Journal
ofPsychosomatic
Research,
Vol.
36. No.
3, pp. 257-266,
1992.
0022-3999192
0
Printed in Great Britain.
SS.OO+.OO
1992 Pergamon Press plc
A PROSPECTIVE TREATMENT STUDY OF PREMENSTRUAL SYMPTOMS USING A TRIPHASIC ORAL CONTRACEPTIVE CYNTHIA A. GRAHAM* and BARBARA B. SHERWIN*? (Received 18 February 1991; accepted in revised form
25 September 1991)
Abstract-Eighty-two women with complaints of moderate to severe premenstrual symptoms were recruited for a double-blind, controlled trial of a triphasic oral contraceptive (o.c.). Subjects made daily ratings of symptoms for at least one baseline cycle and were then randomly assigned to receive either placebo or O.C. for three months. Twenty-three women dropped out of the study (18 o.c., 5 placebo), 13 failed to show prospective confirmation of moderate to severe premenstrual symptoms, and one placebo subject had an anovulatory cycle. Forty-five women with prospectively-confirmed premenstrual changes (20 o.c., 25 placebo) completed the study. Premenstrual breast pain and bloating were significantly reduced with active treatment compared to placebo @ < 0.03) but there were no beneficial effects of the O.C. over placebo for any of the mood symptoms. Women who received o.c.s reported decreased sexual interest after starting treatment and this effect was independent of any adverse influence on mood.
INTRODUCTION CURRENT state of research and treatment related to the premenstrual syndrome (PMS) is characterized by confusion and controversy. Various hormonal mechanisms have been proposed as causal factors in PMS but to date no clear conclusions regarding etiology can be drawn. There is an assumption that premenstrual changes are related to ovulatory cycles and that PMS is unlikely to occur in anovulatory cycles [ 1 ] . Suppression of ovulation has therefore been an important rationale for the use of hormonal treatments for PMS. One method of inhibiting ovulation is by using oral contraceptives (o.c.s), yet there is little systematic evidence of the effect of 0.c.s on PMS [2]. The few studies which have looked at the cyclical changes experienced by women using steroidal contraceptives suffer from a number of methodological shortcomings including the lack of controls for placebo effects, inappropriate subject selection, and infrequent and retrospective assessment of symptoms. Studies comparing women who are established on o.c.s with non-pill users are of limited value because women whose symptoms are worsened by o.c.s or who develop other adverse effects would be more likely to discontinue the pill. Moreover, most earlier studies have used o.c.s with much higher doses of synthetic hormones than those currently available. To date there have been no placebo-controlled studies of the effects of 0.c.s on women with prospectively-confirmed PMS. An important observation has been that women appear to vary in their response THE
*Department of Psychology, McGill University, Montreal, Canada. l-Department of Obstetrics and Gynecology, SMBD-Jewish General Hospital, McGill University, Montreal, Canada. Address reprint requests to: Dr C. A. Graham, MRC Reproductive Biology Unit, Behaviour Research Group, The Kennedy Tower, Royal Edinburgh Hospital, Morningside Park, Edinburgh, EHlO 5HF, U.K. 251
258
C. A. GRAHAM and B. B. SHERWIN
to o.c.s, with some reporting fewer premenstrual symptoms and others experiencing no change or intensified symptoms [3] . However, the identifying characteristics of women with different treatment responses have not been established. The present study was designed as a double-blind, placebo-controlled trial of the effects of a triphasic O.C. on women seeking treatment for moderate to severe premenstrual symptoms. The major goals were: (1) to assess the efficacy of a triphasic O.C. as a treatment for moderate to severe premenstrual symptoms; (2) to determine whether differential response of women with PMS to an O.C. might be associated with a specific symptom profile and; (3) to assess the effect of blocking ovulation on the timing and severity of premenstrual symptoms in an effort to elucidate the mechanisms underlying PMS. METHODS
Between April 1987 and June 1988, women seeking treatment for self-reported moderate to severe physical and psychological premenstrual symptoms were invited to participate through newspaper advertisements or articles about the research. Subject inclusion criteria required that women: (1) provide a retrospective account of premenstrual symptoms of sufficient severity so as to lead to impaired social or occupational functioning; (2) rate a minimum of one mood and one physical symptom as at least ‘moderate’ (i.e. s 4 on a six-point scale) on the Premenstrual Assessment Form; (3) be between the ages of 18-35 yr; (4) have no current and/or chronic medical or gynecologic disorder; (5) not be currently receiving treatment from a mental health professional; (6) not be taking psychotropic or hormone medication; (7) have regular menstrual cycles (23-35 days); (8) not be pregnant or planning a pregnancy over the course of the study; (9) had not used o.c.s in the previous 3 months and; (10) if sexually active, be using a nonhormonal form of contraception. In accordance with published guidelines on the use of o.c.s, [4] all of the women who volunteered for the study were carefully screened by a gynecologist to rule out any gynecological disorder and/or contraindications to O.C. use. Measures Baseline assessmenf and selection. (a) A Menstrual History Form which included questions on O.C. history, menstrual cycle characteristics, and the pattern and severity of premenstrual changes was administered at the initial interview. (b) The Premenstrual Assessment Form (PAF) [ 51 is a 9%item retrospective questionnaire, designed to measure the severity of change from the nonpremenstrual state during the three preceding premenstrual periods. Each item is rated on a six-point scale, from ‘no change’ to ‘extreme change’. The PAF was used for the initial selection of subjects for the study. Treutmenf phase. (a) A shortened, 15-item version of the Daily Ratings Form (DRF) [6] was used to monitor changes daily throughout the course of the study. Items include physical symptoms (edema, breast pain, headaches, low energy, increased appetite and less sexual interest), psychological symptoms (irritable, anxious, mood swings and depressed), and behavioral changes (more sleep, less work, stay at home). Two positive changes (‘increased enjoyment and efficiency’ and ‘more sexual interest’) are also included. The scale for each symptom or behavior ranges from 1 (‘not present at all’) to 6 (‘extreme’). Each day’s ratings were made on a separate page. Subjects also indicated whether they were menstruating or spotting, and noted any physical illness, medication use, or significant life events. (b) Nine visual analogue scales (VAS) which indicate the intensity of a mood or physical symptom during the previous 24 hr were also completed daily. Five mood states (‘cheerful and happy’, ‘fatigued and tired’, ‘tense and anxious’, ‘irritable’, and ‘depressed and unhappy’) and four physical items (‘bloated’, ‘breast pain’, ‘sexual interest’, and ‘headache’) were included. Post-trearmenr phase. (a) The Post-treatment Questionnaire assessed subjects’ beliefs about the treatment they had received. Women were asked to indicate whether they thought they had received the active treatment, inactive tablets, or both active and inactive tablets at different times during the study and if they had any idea about what type of hormone(s) were contained in the tablets. Drugs The triphasic
O.C. used (Synphasic,
Syntex Laboratories,
Mississauga,
Canada)
contained
a fixed dose
Oral contraceptives
and PMS
259
of ethinyl estradiolO.035 mg from days l-21 and norethindrone, 0.5 mg during days l-7, 1 mg during days 8-16, and 0.5 mg during days 17-21. Subjects in the placebo group received 21 tablets containing lactose and cornstarch. Placebos were contained in blister packets identical to the active preparation. To determine whether ovulation occurred, 10 ml of blood were collected from each subject once during the premenstrual, menstrual, and postmenstrual phases of the third treatment cycle. Although it would clearly have been useful to have confirmed the presence of ovulation throughout the study, it was felt that the additional demands on subjects which this procedure would entail might affect compliance. Subjects were instructed to take the study medication approximately 3 hr prior to the drawing of the blood samples [7] The blood was immediately centrifuged and the plasma stored at -20°C until the completion of the study. Progesterone was measured by radioimmunoassay using a no-extraction, solid phaselz51 radioimmunoassay (Diagnostic Products Corp., Los Angeles, CA, U.S.A.). Plasma concentrations of progesterone greater than 9.6 nmols/l on days -3 to -6 of the cycle were regarded as evidence of ovulation. The primary purpose in obtaining blood samples was to detect ovulation in the group receiving placebo, since the women taking o.c.s would not be expected to ovulate. Procedure A two (treatment groups) X four (time periods) design was used. Following one cycle of baseline monitoring, subjects were randomly assigned to either the O.C. or placebo group for three cycles. At an initial interview women who met the selection criteria signed a consent form that had been approved by ethics committees of the university hospital and the community health center where the study was based. Subjects were told that the aim of the study was to investigate the effects of a ‘low-dose hormone’ on PMS. The treatment was specifically not introduced as an o.c.. They were also informed that they might receive inert tablets instead of the active medication and that their treatment could be changed during the course of the study. No instruction was given to subjects regarding the possible experience of side effects; if during the first treatment cycle women reported side effects such as breast pain or spotting, they were reassured that these were not harmful and would most likely subside with continuing use. The DRF and VAS were completed each evening and returned by mail each week. Women who failed to return their ratings were contacted by telephone. DRF ratings for the baseline cycle were scored to assess whether these confirmed subjects’ retrospective reports of moderate to severe premenstrual changes using the method employed by Endicott et al. [ 61 Change scores were calculated for each of the DRF items by contrasting the mean score of the three highest consecutive days of the last five premenstrual days with mean scores of the five postmenstrual days. The criteria used for prospective confirmation required that at least two DRF items show a two-point level of change during the premenstrual compared to the postmenstrual phase. This level of change, although somewhat arbitrary, was chosen because it is roughly equivalent to the criteria established at an NIMH workshop (1983). Two additional criteria were applied: on the above items, the mean premenstrual scores must have been 2 4 (at least moderate) and the mean postmenstrual scores must have not exceeded 3 (mild), unless there was a reported stressful event which could have accounted for the elevated score. Following the baseline cycle, subjects were given a one month supply of either O.C. or placebo and were instructed to begin taking one tablet per day on the first Sunday after menstrual bleeding began. No piils were taken from days 22-28, after which a new packet was started. This is the standard method of administration recommended for this preparation. During monthly visits to the research unit, a one month supply of daily forms and of medication were distributed. At the end of the third treatment cycle, all women completed the Post-treatment Questionnaire and were then debriefed. Statistical
analyses
Each cycle was divided into the following four phases: (1) premenstrual: the five days before the onset of menses; (2) menstrual: all days on which bleeding occurred; (3) postmenstrual: the five days after the cessation of bleeding and; (4) intermenstrual: the remainder of the cycle. A FORTRAN program developed for this study (Amsel and Amsel, unpublished computer program 1988) computed the mean score of every DRF and VAS item in each of the four cycle phases for each subject. SPSS-X three-way analyses of variance (ANOVA) (group x cycle x phase) with repeated measures (on phase and cycle) were performed on the mean phase scores for every symptom in each of the four cycles. Post-hoc tests, using Newman-Keuls method of multiple comparisons, were carried out on all interaction effects which were significant. To investigate the possibility that pretreatment variables might predict response to treatment, subjects in O.C. and placebo groups were divided into different subgroups based on whether or not they had prospectively confirmed on four DRF symptoms (depression, irritability, breast pain and edema) during the baseline cycle. A series of four-way ANOVAs (group X subgroup x cycle x phase) were carried out on the mean phase scores for each of these variables.
