Although many women encounter mildly troublesome premenstrual symptoms during their reproductive years, approximately 3–5% of women suffer from the more debilitating condition, premenstrual dysphoric disorder. Women with premenstrual dysphoric disorder experience moderate-to-severe mood, behavioral, and physical symptoms that disrupt their lives at home or work.1,2 Given that women with premenstrual dysphoric disorder face symptoms on a monthly basis from their early 20s until menopause (age ∼51), they may be symptomatic for about 2,800 days, or 7–8 years, over their lifetime. This extended symptomatic period, in addition to the prevalence of the illness, places premenstrual dysphoric disorder in a similar burden-of-illness category as dysthymic disorder, major depression, and other common medical disorders.3
The biological processes underlying premenstrual dysphoric disorder are not known. However, symptoms are diminished by suppressing ovarian activity4,5 and can be provoked in medically ovariectomized women by exposing them to exogenous gonadal steroids.6 Combined oral contraceptive pills (OCPs) can prevent ovulation and replace endogenous fluctuations of ovarian steroids with stable hormone levels. For this reason combined OCPs are used to treat premenstrual symptoms.7 However, significant differences between active medication and placebo were not found in the only 2 published randomized, placebo-controlled trials.8,9
The purpose of this study was to compare posttreatment symptom scores among women suffering from premenstrual dysphoric disorder who were randomized to either a new OCP formulation containing drospirenone 3 mg and ethinyl estradiol 20 μg or placebo. Drospirenone, a novel progestin with antimineralocorticoid and antiandrogenic activity, is an analogue of spironolactone, which has been shown to ameliorate symptoms of premenstrual syndrome.10 The active hormones were administered for 24 consecutive days rather than the conventional 21 days in a 28-day cycle (21/7). This new treatment platform is associated with greater suppression of follicle development and a more stable hormonal milieu than is found with the standard 21/7 regimen.11,12 Given the prior efficacy of spironolactone for premenstrual symptoms, the follicular suppression found with a 24/4 regimen as well as the lower estrogen dose, we hypothesized that women randomized to active treatment would experience fewer premenstrual symptoms than women randomized to placebo.
MATERIALS AND METHODS
This double-blind, randomized, placebo-controlled, parallel-design study consisted of 7 visits: a screening visit, 2 single-menstrual-cycle qualification visits, 3 single-menstrual-cycle treatment visits in which subjects took either drospirenone 3 mg/ethinyl estradiol 20 μg or placebo, and an end-of-treatment visit. (Participant flow is shown in Fig. 1.) Subjects used barrier contraception throughout the study. All protocols were approved by a site or central institutional review board, and all subjects provided verbal and written informed consent.
Subjects were recruited from 64 centers in the United States; recruitment and follow-up occurred from January 2001 through September 2003. Eligibility criteria included 1) aged 18 to 40 years; 2) a diagnosis of premenstrual dysphoric disorder according to the Diagnostic and Statistical Manual of Mental Disorders, 4th edition2; 3) no depressive, anxiety, eating, bipolar, psychotic, somatoform, dysthymic, or drug/alcohol use disorders during the last 2 years; 4) no OCP use for the 3 months before study entry; 5) no concurrent therapy for premenstrual dysphoric disorder; 6) regular menstrual cycles (25–34 days); 7) no contraindication to OCP treatment; 8) a nonsuspicious Pap test within the prior 6 months; and 9) no current pregnancy.
To meet criteria for randomization, subjects were required to have 2 consecutive menstrual cycles of moderate-to-severe premenstrual symptoms and minimal-to-no follicular-phase symptoms immediately before beginning the 3 double-blind placebo or active treatment cycles. Premenstrual dysphoric disorder severity criteria included 1) a premenstrual-phase (days –5 to –1 before bleeding) daily average of 3.0 or greater for 5 distinct items in the Daily Record of Severity of Problems13; 2) a premenstrual-phase daily average that was twice as high as the corresponding postmenstrual-phase average for the 5 distinct items; 3) a postmenstrual phase (days 6–10) daily average of 2.5 or less for each of the 11 distinct items in the Daily Record of Severity of Problems; and 4) a score of 3 or greater on one Daily Record of Severity of Problems functional impairment item for at least 2 premenstrual days.
