Table 3.
Meta-analyses and Post-hoc Analyses of Pooled Data Evaluating Effectiveness of Treatments for Depression Among Women of Reproductive Age
| Treatment | Description |
|---|---|
| Exercise | |
| Authors, year | North et al, 1990 (35) |
| Type of analysis | Meta-analysis |
| Studies included: design(s); total number/number women only; years studies published | RCT, non-RCT comparative, 1 group pre-post, matching groups, convenience sample groups, and pre-test/post-test studies with any measure of depression as dependent variable; 80/16; English language; published or scheduled for publication on or before June 1, 1989. |
| Population of interest and exclusion criteria | Studies of any reported depression including mood disorders, psychogeneous or endogenous types of depression, primary or secondary |
| Outcome, intervention, comparison group, and covariates | Outcome: Level of depression; intervention: aerobic and resistance exercise; comparison groups: no treatment, wait-list, psychotherapy, enjoyable activity, relaxation, less exercise, anaerobic exercise, exercise and psychotherapy; covariates: source of subjects, group assignment, degree of internal validity, initial level of depression, age, sex, exercise duration per episode, type, frequency and intensity, additional therapy, health status. |
| Results | The overall ESa of exercise on depression was −0.53 ± 0.85.Significant main effects were found for source of subjects (medical/psychological patients > students/citizens); source of study (published > unpublished); purpose of exercise (medical rehabilitation > general health or psychological rehabilitation); health status (sicker > healthier); type of exercise (weight training > aerobic); duration (longer > shorter). No significant moderating effects found for group assignment, internal validity, sex, age, or depression diagnosis. |
| Limitations | Included non-RCT studies and studies with depression as secondary diagnosis to another mental health condition. |
| Authors, year | Craft and Landers, 1998 (33) |
| Type of analysis | Meta-analysis |
| Studies included: design(s); total number/number women only; years studies published | RCT, non-RCT comparative and pretest/posttest studies with measure of depression as dependent variable; 37/4; published or scheduled for publication on or before November 1996. |
| Population of interest and exclusion criteria | 2,158 people with depression as primary diagnosis or secondary diagnosis to another mental health condition; excludes studies on depression as a result of physical health problem. |
| Outcome, intervention, comparison group, and covariates | Outcome: level of depression; intervention: aerobic and resistance exercise; comparison groups: wait list, group or individual therapy, behavioral interventions; covariates: initial level of depression; age; sex; exercise duration per episode, type, frequency, and intensity; additional therapy. |
| Results | Compared with wait-list controls, people who exercised were ES 0.77 (95% CI, −1.08 to −0.47) less depressed than individuals on a wait list. Exercise was as beneficial as group/individual therapy and behavioral interventions.Significant main effects were found for initial level of depression (moderate to severe group > mild to moderate group); source of study (published > unpublished); primary versus secondary depression; no significant moderating effects were found for sex, age, or exercise type, duration, frequency, or intensity on the relationship between exercise and depression. Individuals who exercised 9–12 weeks were less depressed than those who exercised ≤8 weeks. |
| Limitations | Included non-RCT studies; included depression as secondary diagnosis to another mental health condition; limited generalizability because of exclusion criteria; measures of level of depression not stated. |
| Authors, year | Rethorst et al, 2009 (34) |
| Type of analysis | Meta-analysis |
| Studies included design(s); total number/number women only; years studies published | RCTs only; 58/7; 1981-2005. |
| Population of interest and exclusion criteria | 2,982 people with depression (574 clinical, 2,408 nonclinical); depression associated with physical or psychological illness excluded |
| Outcome, intervention, comparison group and covariates | Outcome: level of depression; intervention: moderate to vigorous aerobic or resistance exercise; comparison group: no treatment or wait list, secondary comparison groups: psychotherapy and antidepressant medication; covariates: depression type, intervention duration, exercise type, frequency, bout duration, sex, methodological characteristics, treatment adherence, dose response. |
