افسردگی مادران و رفتار پدر و مادر: بررسی فراتحلیلی

Abstract The results of 46 observational studies were analyzed to assess the strength of the association between depression and parenting behavior and to identify variables that moderated the effects. The association between depression and parenting was manifest most strongly for negative maternal behavior and was evident to a somewhat lesser degree in disengagement from the child. The association between depression and positive maternal behavior was relatively weak, albeit significant. Effects for negative maternal behavior were moderated by timing of the depression: Current depression was associated with the largest effects. However, residual effects of prior depression were apparent for all behaviors. Socioeconomic status, child age, and methodological variables moderated the effects for positive behavior: Effects were strongest for studies of disadvantaged women and mothers of infants. Studies using diagnostic interviews and self-report measures yielded similar effects, suggesting that deficits are not specific to depressive disorder. Research is needed to identify factors that affect the magnitude of parenting deficits among women who are experiencing depression and other psychological difficulties.

Results The mean weighted effect sizes for the three categories of maternal behavior are presented in Table 3. The effect size for each category of behavior differed significantly from zero. For measures of disengaged and negative behavior, depressed mothers exhibited significantly higher levels of behavior than nondepressed mothers. On measures of positive behavior, depressed mothers exhibited significantly lower levels of behavior than nondepressed mothers. The effect sizes for measures of negative and disengaged behaviors were moderate in magnitude. The effect size for measures of positive maternal behavior was small in magnitude. The fail-safe Ns for each category of behavior are also presented in Table 3. All of the Ns are sufficiently large to suggest that the presence of unpublished studies with nonsignificant results would not affect the results of our analyses. Table 3. Effect Sizes for Three Categories of Maternal Behavior k d 95% CI r Fail-Safe N Qw Negative 30 0.40 0.31/0.49 .20 763 66.69** Disengaged 20 0.29 0.17/0.41 .14 203 38.18** Positive 34 0.16 0.08/0.23 .08 458 76.32** ** p < .01. Table options Stem and leaf displays of the effect sizes (r s) for the three categories of maternal behavior are presented in Table 4, Table 5 and Table 6. Correlations between depression and negative behavior ranged from −.44 to .58. Of the 30 effect sizes for negative maternal behavior, 12 (40%) differed significantly from zero (p < .05) in the expected direction and 1 (3%) differed significantly from zero in the direction opposite expectations. The remaining 17 (57%) did not differ significantly from zero. For disengaged maternal behavior, correlations ranged from −.12 to .50. Of the 20 effect sizes for disengaged maternal behavior, 9 (45%) were significantly different from zero (p < .05) and in the expected direction; 11 effect sizes (55%) did not differ significantly from zero. For positive maternal behavior, correlations ranged from −.13 to .69. Of the 34 effect sizes for positive maternal behavior, 12 (24%) were significantly different from zero (p < .05) and in the expected direction. The remaining 26 effect sizes (76%) did not differ from zero. Q statistics, used to assess heterogeneity among effect sizes, are reported in Table 3. The Q values indicate that there was significant heterogeneity among the effect sizes for each of the three categories of maternal behavior. To identify sources of heterogeneity among the studies, six potential moderator variables were examined: (a) timing of depression (current symptoms or diagnosis vs. lifetime diagnosis); (b) definition of depression (self-report vs. interview); (c) socioeconomic status (disadvantaged vs. non-disadvantaged); (d) child age; (e) duration of the observation; and (f) type of observation (structured vs. unstructured laboratory settings vs. home observations). Table 4. Stem and Leaf Display for Negative Maternal Behavior Effect Sizes (r) Stem Leaf .5 1, 5, 8 .4 2, 3 .3 0, 3, 4, 7 .2 1, 3, 5, 6, 7, 7, 8 .1 4, 4, 7, 7 0 0, 0, 0, 0, 4, 4, 8 −.0 2 −.1 1 −.2 −.3 −.4 4 Note: Correlations in bold indicate samples with current depression. Italic correlations indicate economically disadvantaged samples. Table options Table 5. Stem and Leaf Display for Disengaged Maternal Behavior Effect Sizes (r) Stem Leaf .5 0 .4 5, 6, 8 .3 1, 5, 6, 8 .2 3, 6 .1 4, 0 .0 0, 3, 4, 5, 6 −.0 1, 8 −.1 2 Note: Correlations in bold indicate samples with current depression. Italic correlations indicate economically disadvantaged samples. Table options Table 6. Stem and Leaf Display for Positive Maternal Behavior Effect Sizes (r) Stem Leaf .6 9 .5 0, 1, 6 .4 0, 1 .3 6 .2 2, 2, 2, 3, 