Randomization numbers were generated in fixed blocks of 4 and were allocated in a 1:1 ratio. The sponsor’s central coordinating group assigned and sent out the blocks in advance of enrollment. Because we anticipated that relatively few patients would be enrolled per site, patients were not stratified by site.
The Daily Record of Severity of Problems scale has 24 questions that are grouped into 11 distinct symptom items and 3 functional impairment items, which are self-rated on a 6-point scale from 1 (not at all) to 6 (extreme). The averages of the last 5 days before menses for the 11 distinct symptom items were summed and constitute the total Daily Record of Severity of Problems symptom score. The a priori primary efficacy variable was the difference between the average luteal phase Daily Record of Severity of Problems total scores from the 2 qualification cycles and the average luteal phase Daily Record of Severity of Problems total scores from the 3 treatment cycles. Responders were defined as subjects who had a reduction from baseline of 50% or greater in their total Daily Record of Severity of Problems symptom score.
Secondary outcome measures included each of the 11 individual items from the Daily Record of Severity of Problems, the 3 functional impairment items from this scale, the Premenstrual Tension Scales observer-rated and self-rated questionnaires,14 the Endicott Quality of Life Enjoyment and Satisfaction Questionnaire,15 and the Clinical Global Impression-Improvement scale.16 Scores for both Premenstrual Tension Scales range between 0 and 36; a premenstrual score of 18–21 is typically found in patients with premenstrual dysphoric disorder.17 Baseline and treatment-related quality of life were assessed with the general activities subscale of the Endicott Quality of Life Enjoyment and Satisfaction Questionnaire.15 This self-report measure includes 14 items rated on a scale of 1–5 (range 14–70), with higher scores indicating a better quality of life. The Clinical Global Impression-Improvement scale is a single-item measure that observers and subjects used to rate global improvement from baseline on a scale of 1 (very much improved) to 7 (very much worse).16
After screening, subjects were administered the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 4th edition,18 to determine the presence of exclusionary psychiatric diagnoses. Daily ratings were kept throughout the study. Clinical Global Impression-Improvement scales were completed during each treatment visit in each cycle. The other scales were completed once before randomization and after the first and third treatment cycles.
Women underwent general physical and pelvic examinations and had routine blood, urine, and pregnancy tests performed during the diagnostic period and after the 3 treatment cycles. Vital signs, weight, and adverse event information were collected at all visits. Randomized subjects started drospirenone/ethinyl estradiol or matching placebo on day 1 or 2 of the menstrual cycle and continued until day 24. The last 4 days of the pill pack consisted of inert placebo pills for both groups. Subjects took pills for 3 menstrual cycles.
Data were analyzed from the intent-to-treat cohort that included all randomized subjects who took at least 1 dose of study pills. Study centers were pooled from largest to smallest until the pooled center had more than 5 subjects with postbaseline data in each treatment group. No pooled center had more than 15% of the total number of subjects. Analysis of variance (ANOVA) was used to compare differences in baseline scores between the active-treatment and placebo groups. The ANOVA model included terms for treatment assignment and pooled center. Comparisons of categorical measures were made using a generalized Cochran Mantel Haenszel test stratified by pooled center.
Daily Record of Severity of Problems scores from the final 5 days of the menstrual cycle were averaged to obtain mean pretreatment (qualifying cycles 1 and 2) or posttreatment (treatment cycles 1–3) individual-item and total values. Missing data points were imputed by averaging days bordering the missing value. There were fewer than 0.25% missing data points, which were distributed equally in both groups. The individual-item and summary scores were carried forward for noncompleters. Differences between daily ratings for the 2 treatment groups were estimated initially using an analysis of covariance (ANCOVA) model containing terms for treatment and pooled center, with the baseline value as a covariate. Normality of the residuals was assessed with the Shapiro-Wilk statistic. Because the data violated the normality assumption in many instances, analyses were repeated using a rank ANCOVA with the same terms. In these instances, the rank ANCOVA was reported. The Fisher exact test was used for analyses where cells included fewer than 10 observations.