| Results | Overall ES of −0.80 indicates participants in the exercise treatment had significantly lower depression scores than controls. Exercise was more effective among clinically depressed participants (ES = −1.03) than nonclinical samples (ES = −0.59). Aerobic and resistance exercises were equally effective. No significant differences in the effect of exercise compared with psychotherapy or antidepressant medication. No significant moderating effects found for sex on the relationship between exercise and depression. |
| Limitations | Limited generalizability because of exclusion criteria. |
| Cognitive therapy | |
| Authors, year | Gloaguen et al, 1998 (36) |
| Type of analysis | Meta-analysis |
| Studies included: design(s); total number/number women only; years studies published | RCTs only; 48/4; Published studies and those presented at international congresses, 1977–1996 |
| Population of interest and exclusion criteria | 2,765 people with major depression or mild/moderate dysthymia; excluded psychotic and bipolar disorder |
| Outcome, intervention, comparison group and covariates | Outcome: level of depression assessed by BDI; intervention: cognitive therapy; Comparison group: waiting list or placebo, antidepressant medication, behavioral therapy, other psychotherapeutic treatment (psychodynamic therapies, interpersonal therapies, nondirective, supportive, relaxation and alternative bibliotherapy); covariates: BDI scores, sex, age. |
| Results | Cognitive therapy was significantly better than waiting-list or placebo (P <.001), antidepressants (P < .001), and a group of miscellaneous therapies (P <.001). No covariates modified effect size in multivariable analysis. |
| Limitations | Limited to published articles and studies presented at international congresses. Limited generalizability because of exclusion criteria. |
| Psychotherapy | |
| Authors, year | Robinson et al, 1990 (37) |
| Type of analysis | Meta-analysis |
| Studies included design(s); total number/number women only; years studies published | Studies comparing treatment to no treatment or different types of therapy; excluded case histories and pre-post designs and treatments without a prominent verbal component, and marriage/family therapy; 58/unknown; Mean percentage of female clients per study was 80%; range: 50%–100%; psychological abstracts 1976-1986 and relevant journals 1985-1986. |
| Population of interest and exclusion criteria | People with depression either meeting formal diagnostic criteria or screening positive; excluded subjects described in more general terms or by other specific diagnoses; studies examining inpatients or children and adolescents also excluded; mean age: 39 years, range: 19-71 |
| Outcome, intervention, comparison group, and covariates | Outcome: depression symptoms; assessed by multiple different validated screeners for depression, general mental health and functioning; intervention: 1 of 4 types of therapies: 1) cognitive, 2) behavioral, 3) cognitive-behavioral, 4) general verbal, 7 mean weeks of treatment and 8.7 mean number of sessions; comparison: no treatment (n = 46 studies), wait list (n = 29 studies), placebo (n = 9 studies); covariates: sex and age, weeks of treatment, number of sessions. |
| Results | Psychotherapy was more effective than no treatment (ES = 0.73 at posttreatment and 0.68 at follow-up, average 13 weeks after treatment) and wait list (ES = 0.84) (P < .05 for all). In the 29 studies using outcome measures specific to depression, psychotherapy was more effective than wait list (ES = 0.93, P < .05). No differences in effect sizes found by sex, age, weeks of treatment, or number of sessions. |
| Limitations | Included non-RCTs; limited to published articles and abstracts presented at conferences. |
| SSRIs/SNRIs | |
| Authors, year | Entsuah et al, 2001 (38) |
| Type of analysis | Post-hoc analysis of pooled data |
| Studies included design(s); total number/number women only; years studies published | Included placebo controlled double-blind, active-controlled phase 2,3, or 5 trials; 8/0; 62%-65% of participants female, depending on intervention type; published or reported 1992 to 1998. |
| Population of interest and exclusion criteria | 2,045 people meeting DSM-III or IV criteria for major depressive disorder and ≥20 on HAM-D-21 or 25 on Montgomery Asberg Depression Rating Scale; age range 18-83 years; inpatients (n = 67) and outpatients (n = 1,977); excluded pregnant, lactating; significant history of cardiovascular, renal, hepatic, or seizure disorders; abnormal physical examination or electrocardiogram; history of alcohol or drug abuse; use of investigational or antipsychotic drugs in last 30 days, monoamine oxidase inhibitors within 14 days, or antidepressants anxiolytics or sedative-hypnotic drugs within 7 days. |