7, 8 .1 0, 1, 2, 5, 8, 8, 9 .0 0, 0, 0,0, 0, 0, 6, 6, 6, 6 −.0 2, 4, 5 −.1 3 Note: Correlations in bold indicate samples with current depression. Italic correlations indicate economically disadvantaged samples. Table options For negative parenting behaviors, timing of depression moderated the relationship between maternal depression and parenting behavior. These results are summarized in Table 7. There was a significant difference, Qb = 4.40, p < .05, between the mean effect size for studies with samples of currently depressed women and the mean effect size for studies with samples of women with a lifetime diagnosis of depression. The effect size for currently depressed samples was moderate in magnitude while the effect size for samples of women with lifetime diagnoses of depression was small in magnitude. There was significant heterogeneity in the effect sizes of the studies of mothers with current symptoms and additional moderator analyses were conducted on the group of studies using currently depressed samples. However, no other significant moderating effects were found. Table 7. Moderators for Negative Behaviors k d 95% CI r Qw Timing of Depression Current Symptoms/Diagnosis 23 .45 .35/.55 .22 52.22** Lifetime Diagnosis 7 .22 .02/.41 .11 10.06 ** p < .01. Table options For disengaged behaviors, none of the moderators we selected accounted for the heterogeneity among the studies. That is, timing of depression, definition of depression, socioeconomic status, child age, length of observation, and type of observation all failed to explain the heterogeneity among the studies. Among studies that assessed positive maternal behavior, socioeconomic status, age of child, length of observation, and type of observation moderated the association between maternal depression and parenting behavior. Table 8 presents a summary of these results. There was a significant difference, Qb = 14.87, p < .001, between the mean effect size for studies with samples of economically disadvantaged women and studies with samples of women who were not economically disadvantaged. The association between depression and positive maternal behavior was moderate in magnitude among economically disadvantaged; among studies of nondisadvantaged women, the effect was zero. There was also a significant difference between the mean effect sizes for studies with samples of children of different ages. Due to the limited number of studies involving children 6 and older, only mean effect sizes for the two younger groups (infants vs. toddlers and preschoolers) were compared. A significant difference between the mean effect sizes for these two groups was found, Qb = 6.78, p < .01. A moderate effect size was obtained for studies of mothers of infants and a small effect size was found for studies of mothers of toddlers and preschool-aged children. A significant difference was found among mean effect sizes based on length of observation, Qb = 9.7, p < .01. Pairwise comparisons of the three lengths of observations demonstrated that the effects of depression on positive maternal behaviors were stronger when observations were 10 minutes or shorter than when observations were 11 to 59 minutes in length, z2 = 7.86, p < .05, or 60 minutes or longer, z2 = 6.36, p < .05. The effect sizes for the studies in which observations were 11 to 59 minutes in length and studies in which observations were 60 minutes or longer did not differ, z2 = 0.32, p > .10. Finally, there was a significant difference, Qb = 23.59, p < .001, among the mean effect sizes for studies with samples based on different types of observations. Pairwise comparisons of the three types of observations demonstrated that the effects of depression on positive maternal behavior were stronger when observations were conducted in unstructured laboratory settings than when observations were conducted in structured laboratory settings, z2 = 9.47, p < .01, or in the home, z2 = 23.59, p < .001. The effect sizes for studies using structured laboratory observations and home observations did not differ, z2 = 1.01, p > .10. Table 8. Moderators for Positive Behaviors k d 95% CI r Qw Total sample Socioeconomic status Disadvantaged 11 .42 .26/.59 .21 34.93*** Nondisadvantaged 21 .06 −.03/.15 .03 21.53 Child Age <1 year 12 .47 .30/.64 .23 36.21*** 1–5 years 15 .19 .06/.32 .10 10.00 Length of observation 1–10 minutes 13 .48 .32/.65 .23 34.46** 11–59 minutes 7 .12 −.02/.47 .06 2.56 60 or more minutes 11 .19 .04/.35 .10 6.92 Type of observation Home 14 .05 −.04/.15 .03 20.29 Lab-Structured 9 .16 −.03/.36 .08 9.16 Lab-Unstructured 9 .60 .40/.79 .29 23.85** ** p < .01. *** p < .001. Table options In that there was significant heterogeneity in the effect sizes for positive maternal behavior among studies of mothers who were economically disadvantaged, additional moderator analyses were attempted. However, the small number of studies in each group precluded further moderator analyses.