C. A. GRAHAM and B. B. SHERWIN
260 Student’s r-tests and Chi-squares treatment Questionnaire.
were used to analyse
the retrospective
questionnaires
and the Post-
RESULTS
Two-hundred and twelve women responded to the recruitment notices. Of these, 54 decided against taking part after an initial telephone interview and 76 were ineligible for the trial, primarily because they had physical or psychiatric disorders (N = 24), symptoms of insufficient severity (N = 13), were taking other medications (N = 12), or failed to satisfy the age restrictions (N = 10). Eighty-two women who met the selection criteria were entered into the study. Table I contains demographic and menstrual cycle data for this overall sample.
TABLE L-SUBJECT
CHARACTERISTICSfN = 82)
Percentage Marital status: married/cohabiting single divorced/separated Education: 15+ yr 12-15 yr < 12 yr Employment: full-time part-time student homemaker No. of pregnancies: 0 1 2 3 or more Previously sought treatment Yes No
of sample
40.5 53.2 6.3 48.1 41.7 10.2 55.1 12.8 21.8 10.3 52.5 20.7 12.2 14.6 for PMS:
Mean age: 29.5 yr (ft SD 5.04) Mean cycle length: 28.5 days (*
57.3 42.7
SD
2.7)
Consistent with previous treatment studies, not only was the attrition rate fairly high overall, but it was higher in the O.C. group than the placebo group (x2 = 8.7, p < 0.01) (Table II). Nine women dropped out during the baseline cycle (five o.c., four placebo), nine in the first month of treatment (eight o.c., one placebo) and five during the second treatment cycle (all o.c.). The 59 subjects (23 o.c., 36 placebo) who completed the trial were compared with the women who dropped out (N = 23) on a number of variables. Completers were employed in higher-status occupations compared to drop-outs @ < 0.007). The drop-outs had significantly higher severity scores on 4 of the 18 subscales of the retrospective PAF: lability @ < 0.04), hysteroid depressive features @ < 0.03), anxiety (p < O.Ol), and water retention @ < 0.04). No other significant pretreatment differences between completers and drop-outs were apparent. As the O.C. and
Oral contraceptives
and PMS
261
placebo groups did not differ on any of the above variables, taking the active pill increased the likelihood that a woman would drop out of the study. TABLE
II.-REASONS
GIVEN FOR DROPPING OUT (N=23)
O.C. Physical illness Decided against taking part Moving away Side effects* spotting (N = 10) breast pain (N = 6) nausea (N = 5) weight gain (N = 2) abdominal cramps (N = 2) pains in legs (N = 1) water retention (N = 1) yeast infection (N = 1) *Some
subjects
reported
(N=
18)
Placebo
3 1 1 13
(N = 5) 2 2 1
more than one side effect.
Table III lists the incidence of side effects reported by the pill subjects who completed the study (N = 23). The number of days of spotting or breakthrough bleeding was highest in the first pill cycle and occurred less frequently in the third treatment cycle (mean = 3.74 (* 4.53) vs 1.75 (* 3.1 l), respectively). TABLE III.-INCIDENCE OF SIDE EFFECTS IN THE ORAL CONTRACEPTIVE SUBJECTS (N=23)WHO COMPLETED THE STUDY
Spotting/breakthrough Nausea Abdominal cramps Breast pain/swelling Acne Dizziness
bleeding
N
%
16 7 7 4 4 4
69.6 30.4 30.4 17.4 17.4 17.4
There were no significant differences between the O.C. and placebo groups in educational background, occupational status, age, menstrual cycle length, or in the proportion of women in each group who met criteria for prospective confirmation. Eighty-three percent of the overall sample confirmed on at least two DRF items in the baseline cycle. Of the 82 subjects studied, 13 failed to meet criteria for prospective confirmation, 1 had progesterone levels which indicated an anovulatory cycle, and 23 did not complete the study. The remaining 45 women (20 o.c., 25 placebo) with prospectively confirmed premenstrual symptoms who completed the trial were included in the treatment analyses. Both O.C. and placebo groups showed a significant clinical improvement on every symptom except headache. Post-hoc analyses of cycle x phase interactions showed a significant reduction in premenstrual symptom scores between the baseline cycle and the third treatment cycle. There was also a decrease in menstrual phase scores for four variables (‘mood swings’, ‘more sleep’, ‘unhappy’ and ‘tense’) in the second
C. A. GRAHAM and B. B. SHERWIN
262
treatment cycle compared to the first month on treatment, irrespective of treatment group. However, scores in postmenstrual and intermenstrual phases did not change across the trial. Significant group x cycle x phase effects occurred for ratings of breast pain on both the DRF @ < 0.03) and VAS @J < 0.04). Post-hoc analyses yielded very similar results for this symptom on both measures. In the O.C. group, ratings of premenstrual breast pain were significantly lower during the third treatment cycle compared both to baseline @ < 0.05) and to the first treatment cycle (p < 0.01). No significant changes in premenstrual breast pain ratings occurred across the trial in the placebo group (Fig. 1). W Baseline
T
0
3rd. treatment cycle
TT
tV&S
O.C.