Rates of response (≥ 50% decrease in daily ratings after treatment) in both groups were compared using logistic regression, with additional terms for treatment and pooled center. The addition of body mass index and age to the logistic model did not add explanatory power (P value for regression coefficients > 0.5) and therefore were not included in the final model.
A similar statistical approach was used for additional secondary measures. For subjects who dropped out of the trial, scores were carried forward to the end of the double-blind treatment phase. Differences among continuous measures were estimated using ANCOVA models that included terms for treatment assignment, pooled center, and baseline score as a covariate. If the assumption of homogeneity of slopes was violated at the 0.05 level, a supportive rank ANCOVA was performed and was reported. We estimated that by recruiting 408 subjects, we would have 90% power to show a significant (P < .05) difference between groups of at least 6.5 (standard deviation 18) points on the total Daily Record of Severity of Problems, if a difference truly exists. For a repeated measures design, we assumed that 80% of the randomized subjects would have at least one response during the treatment period. Data were analyzed with SAS 8.12 (SAS Institute, Cary, NC).
Women were recruited by advertisement and referral. Of the 3,496 women who requested participation, 449 subjects were included in the intent-to-treat analysis. Screening flow is shown in Figure 1. Eligibility among women interested in participation was similar to that found in other clinical trials of premenstrual dysphoric disorder.13,17,19,20 The only statistically significant differences in baseline characteristics between the drospirenone/ethinyl estradiol and placebo groups were in mean age (31.0 versus 32.0 years, respectively, P = .048) and body mass index (26.1 versus 25.1 kg/m2, respectively, P = .015) (Table 1). Of the subjects randomized, 161 (69.4%) in the drospirenone/ethinyl estradiol group and 167 (76.6%) in the placebo group completed the study. There were no statistically significant differences (P = .473 by ANOVA) in the main outcome measure between treatment and placebo groups for those women who failed to complete the study. There was no significant difference in the number of women who completed active treatment and those who received placebo (P = .09 by Fisher exact test).
The on-treatment differences in cycle lengths for patients receiving active treatment and placebo were small and nonsignificant (27.3 days for drospirenone/ethinyl estradiol and 27.6 days for placebo; P = .23). Bleeding began an average of 3 days after cessation of active pills when subjects took drospirenone/ethinyl estradiol; therefore, approximately 3 days of the hormone-free interval were included in the premenstrual phase.
The total Daily Record of Severity of Problems symptom score in subjects receiving drospirenone/ethinyl estradiol decreased by 47%, from a mean (± standard deviation [SD]) of 77.40 (± 16.7) during the qualification cycles to an average of 41.2 (± 17.3) for the 3 treatment cycles; the corresponding reduction in the placebo group was 38%, from 78.1 (± 17.8) to 48.1 (± 21.1). The estimated adjusted mean change from baseline over 3 cycles in the total Daily Record of Severity of Problems symptom score was –37.49 for the drospirenone/ethinyl group and – 29.99 for the placebo group, for an adjusted mean difference of –7.50 (95% confidence interval [CI] –11.2 to –3.8; P = .001 by rank ANCOVA). The estimated improvement was significantly greater for subjects assigned to drospirenone/ethinyl estradiol than for those receiving placebo for all 11 distinct items in the Daily Record of Severity of Problems (Table 2). When the estimated difference was corrected for multiple testing (n = 11, P < .005), differences between groups for items “tired/fatigued” and “slept more” were no longer statistically significant.
Symptoms from the Daily Record of Severity were summed into clinically derived subscales reported by others.13,17,20 Active treatment reduced physical symptoms by –10.7 points, compared with –8.6 points in the placebo group (adjusted mean difference –2.1, 95% CI –3.3 to –0.95; P < .001 by rank ANCOVA). Similarly, mood symptoms were decreased by –19.2 and –15.3 points in active-treatment and placebo groups, respectively (adjusted mean difference –3.9, 95% CI –5.84 to –2.010; P = .003 by rank ANCOVA), and behavioral symptoms improved by –7.7 and –6.2, in active-treatment and placebo groups, respectively (adjusted mean difference –1.5, 95% CI –2.25 to –0.73; P < .001).
