Postpartum depression or postnatal depression is a term that describes the occurrence of moderate to severe depression in a woman after she has given birth (although sometimes men are given this diagnosis when severe depression occurs after the birth of a child). This depression may occur soon after delivery and may linger up to a year or longer. In the majority of recognized cases the depression occurs within the first three months following the delivery of the child. The DSM-IV does not recognize postpartum depression as a distinct disorder. People who receive a diagnosis of postpartum depression must first meet the standard diagnostic criteria for a major depressive episode and then they must satisfy the additional specifier criteria for the postpartum onset (American Psychiatric Association [APA], 2000). This criterion states that the onset of the major depressive episode must occur within four weeks after delivery.
Postpartum depression then should be responsive to traditional treatments for depression. However, because of the potential complications in using antidepressant medications to treat postpartum depression in nursing mothers an area of focus in the treatment of postpartum depression has been psychotherapy. O'Hara, Stuart, Gorman, and Grant (2000) report that investigating the effectiveness of psychotherapy for postpartum depression is important because of previous research that has identified mother-infant bonding is impaired by maternal depression and the increased social role adjustments of the new mother are seriously hampered by depressive symptoms. The authors identify previous research that has demonstrated that cognitive-behavioral therapy and interpersonal therapy are effective in the reduction of the symptoms in postpartum depression, but cite several major flaws in previous research such as: (1) Including participants with both minor and major depression in the analysis (thus, in patients with minor depression small changes would appear as if there was a large treatment effect). (2) Many studies included non-standardized or "non-manualized" therapies (thus, treatment across different patients or therapists could differ substantially). (3) Many studies utilized non-professional therapists such as nurses or visitors (this obviously could affect the treatment outcomes). (4) Some studies used therapies that were not targeted at treating depression. O'Hara et al. (2000) concluded that findings from such studies had poor validity and that a more controlled study on the effectiveness of psychotherapy for the treatment of postpartum depression was in order. They hypothesized that under tightly controlled experimental conditions those with postpartum depression who received interpersonal therapy (IPT) would demonstrate a significant reduction in depressive symptoms and increases in social functioning compared to a wait list control group with postpartum depression (WLC). The researchers chose IPT as the manipulation over other therapies as IPT's focus on personal relationships with others directly addresses problems previously demonstrated to be of particular concern in women with postpartum depression. IPT is a time-limited psychotherapy that places emphasis on interpersonal contexts such as building interpersonal skills. This is because interpersonal factors are believed to significantly contribute to psychological problems such as depression (unlike other therapies that often focus on intrapsychic processes). Originally developed by Harry Stack Sullivan, IPT was initially heavily influenced by psychodynamic psychotherapy but has borrowed from cognitive behavioral approaches as well (Weissman & Markowitz, 1998).
O'Hara et al. (2000) sent recruiting letters 20620 recent mothers. After several screening steps where women were assessed for depression meeting the qualifiers for postpartum depression, subjected to strict exclusion criteria (including comorbid psychiatric disorders and cognitive disorders), and assessed for their level of depression and social functioning 120 Caucasian women were randomly assigned to IPT or WLC groups. There were no significant differences between participants in the two conditions on their initial level of depression (all ranged within the moderate to the severely depressed range), length of depressive episode, or on other demographic variables (age, education, etc.). The groups were re-assessed for depression and social functioning at four, eight, and twelve week periods (twelve weeks was the length of the study). WLC participants were followed to make sure that they did not get treatment for their depression as well.
Therapists in the IPT group were all Ph.D. Or PsyD level psychologists. Therapists were trained in IPT with 40 hours of didactics and themselves attended 12 weeks of IPT therapy with a postpartum depressed patient before they treated the subjects. There were no significant differences concerning demographic factors between the therapists. During the experiment therapists video-recorded therapy sessions and these were reviewed with the researchers to ensure treatment protocol adherence. Therapists saw between one and eleven patients for the 12 weeks. Attrition rates were not significantly different between the two conditions with 12 of 60 withdrawing from IPT and 9 of 60 patients from WLC (Chi-square = ns).
O'Hara et al. (2000) had outlined several limitations of previous studies, all of which related to internal validity, or the ability of the experimental procedure to demonstrate a causal relationship between the independent and dependent variable by controlling for confounds or other variables that could affect a change on the dependent measure other than the independent measure (here the independent variable is the IPT administration). Thus, the confounds of differential levels of depression in the groups, poorly trained therapists, non-standard treatments, and other problems with internal control in previous studies outlined by the researchers were addressed in this study. This study endeavored to provide clear controlled conditions such as equivalence between treatment and control conditions, homogeneous measurements, and therapists suing a standardized protocols for treatment. The randomization of the participants to the treatment conditions is a powerful internal control to ensure equivalence between groups and accomplished its goal. In some cases when researchers want stricter control subjects can be matched on certain subject variables such as age and then randomly assigned to the conditions, but that was not necessary in this design. And of course the WLC represents a true control condition, crucial in experimental research.
The major analyses were performed with repeated measures ANOVAs on standardized measures of depression (Hamilton Depression Inventory [HAMD] and the Beck Depression Inventory [BDI] for depression and the Social Adjustment Scale-self-report scale [SAS-SR], Postpartum Adjustment Questionnaire [PAQ] and the Dyadic Adjustment Scale [DAS] for changes in interpersonal relations). There were significant reductions in the level of depression for the IPT group starting at four weeks. These differences indicated significantly lower depression scores in the IPT group compared to the WLC group at four weeks and continuing across the length of the study on both measures of depression. The WLC did demonstrate a slight decline in depression, but this was not significant and the level of the WLC group's depression was always significantly higher than the IPT group across all but the baseline measurement. The SAS-SR demonstrated a modest increase of a third of a standard deviation in the IPT group but still a significant difference compared the WLC group. The PAQ and DAS displayed modest rises as well. The authors' hypothesis was supported: IPT did demonstrate a significant reduction of depression in postpartum depressed mothers compared to a WLC condition.
Normally that would be a happy end to this tale, but let us look more closely now that we know about research. First, even though the researchers worked hard to provide an experiment with good internal validity, they missed the boat when the study is evaluated in terms of its external validity. There was no attempt at random sampling. The authors do acknowledge this limitation. The recruiting letters were sent out throughout the state of Iowa resulting in a sample of Caucasian women with a mean age of about 30 (standard deviation of about 4.5) with a mean of about 14.5 years of education (standard deviation about 4.5). The generalziability of the findings is quite limited and again the authors do acknowledge this; however, what they do not acknowledge is that their sample may have fit the old YAVIS notion of successful therapy responders (Young, Attractive, Verbal, Intelligent, and Successful) given the review of the demographic data of the sample (Lewis, Davis, Walker, & Jennings, 1981). YARVIS clients respond well to any therpautic intervention. This may be the only group that would demonstrate such an effect compared to a WLC. People with different ages, levels of education, or ethnic backgrounds may not respond to IPT. Moreover, certain exclusion criteria applied in this study such as being under 20 years of age, a history of mood disorder, alcohol use, poor spusal relationships, etc. are known to be risk factors for postpartum depression (APA, 2000) and by excluding participants with these factors generalizibilty may be limited.
The dropout rate in this study was not significantly different between the two conditions, but the researchers made a mistake by not trying to follow up on those that prematurely left the study. Often, in medical and psychotherpautic research there is something significantly different about those that leave the research compared to those that stay (Hamilton, Moore, Crane, & Payne, 2011). These differences may have influnced the final outcome measure had those subjects remained. For instance, those leaving the treatment condition may not be responding to the treatment, the treatment…