F&T
INI&
i
F&T
lNliF7
Placebo
FIG. 1. Mean VAS breast pain ratings (f SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, premenstrual scores lower in third treatment cycle compared to baseline @ < 0.05); other comparisons reported in text.
There was also a differential treatment response for DRF edema scores, although this finding is qualified by the fact that there was a pretreatment difference on this variable, with the O.C. group having higher premenstrual scores than the placebo group. Analyses of the significant group x cycle x phase effect @ < 0.02) revealed a significant decrease in edema premenstrually during the second @ < 0.01) and third @ < 0.01) treatment cycles for the O.C. group compared to their scores at baseline. In contrast, premenstrual edema scores of the placebo group remained stable across the trial (Fig. 2). A comparison of the four phase scores within each cycle also showed a greater reduction in premenstrual edema and breast pain in the O.C. group. Although at baseline premenstrual edema and breast pain ratings were significantly higher than postmenstrual scores for both groups @ < O.Ol), by the third treatment cycle only the placebo group showed any between-phase differences on these variables. On one variable, sexual interest, the O.C. group showed a significant deterioration during treatment. For the VAS variable ‘sexual interest’ there was a significant group X cycle X phase effect @ < 0.007); in the O.C. group, sexual interest scores were lower during both the menstrual (p < 0.01) and postmenstrual 0, < 0.01) phases of the third treatment cycle compared to baseline (Fig. 3). There were no significant
Oral
contraceptives and PMS
263
changes in sexual interest scores during treatment in the placebo group. Once again, there was one pretreatment group difference, with the O.C. group showing higher sexual interest than the placebo group during the menstrual phase @ < O.Ol), but no other significant group differences during the baseline cycle. 5
I
4
1
n Baseline
T
0
h&S
INI&
F&T
i
PA
3rd. treatment cycle
bit43
P&T
ltiR
Placebo
O.C.
FIG. 2. Mean premenstrual edema ratings (+ SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, premenstrual scores lower in third treatment cycle compared to baseline @ < 0.01); at baseline, premenstrual scores higher in O.C. group than in placebo @ < 0.05); other comparisons reported in text
n Baseline 0
3rd. treatment cycle
11 dlL _TT
l
TT
i
PRE
MENS
O.C.
fa3T
INTER
I
PRE
MENS
POST
INTER
Placebo
FIG. 3. Mean VAS sexual interest ratings (+ SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, menstrual and postmenstrual scores lower in third treatment cycle compared to baseline @ < 0.01); other comparisons reported in text.
When women were divided into subgroups based on whether they had prospectively confirmed on specific DRF items (depression, irritability, breast pain, edema) only one symptom-depression-showed any relationship to treatment response. For the sake of clarity, women who had prospectively confirmed on depression at
264
C. A. GRAHAM and B. B. SHERWIN
baseline were designated ‘depressed’ and those who failed to meet confirmation criteria were called ‘non-depressed’. Thus, there were four subgroups: ‘depressed’ O.C. (N= 12), ‘non-depressed’ O.C. (N= 8), ‘depressed’ placebo (N= lo), and ‘non-depressed’ placebo (N = 15). The ‘depressed’ O.C. group reported greater improvement in three behavioral symptoms; specifically, these women were less impaired at work @ < 0.02), required less sleep @ < 0.05), and had more energy (p < 0.04) premenstrually whilst taking O.C.S. In contrast, women with premenstrual depression who received placbeo, and those without a tendency to cyclical depression in both treatment groups, either showed no change in these symptoms or less consistent improvement. On the Post-treatment Questionnaire subjects in the O.C. group were more likely to identify which treatment they had received compared to the placebo subjects @ < 0.005). Fourteen (38.9%) placebo subjects believed that they had received both active and inactive tablets and seven (19.4%) incorrectly guessed that they had taken active tablets throughout the study. In comparison, four women (17.4%) in the O.C. group incorrectly guessed that they had received both active and inactive tablets and one (4.3 %) thought that she had taken a placebo throughout the study. Only 24 % of the overall sample correctly guessed that the active tablets contained a synthetic estrogen and progesterone; 45 % stated that they did not know what type of hormone(s) were contained in the active tablets and 3 1% gave an incorrect response.