To determine whether subjects experienced “hormone withdrawal” symptoms or a shift in their symptomatic period, we compared the total Daily Record of Severity of Problems score from days 21–24 (active hormone administration) with that from days 25–28 (placebo administration) for women assigned to drospirenone/ethinyl estradiol. Terms in the ANOVA model included “subject,” “pill pack,” “pill-pack days,” and the “interaction of pill-pack days with pill pack.” There was no significant difference in model-estimated mean total Daily Record of Severity of Problems scores between these 2 intervals (38.4 ± 17.6 for days 21–24 versus 39.6 ± 18.8 for days 25–28, 95% CI –3.19 to 0.12, P = .069). This lack of difference is particularly notable because, aside from withdrawal, there is a tendency for women to experience more symptoms as menses approach.
Overall improvement according to the Clinical Global Impression-Improvement observer-rated scale was greater for subjects receiving active treatment than for those in the placebo group (2.2 versus 2.5, adjusted mean difference –0.30, 95% CI –0.55 to –0.05; P = .004 by rank ANCOVA). Similarly, self-rated improvement was greater in the active-treatment than in the placebo group (2.3 versus 2.5, adjusted mean difference –0.26, 95% CI –0.53 to 0.008; P = .014 by rank ANCOVA). Active treatment was significantly more effective than placebo in observer-rated (adjusted mean difference –1.9, 95% CI –3.8 to –.05; P = .023 by rank ANCOVA) and self-rated (–3.5, 95% CI –5.7 to –1.3, P = .004 by rank ANCOVA) Premenstrual Tension Scales scores. The median pretreatment and on-treatment values and the adjusted mean differences between groups for the Premenstrual Tension Scales are shown in Table 3.
Functional-impairment and quality-of-life domains were included as secondary outcome measures (Table 3). Compared with the placebo group, the subjects randomized to drospirenone/ethinyl estradiol experienced significant improvement in the functional items from the Daily Record of Severity of Problems: productivity (adjusted mean difference –0.33, 95% CI –0.55 to –0.12; P = .003 by rank ANCOVA); enhanced enjoyment in social activities (adjusted mean difference –0.34, 95% CI –0.55 to –0.12; P = .003 by rank ANCOVA); and better quality of relationships (adjusted mean difference –0.42, 95% CI –0.64 to –0.20; P < .001 by rank ANCOVA). Subjects using drospirenone/ethinyl estradiol reported statistically greater improvement in items 1–14 of the Endicott Quality of Life Enjoyment and Satisfaction Questionnaire (P = .047 by rank ANCOVA) compared with those subjects randomized to placebo, although statistical significance was not achieved for medication satisfaction (item 15) and overall life satisfaction (item 16). By cycle 3, 48.4% (95% CI 41.1% to 55.8%) of actively treated and 36.1% (95% CI 29.3% to 43.3%) of placebo-treated subjects met the criteria for response (odds ratio 1.7, 95% CI 1.1 to 2.6; P = .015). The number-needed-to-treat was 8 patients.
Adverse events considered by investigators to be possibly related to the study drug occurred in 118 (51.1%) subjects in the drospirenone/ethinyl estradiol group and 66 (30.3%) subjects in the placebo group. Adverse events that occurred in 5% or more of either treatment assignment are shown in Table 4. There was one severe adverse event (dysplasia) considered by the investigator to be possibly related to study drug in the active-treatment group. Thirty-five subjects receiving drospirenone/ethinyl estradiol and 9 subjects in the placebo group discontinued due to adverse events. The reasons most commonly given by subjects in the active treatment group for discontinuing the study included nausea (11 subjects, 31.4%), intermenstrual bleeding (8 subjects, 22.9%), and asthenia (7 subjects, 20.0%). No deaths were reported in the study.
There were no significant differences in serum potassium levels between active and placebo groups during the treatment period (4.3 mEq/L for both groups). Four subjects receiving active treatment and one subject in the placebo group had serum potassium levels above the normal range (≥ 5.5 mEq/L). One of the 4 subjects receiving drospirenone/ethinyl estradiol had a serum potassium level of 6.0 mEq/L or greater, but a repeat serum potassium value 4 days later was 4.6 mEq/L. None of the subjects with elevated potassium levels reported arrhythmias, dizziness, palpitations, bradycardia, tachycardia, or syncope at any point in the study.
