, risk status predicted initial levels of chronic interpersonal stress (b = − 0.54, SE = 0.27, p = 0.044). Neither offspring sex or age at T1 accounted for any variability in intercepts.

Differences in reported changes in parenting and chronic interpersonal stress over time

There was statistically significant variability in the linear effects of time for the difficult child subscale (Wald Z = 2.27, p = 0.023), which indicates that between-subject effects influence its rate of change. For the total stress score (Wald Z = 1.77, p = 0.076) and the dysfunctional interaction subscale (Wald Z = 1.86, p = 0.063), the amount of inter-individual variability was trend-level. Finally, for the parental distress subscale, the observed between-subject differences in changes over time were not statistically significant (Wald Z = 1.34, p = 0.180). In terms of chronic interpersonal stress, the level of variability in the linear effects of time was non-significant (Wald Z = − 0.723, p = 0.470).

Change across time varied by risk-status for the total stress score (b = 0.95, SE = 0.18, p < 0.001), the dysfunctional interaction subscale (b = 0.64, SE = 0.21, p = 0.003), and for the difficult child subscale (b = 0.94, SE = 0.23, p < 0.001) (see Fig. 1a, c, d, respectively). Families having a parent with BD reported changes which followed a steeper curve than control families. Upon further investigation, families having a parent with BD also reported greater pre-post improvement on both the total stress score (b = 0.54, SE = 0.24, p = 0.027) and on the difficult child subscale (b = 0.56, SE = 0.22, p = 0.012). Finally, the rate of change for chronic interpersonal stress scores did not vary between groups (b = − 0.01, SE = 0.09, p = 0.925); this means that improvements in observer-rated levels of stress were not associated to risk-status. Offspring sex and age were not significantly associated to variability in growth trajectories.

Parallel mediations models

Standardized model results predicting both internalizing and externalizing symptoms are summarized in Tables 3 and 4, respectively. Internalizing and externalizing scores were each averaged across T3 and T4. The mean internalizing score (± SD) for OBD was 12.64 (11.71), while it was 9.11 (8.10) for control offspring. The mean externalizing score for OBD was 17.08 (15.36), and 6.72 (8.72) for control families. Mediations were run using change scores on parenting stress, however changes in chronic interpersonal stress were omitted from these analyses given the non-significant slope terms for families having a parent with BD reported above.

Table 3 Standardized model results of parallel mediations predicting parent-reported internalizing problems across follow-upTable 4 Standardized model results of parallel mediations predicting parent-reported externalizing problems across follow-up

Improvements on the total stress subscale mediated the relationship between participation in the RUSH intervention program and internalizing symptoms across follow-up (β = − 0.31, SE = 0.10, Cl = − 0.55, − 0.13; Table 2). Additionally, reported improvements on the total stress subscale also mediated the relationship between participation in the RUSH program and externalizing symptoms across follow-up (β = − 0.35, SE = 0.10, Cl = − 0.57, − 0.16; Table 3). None of the other indirect effects were significant.