DISCUSSION
To our knowledge, this is the first double-blind, placebo-controlled trial of o.c.s in women with prospectively-confirmed PMS. Several other features of the experimental design increased the likelihood that any observed differences in the effects of the O.C. and placebo could be attributed to direct hormonal effects on the behavior. Prior studies on the relationship between o.c.s and PMS have looked at premenstrual symptoms in women already taking o.c.s and have not assessed symptom change before and after starting O.C.S. Moreover, because our subjects were not aware that the treatment was an o.c., factors such as attitudes toward the pill, personality characteristics which affect choice of contraception, etc. were controlled. These data do not support the commonly-held theory that premenstrual mood symptoms are a consequence of hormonal fluctuations of the ovarian cycle. However, the fact that breast pain and edema improved to a greater extent on the active pill than placebo suggests that these two symptoms may be related in some way to ovulation and luteal function. In agreement with this, a recent comparison of women using and not using different types of O.C.S. found breast tenderness was the only symptom to show a clear difference between groups [ 1 ] . A specific aim of this study was to relate response to O.C. administration with different types of premenstrual changes in the hope that guidelines for treatment decisions might emerge. Our findings suggest that a triphasic O.C. may provide relief for premenstrual breast pain and edema. However, most women who seek treatment for PMS do so because of an inability to cope with the mood-related changes rather than the physical symptoms [ 81 and cyclical mood changes improved equally with O.C. and placebo treatment. These findings are consistent with those of Cullberg [9] who reported that
Oral contraceptives
and PMS
265
premenstrual irritability and depression were not significantly relieved by o.c.s compared to placebo. In agreement with previous reports [lo], premenstrual mood symptoms seem to be more responsive to placebo than physical symptoms. Previous studies have suggested that the response afforded by placebo is transient and lasts only two or three cycles [ 111. In the present study there was no suggestion of any abatement of the response to placebo by the third treatment cycle, although it is possible that with a longer treatment period, this may have occurred. In a recent cross-over study of estradiol patches in the treatment of PMS [ 121, there was a response to both placebo and active treatment during the first three months of use. However, after three months, patients who switched from placebo to active treatment maintained their initial gains while the scores of those who changed from active treatment to placebo generally deteriorated. In addition to the positive effects of the active treatment on breast pain and edema, there are also possible negative effects of the O.C. which must be considered. It has previously been suggested that women who are prone to PMS are more likely to experience side effects using o.c.s [ 131 . As there is very little information available on what proportion of women taking o.c.s experience side effects, it is difficult to determine whether the present sample had more difficulties tolerating 0.c.s than women without premenstrual complaints. Side effects (particularly spotting) were also fairly commonly reported by completers, especially in the first cycle on O.C.S. A clinical impression was that the women who dropped out, often after a very short time on the tablets, may have been more ambivalent about taking hormones. On the other hand, it may be the case that the women who withdrew from the study did constitute a subgroup who were more likely to experience side effects. The changes in sexual interest induced by the pill were unexpected and of particular interest, given the dearth of evidence on the effects of the triphasic 0.c.s on sexuality. After starting on the triphasic o.c., women showed less sexual interest during the menstrual and postmenstrual phases (corresponding to the pill-free week and the first few days of the next pill cycle). These findings are in contrast with two other studies [ 14, 151 showing that O.C. users report highest sexual interest in the postmenstrual week. A caveat with regard to these data is related to the higher baseline levels of sexual interest in the O.C. group. Nevertheless, the findings provide interesting information with regard to the relationship between sexual interest and mood. These negative effects of the O.C. on sexual interest could not be attributed to a negative influence on mood and suggest a direct hormonal effect of the O.C. on sexuality, although the mechanism by which this may have occurred is unclear. The finding that sexual interest and mood are dissociable is supported by results of a previous placebo-controlled study in which women who had undergone sterilization experienced decreased sexual interest on the O.C. but no alteration in depression scores [ 161. Given the continued uncertainty in the literature on whether o.c.s are associated with depressive mood change [ 21 , it is interesting that none of the women receiving 0.c.s reported any negative effects on mood. In fact, a subgroup of women with a tendency to premenstrual depression responded better to 0.c.s than women without such cyclical depression. Because the number of subjects in the subgroup analyses were small, these findings should be regarded as tentative. However, given the fact
C. A. GRAHAM and 8. B. SHERWIN
266
that there is currently no basis on which to predict a woman’s response findings, if replicated, would have important clinical implications. .4c/tnowledgemenrs-Support
Council
of Canada
for this research
(No. MA-8707)
was provided
by grants
from
to o.c.s,
the Medical
these
Research
and from Fonds pour la Formation
de Chercheurs et 1’Aide a la Graham was supported by a Medical Canada provided the O.C. and placebo
Recherche (No. EQ3632) awarded to B. B. Sherwin. C. A. Research Council of Canada Studentship. Syntex Laboratories, tablets. The authors would like to thank Natalie Rubin for her help with data collection
and analyses.
REFERENCES
1. WALKER
A, BANCROFT J. Relationship between premenstrual symptoms and oral contraceptive use: A controlled study. Psychosom Med 1990; 52: 86-96. 2. BANCROFT J, SAR~ORIUS N. The effects of oral contraceptives on well-being and sexuality. Oxford
Rev Reprod Biol 1990; 12: 57-92. 3. FREEMAN EW, SONDHEIMER S, WE~NBAUM PJ, RICKELS K. Evaluating 4.
5. 6. 7.
8. 9.
premenstrual symptoms in medical practice. Obsfer Gynecol 1985; 65: 500-505. HEALTH AND WELFARE CANADA. Oral Contraceptives. Government Report, 1985. HALBREICH U, ENDICOTT J, SCHACHT S, NEE J. The diversity of premenstrual changes as reflected in the Premenstrual Assessment Form. Acra Psychiat Stand 1982; 65: 46-65. ENDICOTT J, NEE J, COHEN J, HALBREICH U. Premenstrual changes: Patterns and correlates of daily ratings. J Affect Disord 1986; 10: 27-35. SAPERSTEIN S, EDCREN RA, ELLIS DJ, LEE GJL, KUSHINSKY S, OLMXED A, MROSZCZAK E. Bioequivalence of norethindrone and ethinyl estradiol for two different weight tablets with the same hormonal content. Conrracepption 1986; 33: 547-557. CORNEY RH, STANTON R. A survey of 658 women who report symptoms of premenstrual syndrome. J Psychosom Res 1991; 35: 71-82. CULLBERG J. Mood changes and menstrual symptoms with different gestagen/estrogen combinations.