This placebo-controlled, double-blind study shows that the new drospirenone-containing OCP formulation administered for 24 of 28 days in a cycle ameliorates symptoms associated with premenstrual dysphoric disorder. Active treatment was associated with a 47% reduction in the total Daily Record of Severity of Problems scale, while the median reductions in distinct items ranged from 45% to 62%. The palliative effect of OCPs, particularly for physical premenstrual symptoms, has been suggested by previous epidemiological and non–placebo-controlled research.7,21–23 However, before this trial, it was far less certain that an OCP formulation would be therapeutic for women suffering from the more severe condition, premenstrual dysphoric disorder. Moreover, women with premenstrual dysphoric disorder experience severe mood symptoms, and OCPs have been thought by some to worsen or cause symptoms of depression.24,25 Therefore, it is notable that in the current study, active treatment was associated with a 49% reduction in premenstrual depression. Our finding showing efficacy for a hormonal contraceptive intervention contrasts with the only 2 published placebo-controlled studies to evaluate OCP treatment for premenstrual symptoms. The first study enrolled 82 women with premenstrual syndrome, and the active treatment was a triphasic OCP (ethinyl estradiol 35 μg plus norethindrone 0.5 mg days 1–7 and 17–21 and 1.0 mg days 8–16).8 Differences between the OCP and placebo were minimal, although mood worsened in the active-treatment group during the hormone-free, postmenstrual period. In the second trial, either placebo or an OCP with drospirenone 3 mg and ethinyl estradiol 30 μg was administered in a 21/7 platform to 82 women who met premenstrual dysphoric disorder criteria.9 Although active treatment was associated with greater positive change than with placebo, the between-group differences did not reach statistical significance, possibly due to the modest sample size in the setting of a considerable placebo response rate (43%). The current study also used a formulation composed of drospirenone and ethinyl estradiol, but the dose of ethinyl estradiol was lower (ie, 20 μg rather than 30 μg), and the active drug was given for 24 rather than 21 days in a 28-day cycle. These differences in estrogen dose and shortened drug-free interval may contribute to the significant differences between active treatment (48.4%) and placebo (36.1%) found in this study. This finding corresponds to a number-needed-to-treat of 8 to achieve at least a 50% reduction in symptom severity (ie, response) in one patient.
U.S. Food and Drug Administration (FDA)-approved treatments for premenstrual dysphoric disorder are limited to selective serotonin reuptake inhibitors (SSRIs).13,20,26 In the current study, the overall effect size estimating the general therapeutic benefit is roughly similar to that found for SSRIs as reported in a recent meta-analysis.26 In addition, physical symptoms (eg, those found in item 11 of the Daily Record of Severity of Problems) responded well to treatment with drospirenone and ethinyl estradiol. These symptoms are less likely to respond to SSRIs,13 or they require higher doses for response.27,28
In the current study, as in many studies using SSRIs,17,20,27,28 the placebo group had high response rates. Although this could simply be due to subjects’ positive expectations, study methods, such as close contact with study staff and completing daily ratings of mood, may have contributed to this nonspecific response. The latter is particularly important because “mood monitoring” is a time-honored therapeutic intervention in behavioral medicine and contributes to improvement in emotional symptoms for a number of mood and anxiety disorders.
Attrition associated with adverse events occurred in 15% of subjects randomized to active treatment compared with 4% taking placebo. The most common adverse event in the active treatment group was intermenstrual bleeding (25.1% versus 4.6% for placebo). Intracyclic bleeding is a frequent occurrence during the initial cycles of treatment with OCPs containing low doses of estrogen, an effect that typically decreases and then remains stable after the first few cycles.29–31 Nausea was more common with active treatment compared with placebo, but the incidence of nausea (18.6%) is similar to that found with other low-dose OCP formulations.32 Relevant to this trial, complaints of nausea and breast tenderness decrease after the first few cycles of low-dose OCP use. There was no difference between the groups with respect to postbaseline serum potassium levels.