| Outcome, intervention, comparison group and covariates | Outcome: depression absence, response or remission to treatment, measured by score of 0 on depressed mood item of HAM-D-21; score ≤7 on HAM-D-17; and ≥50% decrease in score on HAM-D-21, respectively; intervention: venlaflaxine (n = 865); SSRI (n = 757); 6-12 weeks; comparison group: placebo (n = 450); covariates: age, sex |
| Results | For both men and women, both venlaflaxine and SSRIs were significantly more effective in depression absence, response, and remission than placebo. At 8 weeks, rates of remission among women treated with venlaflaxine were higher (45%) than those among people receiving SSRIs (34%) or placebo (24%), P < .001. Rates of response with venlaflaxine (65%) were higher than placebo (43%), P < .001. Rates of absence of depressed mood among women receiving venlaflaxine (37%) or SSRI (31%) were higher than placebo (20%), P < .003. Among participants ≤40 years (men and women combined), rates of remission (P < .001) and absence of depression (P < .001) were higher with venlafaxine than placebo. Differences between SSRIs and placebo were not statistically significant among participants ≤40 years. No effect of age or sex on effectiveness in depression absence, remission, or response to venlaflaxine or SSRIs. |
| Limitations | Limited generalizability; disproportionately large number of patients treated with fluoxetine in SSRI group; no studies examined setraline or citalopram; no information on menopausal status. |
| Authors, year | Khan et al, 2006 (39) |
| Type of analysis | Post-hoc analysis of pooled data. |
| Studies included design(s); total number/number women only; years studies published | Double-blind RCTs, phase 2, 3, and 4; 15/0; conducted 1996–2003. |
| Population of interest and exclusion criteria | 323 people with depression; 177 (55%) women with depression; n = 80 on placebo; n = 71 on SSRI; n = 26 on SNRI; excluded people with severe illness, suicidal patients, and patients with concomitant disorders. |
| Outcome, intervention, comparison group and covariates | Outcome: level of depression assessed by response and remission rates from HAM-D-17 scores; intervention: SSRI (fluoxetine, paraxetine CR, sertaline, citalopram, excitalopram), SNRI (venlaflaxine ER); comparison group: placebo; covariates: baseline depression score and sex. |
| Results | Response rates greater for women taking SSRIs (64.8%) and SNRIs (69.2%) than placebo (40%) (P < .001). Remission rates higher in women taking SSRIs (45.1%) and SNRIs (46.2%), than placebo (13.8%) (P < .001). Women taking SSRIs had ES 0.82, and women taking SNRI had ES 0.76. Effect of SSRIs greater in women than men (P = .001). No difference between women and men taking SNRI. |
| Limitations | Duration of medication use not stated; limited generalizability because of exclusion criteria. |
| Authors, year | Kornstein, et al, 2006 (40) |
| Type of analysis | Post-hoc analysis of pooled data. |
| Studies included design(s); total number/number women only; years studies published | Randomized, multicenter, double-blind, placebo-controlled trials; 7/0; 2002-2005. |
| Population of interest and exclusion criteria | N = 1622 people with major depressive disorder defined as HAM-D-17 score >15; 1,062 women n = 578 duloxetine; n = 484 placebo; mean age 41 years; excluded people with current and primary Axis I disorder other than depression, an Axis II disorder, lack of response to ≥2 courses of antidepressant therapy, serious medical illness, risk of suicide, history of substance abuse in last year, positive drug screen. |
| Outcome, intervention, comparison group and covariates | Outcome: level of depression; intervention: duloxetine (40, 60, 80, and 120 mg/day); comparison group: placebo; covariates: sex, study design characteristics, methodological characteristics, study duration (7-9 weeks). |
| Results | For women, duloxetine was more effective than placebo in HAM-D-17 (ES = 0.22, P <.001), CGIS (ES = 0.19, P <.001), and PGI-I (ES = 0.30, P <.001) measures. Significantly greater improvement in VAS pain scores in duloxetine-treated compared with placebo-treated women. |
| Limitations | Duration of treatment 7-9 weeks; menopausal status unknown; generalizability limited because of exclusion criteria. |
Abbreviations: RCT, randomized controlled trial; ES, effect size; CI, confidence interval, BDI, Beck Depression Inventory; NR, not reported; SSRI, selective serotonin reuptake inhibitor; SNRI, selective norepinephrine reuptake inhibitor; HAM-D, Hamilton Rating Scale for Depression; CGIS, Clinical Global Impressions-Severity; PGI-I, Patient Global Impression of Improvement scale; VAS, visual analogue scale.
a
ES = standardized mean difference between intervention and control groups.