Acra Psychiar Stand (Suppl) 1972; 236: 1-86. IO. METCALF MG, HUDSON SM. The premens!rual syndrome: Selection of women for treatment trials. J P.~pchosom Res 1985; 29: 631-638. Il. MAGOS AL, BRINCAT M, STUDD JWW. Treatment of the premenstrual syndrome by subcutaneous oestradiol implants and cyclical oral norethisterone: Placebo controlled study. Br Med J 1986; 292: 1620-1633. 12. WATSON NR, Sruoo JWW, SAVVAS M, GARNET.~ T, BABER RJ. Treatment of severe premenstrual syndrome with oestradiol patches and cyclical oral norethisterone. Lancer 1989; iii: 730-?32. 13. DENNERSTEIN L, SPENCER-GARDNER C, BROWN JB, SMITH MA, BURROWS GD. Premenstrual tension: Hormonal profiles. J Psychosom Obsret Gynecol 1984; 3: 37-51. 14. ADAMS DB, GOI.D AR, BURT AD. Rise in female-Initiated sexual activity and its suppression by
oral contraceptives.
N Errgl J Med 1978; 299:
15. WARNER P, BANCROF~ J. Mood,
sexuality,
1145-I
150.
oral contraceptives,
and the menstrual cycle. J Psyrhosom
Rrs 1988; 32: 417-427. 16. LEETON J, MCMASTER R, WORSLEY A. The effects on sexual response and mood after of women taking long-term oral contraception: Results of a double-blind cross-over N Z Obstet Gynaecol 1978; 18: 194-197.
sterilization study. Aust
ofPsychosomatic
Research,
Vol.
36. No.
3, pp. 257-266,
1992.
0022-3999192
0
Printed in Great Britain.
SS.OO+.OO
1992 Pergamon Press plc
A PROSPECTIVE TREATMENT STUDY OF PREMENSTRUAL SYMPTOMS USING A TRIPHASIC ORAL CONTRACEPTIVE CYNTHIA A. GRAHAM* and BARBARA B. SHERWIN*? (Received 18 February 1991; accepted in revised form
25 September 1991)
Abstract-Eighty-two women with complaints of moderate to severe premenstrual symptoms were recruited for a double-blind, controlled trial of a triphasic oral contraceptive (o.c.). Subjects made daily ratings of symptoms for at least one baseline cycle and were then randomly assigned to receive either placebo or O.C. for three months. Twenty-three women dropped out of the study (18 o.c., 5 placebo), 13 failed to show prospective confirmation of moderate to severe premenstrual symptoms, and one placebo subject had an anovulatory cycle. Forty-five women with prospectively-confirmed premenstrual changes (20 o.c., 25 placebo) completed the study. Premenstrual breast pain and bloating were significantly reduced with active treatment compared to placebo @ < 0.03) but there were no beneficial effects of the O.C. over placebo for any of the mood symptoms. Women who received o.c.s reported decreased sexual interest after starting treatment and this effect was independent of any adverse influence on mood.
INTRODUCTION CURRENT state of research and treatment related to the premenstrual syndrome (PMS) is characterized by confusion and controversy. Various hormonal mechanisms have been proposed as causal factors in PMS but to date no clear conclusions regarding etiology can be drawn. There is an assumption that premenstrual changes are related to ovulatory cycles and that PMS is unlikely to occur in anovulatory cycles [ 1 ] . Suppression of ovulation has therefore been an important rationale for the use of hormonal treatments for PMS. One method of inhibiting ovulation is by using oral contraceptives (o.c.s), yet there is little systematic evidence of the effect of 0.c.s on PMS [2]. The few studies which have looked at the cyclical changes experienced by women using steroidal contraceptives suffer from a number of methodological shortcomings including the lack of controls for placebo effects, inappropriate subject selection, and infrequent and retrospective assessment of symptoms. Studies comparing women who are established on o.c.s with non-pill users are of limited value because women whose symptoms are worsened by o.c.s or who develop other adverse effects would be more likely to discontinue the pill. Moreover, most earlier studies have used o.c.s with much higher doses of synthetic hormones than those currently available. To date there have been no placebo-controlled studies of the effects of 0.c.s on women with prospectively-confirmed PMS. An important observation has been that women appear to vary in their response THE
*Department of Psychology, McGill University, Montreal, Canada. l-Department of Obstetrics and Gynecology, SMBD-Jewish General Hospital, McGill University, Montreal, Canada. Address reprint requests to: Dr C. A. Graham, MRC Reproductive Biology Unit, Behaviour Research Group, The Kennedy Tower, Royal Edinburgh Hospital, Morningside Park, Edinburgh, EHlO 5HF, U.K. 251
258
C. A. GRAHAM and B. B. SHERWIN
to o.c.s, with some reporting fewer premenstrual symptoms and others experiencing no change or intensified symptoms [3] . However, the identifying characteristics of women with different treatment responses have not been established. The present study was designed as a double-blind, placebo-controlled trial of the effects of a triphasic O.C. on women seeking treatment for moderate to severe premenstrual symptoms. The major goals were: (1) to assess the efficacy of a triphasic O.C. as a treatment for moderate to severe premenstrual symptoms; (2) to determine whether differential response of women with PMS to an O.C. might be associated with a specific symptom profile and; (3) to assess the effect of blocking ovulation on the timing and severity of premenstrual symptoms in an effort to elucidate the mechanisms underlying PMS. METHODS