There are several possible limitations to the current study. First, OCPs potentially change the characteristics of the menstrual cycle, which could unblind study personnel as well as participants. This is of greater concern for the second and third cycles because subjects and clinicians are more likely to be unblinded by menstrual characteristics. In this study, the first set of premenstrual ratings were done before the first menstrual period. The active treatment-placebo difference occurring in the first study cycle was at least as large as, if not larger than, that for subsequent study cycles (data available upon request). Therefore, if unblinding occurred, it either worked against active treatment or was not of substantial magnitude. Second, it is possible that drospirenone/ethinyl estradiol shifted the symptomatic period to the hormone-free interval, as in a prior OCP study.8 However, our analyses of subjects assigned to drospirenone/ethinyl estradiol failed to show differences between premenstrual medication and premenstrual placebo days.
In summary, the OCP formulation with drospirenone 3 mg and ethinyl estradiol 20 μg administered in a 24/4 regimen was more effective than placebo at ameliorating premenstrual symptoms in women with premenstrual dysphoric disorder. This OCP formulation may have a unique role in women who both desire contraception and suffer from symptoms of premenstrual dysphoric disorder.
1. Wittchen HU, Becker E, Lieb R, Krause P. Prevalence, incidence and stability of premenstrual dysphoric disorder in the community. Psychol Med 2002;32:119–32.
2. Premenstrual dysphoric disorder. In: First MB, editor. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 2000. p. 771–4.
3. Halbreich U, Borenstein J, Pearlstein T, Kahn LS. The prevalence, impairment, impact, and burden of premenstrual dysphoric disorder (PMS/PMDD). Psychoneuroendocrinology 2003;28 suppl 3:1–23.
4. Hahn PM, Van Vugt, DA, Reid RL. A randomized, placebo-controlled, crossover trial of danazol for the treatment of premenstrual syndrome. Psychoneuroendocrinology 1995;20:193–209.
5. Brown CS, Ling FW, Andersen RN, Farmer RG, Arheart KL. Efficacy of depot leuprolide in premenstrual syndrome: effect of symptom severity and type in a controlled trial. Obstet Gynecol 1994;84:779–86.
6. Schmidt PJ, Nieman LK, Danaceau MA, Adams LF, Rubinow DR. Differential behavioral effects of gonadal steroids in women with and in those without premenstrual syndrome. N Engl J Med 1998;338:209–16.
7. Hylan TR, Sundell K, Judge R. The impact of premenstrual symptomatology on functioning and treatment-seeking behavior: experience from the United States, United Kingdom, and France. J Womens Health Gend Based Med 1999;8:1043–52.
8. Graham CA, Sherwin BB. A prospective treatment study of premenstrual symptoms using a triphasic oral contraceptive. J Psychosom Res 1992;36:257–66.
9. Freeman EW, Kroll R, Rapkin A, Pearlstein T, Brown C, Parsey K, et al. Evaluation of a unique oral contraceptive in the treatment of premenstrual dysphoric disorder. J Womens Health Gend Based Med 2001;10:561–9.
10. Wang M, Hammarback S, Lindhe BA, Backstrom T. Treatment of premenstrual syndrome by spironolactone: a double-blind, placebo-controlled study. Acta Obstet Gynecol Scand 1995;74:803–8.
11. Spona J, Elstein M, Feichtinger W, Sullivan H, Ludicke F, Muller U, et al. Shorter pill-free interval in combined oral contraceptives decreases follicular development. Contraception 1996;54:71–7.
12. Sulak PJ, Scow RD, Preece C, Riggs MW, Kuehl TJ. Hormone withdrawal symptoms in oral contraceptive users. Obstet Gynecol 2000;95:261–6.
13. Yonkers KA, Halbreich U, Freeman E, Brown C, Endicott J, Frank E, et al. Symptomatic improvement of premenstrual dysphoric disorder with sertraline treatment: a randomized controlled trial. Sertraline Premenstrual Dysphoric Collaborative Study Group. JAMA 1997;278:983–8.
14. Steiner M, Haskett RF, Carroll BJ. Premenstrual tension syndrome: the development of research diagnostic criteria and new rating scales. Acta Psychiatr Scand 1980;62:177–90.
15. Endicott J, Nee J, Harrison W, Blumenthal R. Quality of Life Enjoyment and Satisfaction Questionnaire: a new measure. Psychopharmacol Bull 1993;29:321–6.