Between April 1987 and June 1988, women seeking treatment for self-reported moderate to severe physical and psychological premenstrual symptoms were invited to participate through newspaper advertisements or articles about the research. Subject inclusion criteria required that women: (1) provide a retrospective account of premenstrual symptoms of sufficient severity so as to lead to impaired social or occupational functioning; (2) rate a minimum of one mood and one physical symptom as at least ‘moderate’ (i.e. s 4 on a six-point scale) on the Premenstrual Assessment Form; (3) be between the ages of 18-35 yr; (4) have no current and/or chronic medical or gynecologic disorder; (5) not be currently receiving treatment from a mental health professional; (6) not be taking psychotropic or hormone medication; (7) have regular menstrual cycles (23-35 days); (8) not be pregnant or planning a pregnancy over the course of the study; (9) had not used o.c.s in the previous 3 months and; (10) if sexually active, be using a nonhormonal form of contraception. In accordance with published guidelines on the use of o.c.s, [4] all of the women who volunteered for the study were carefully screened by a gynecologist to rule out any gynecological disorder and/or contraindications to O.C. use. Measures Baseline assessmenf and selection. (a) A Menstrual History Form which included questions on O.C. history, menstrual cycle characteristics, and the pattern and severity of premenstrual changes was administered at the initial interview. (b) The Premenstrual Assessment Form (PAF) [ 51 is a 9%item retrospective questionnaire, designed to measure the severity of change from the nonpremenstrual state during the three preceding premenstrual periods. Each item is rated on a six-point scale, from ‘no change’ to ‘extreme change’. The PAF was used for the initial selection of subjects for the study. Treutmenf phase. (a) A shortened, 15-item version of the Daily Ratings Form (DRF) [6] was used to monitor changes daily throughout the course of the study. Items include physical symptoms (edema, breast pain, headaches, low energy, increased appetite and less sexual interest), psychological symptoms (irritable, anxious, mood swings and depressed), and behavioral changes (more sleep, less work, stay at home). Two positive changes (‘increased enjoyment and efficiency’ and ‘more sexual interest’) are also included. The scale for each symptom or behavior ranges from 1 (‘not present at all’) to 6 (‘extreme’). Each day’s ratings were made on a separate page. Subjects also indicated whether they were menstruating or spotting, and noted any physical illness, medication use, or significant life events. (b) Nine visual analogue scales (VAS) which indicate the intensity of a mood or physical symptom during the previous 24 hr were also completed daily. Five mood states (‘cheerful and happy’, ‘fatigued and tired’, ‘tense and anxious’, ‘irritable’, and ‘depressed and unhappy’) and four physical items (‘bloated’, ‘breast pain’, ‘sexual interest’, and ‘headache’) were included. Post-trearmenr phase. (a) The Post-treatment Questionnaire assessed subjects’ beliefs about the treatment they had received. Women were asked to indicate whether they thought they had received the active treatment, inactive tablets, or both active and inactive tablets at different times during the study and if they had any idea about what type of hormone(s) were contained in the tablets. Drugs The triphasic
O.C. used (Synphasic,
Syntex Laboratories,
Mississauga,
Canada)
contained
a fixed dose
Oral contraceptives
and PMS
259
of ethinyl estradiolO.035 mg from days l-21 and norethindrone, 0.5 mg during days l-7, 1 mg during days 8-16, and 0.5 mg during days 17-21. Subjects in the placebo group received 21 tablets containing lactose and cornstarch. Placebos were contained in blister packets identical to the active preparation. To determine whether ovulation occurred, 10 ml of blood were collected from each subject once during the premenstrual, menstrual, and postmenstrual phases of the third treatment cycle. Although it would clearly have been useful to have confirmed the presence of ovulation throughout the study, it was felt that the additional demands on subjects which this procedure would entail might affect compliance. Subjects were instructed to take the study medication approximately 3 hr prior to the drawing of the blood samples [7] The blood was immediately centrifuged and the plasma stored at -20°C until the completion of the study. Progesterone was measured by radioimmunoassay using a no-extraction, solid phaselz51 radioimmunoassay (Diagnostic Products Corp., Los Angeles, CA, U.S.A.). Plasma concentrations of progesterone greater than 9.6 nmols/l on days -3 to -6 of the cycle were regarded as evidence of ovulation. The primary purpose in obtaining blood samples was to detect ovulation in the group receiving placebo, since the women taking o.c.s would not be expected to ovulate. Procedure A two (treatment groups) X four (time periods) design was used. Following one cycle of baseline monitoring, subjects were randomly assigned to either the O.C. or placebo group for three cycles. At an initial interview women who met the selection criteria signed a consent form that had been approved by ethics committees of the university hospital and the community health center where the study was based. Subjects were told that the aim of the study was to investigate the effects of a ‘low-dose hormone’ on PMS. The treatment was specifically not introduced as an o.c.. They were also informed that they might receive inert tablets instead of the active medication and that their treatment could be changed during the course of the study. No instruction was given to subjects regarding the possible experience of side effects; if during the first treatment cycle women reported side effects such as breast pain or spotting, they were reassured that these were not harmful and would most likely subside with continuing use. The DRF and VAS were completed each evening and returned by mail each week. Women who failed to return their ratings were contacted by telephone. DRF ratings for the baseline cycle were scored to assess whether these confirmed subjects’ retrospective reports of moderate to severe premenstrual