16. Guy W. ECDEU Assessment manual for psychopharmacology. Publication ADM 76-338. Rockville (MD): National Institute of Mental Health, U.S. Department of Health, Education, and Welfare; 1976.
17. Cohen LS, Miner C, Brown EW, Freeman E, Halbreich U, Sundell K, et al. Premenstrual daily fluoxetine for premenstrual dysphoric disorder: a placebo-controlled, clinical trial using computerized diaries. Obstet Gynecol 2002;100:435–44.
18. First MB, Spitzer RL, Gibbon M, Williams JB. Structured Clinical Interview for DSM-IV Axis I Disorders—Patient Edition (SCID-I/P, Version 2.0). Washington, DC: American Psychiatric Association; 1994.
19. Steiner M, Steinberg S, Stewart D, Carter D, Berger C, Reid R, et al. Fluoxetine in the treatment of premenstrual dysphoria. Canadian Fluoxetine/Premenstrual Dysphoria Collaborative Study Group. N Engl J Med 1995;332:1529–34.
20. Halbreich U, Bergeron R, Yonkers KA, Freeman E, Stout AL, Cohen L. Efficacy of intermittent, luteal phase sertraline treatment of premenstrual dysphoric disorder. Obstet Gynecol 2002;100:1219–29.
21. Walker A, Bancroft J. Relationship between premenstrual symptoms and oral contraceptive use: a controlled study. Psychosom Med 1990;52:86–96.
22. Backstrom T, Hansson-Malmstrom Y, Lindhe BA, Cavalli-Bjorkman B, Nordenstrom S. Oral contraceptives in premenstrual syndrome: a randomized comparison of triphasic and monophasic preparations. Contraception 1992;46:253–68.
23. Backstrom T, Andreen L, Birzniece V, Bjorn I, Johansson IM, Nordenstam-Haghjo M, et al. The role of hormones and hormonal treatments in premenstrual syndrome. CNS Drugs 2003;17:325–42.
24. Slap GB. Oral contraceptives and depression: impact, prevalence and cause. J Adolesc Health Care 1981;2:53–64.
25. Kahn LS, Halbreich U. Oral contraceptives and mood. Expert Opin Pharmacother 2001;2:1367–82.
26. Wyatt KM, Dimmock PW, O’Brien PM. Selective serotonin reuptake inhibitors for premenstrual syndrome (Cochrane Review). In: The Cochrane Library, Issue 2, 2002. Oxford: Update Software.
27. Miner C, Brown E, McCray S, Gonzales J, Wohlreich M. Weekly luteal-phase dosing with enteric-coated fluoxetine 90 mg in premenstrual dysphoric disorder: a randomized, double-blind, placebo-controlled clinical trial. Clin Ther 2002;24:417–33.
28. Cohen LS, Soares CN, Yonkers KA, Bellew KM, Bridges IM, Steiner M. Paroxetine controlled release for premenstrual dysphoric disorder: a double-blind, placebo-controlled trial. Psychosom Med 2004;66:707–13.
29. Stone S. Clinical review of a monophasic oral contraceptive containing desogestrel and ethinyl estradiol. Int J Fertil Menopausal Stud 1993:38 suppl:117–21.
30. Endrikat J, Muller U, Dusterberg B. A twelve-month comparative clinical investigation of two low-dose oral contraceptivescontaining 20 micrograms ethinylestradiol/75 micrograms gestodene and 30 micrograms ethinylestradiol/75 micrograms gestodene, with respect to efficacy, cycle control, and tolerance. Contraception 1997;55:131–7.
31. Bachmann G, Sulak PJ, Sampson-Landers C, Benda N, Marr J. Efficacy and safety of a low-dose 24-day combined oral contraceptive containing 20 micrograms ethinylestradiol and 3 mg drospirenone. Contraception 2004;70:191–8.
32. Audet MC, Moreau M, Koltun WD, Waldbaum AS, Shangold G, Fisher AC, et al. Evaluation of contraceptive efficacy and cycle control of a transdermal contraceptive patch vs an oral contraceptive: a randomized controlled trial. JAMA 2001;285:2347–54.
Figure. No caption available.