changes using the method employed by Endicott et al. [ 61 Change scores were calculated for each of the DRF items by contrasting the mean score of the three highest consecutive days of the last five premenstrual days with mean scores of the five postmenstrual days. The criteria used for prospective confirmation required that at least two DRF items show a two-point level of change during the premenstrual compared to the postmenstrual phase. This level of change, although somewhat arbitrary, was chosen because it is roughly equivalent to the criteria established at an NIMH workshop (1983). Two additional criteria were applied: on the above items, the mean premenstrual scores must have been 2 4 (at least moderate) and the mean postmenstrual scores must have not exceeded 3 (mild), unless there was a reported stressful event which could have accounted for the elevated score. Following the baseline cycle, subjects were given a one month supply of either O.C. or placebo and were instructed to begin taking one tablet per day on the first Sunday after menstrual bleeding began. No piils were taken from days 22-28, after which a new packet was started. This is the standard method of administration recommended for this preparation. During monthly visits to the research unit, a one month supply of daily forms and of medication were distributed. At the end of the third treatment cycle, all women completed the Post-treatment Questionnaire and were then debriefed. Statistical
analyses
Each cycle was divided into the following four phases: (1) premenstrual: the five days before the onset of menses; (2) menstrual: all days on which bleeding occurred; (3) postmenstrual: the five days after the cessation of bleeding and; (4) intermenstrual: the remainder of the cycle. A FORTRAN program developed for this study (Amsel and Amsel, unpublished computer program 1988) computed the mean score of every DRF and VAS item in each of the four cycle phases for each subject. SPSS-X three-way analyses of variance (ANOVA) (group x cycle x phase) with repeated measures (on phase and cycle) were performed on the mean phase scores for every symptom in each of the four cycles. Post-hoc tests, using Newman-Keuls method of multiple comparisons, were carried out on all interaction effects which were significant. To investigate the possibility that pretreatment variables might predict response to treatment, subjects in O.C. and placebo groups were divided into different subgroups based on whether or not they had prospectively confirmed on four DRF symptoms (depression, irritability, breast pain and edema) during the baseline cycle. A series of four-way ANOVAs (group X subgroup x cycle x phase) were carried out on the mean phase scores for each of these variables.
C. A. GRAHAM and B. B. SHERWIN
260 Student’s r-tests and Chi-squares treatment Questionnaire.
were used to analyse
the retrospective
questionnaires
and the Post-
RESULTS
Two-hundred and twelve women responded to the recruitment notices. Of these, 54 decided against taking part after an initial telephone interview and 76 were ineligible for the trial, primarily because they had physical or psychiatric disorders (N = 24), symptoms of insufficient severity (N = 13), were taking other medications (N = 12), or failed to satisfy the age restrictions (N = 10). Eighty-two women who met the selection criteria were entered into the study. Table I contains demographic and menstrual cycle data for this overall sample.
TABLE L-SUBJECT
CHARACTERISTICSfN = 82)
Percentage Marital status: married/cohabiting single divorced/separated Education: 15+ yr 12-15 yr < 12 yr Employment: full-time part-time student homemaker No. of pregnancies: 0 1 2 3 or more Previously sought treatment Yes No
of sample
40.5 53.2 6.3 48.1 41.7 10.2 55.1 12.8 21.8 10.3 52.5 20.7 12.2 14.6 for PMS:
Mean age: 29.5 yr (ft SD 5.04) Mean cycle length: 28.5 days (*
57.3 42.7
SD
2.7)
Consistent with previous treatment studies, not only was the attrition rate fairly high overall, but it was higher in the O.C. group than the placebo group (x2 = 8.7, p < 0.01) (Table II). Nine women dropped out during the baseline cycle (five o.c., four placebo), nine in the first month of treatment (eight o.c., one placebo) and five during the second treatment cycle (all o.c.). The 59 subjects (23 o.c., 36 placebo) who completed the trial were compared with the women who dropped out (N = 23) on a number of variables. Completers were employed in higher-status occupations compared to drop-outs @ < 0.007). The drop-outs had significantly higher severity scores on 4 of the 18 subscales of the retrospective PAF: lability @ < 0.04), hysteroid depressive features @ < 0.03), anxiety (p < O.Ol), and water retention @ < 0.04). No other significant pretreatment differences between completers and drop-outs were apparent. As the O.C. and
Oral contraceptives
and PMS
261
placebo groups did not differ on any of the above variables, taking the active pill increased the likelihood that a woman would drop out of the study. TABLE
II.-REASONS
GIVEN FOR DROPPING OUT (N=23)
O.C. Physical illness Decided against taking part Moving away Side effects* spotting (N = 10) breast pain (N = 6) nausea (N = 5) weight gain (N = 2) abdominal cramps (N = 2) pains in legs (N = 1) water retention (N = 1) yeast infection (N = 1) *Some
subjects
reported
(N=
18)
Placebo
3 1 1 13
(N = 5) 2 2 1
more than one side effect.
Table III lists the incidence of side effects reported by the pill subjects who completed the study (N = 23). The number of days of spotting or breakthrough bleeding was highest in the first pill cycle and occurred less frequently in the third treatment cycle (mean = 3.74 (* 4.53) vs 1.75 (* 3.1 l), respectively). TABLE III.-INCIDENCE OF SIDE EFFECTS IN THE ORAL CONTRACEPTIVE SUBJECTS (N=23)WHO COMPLETED THE STUDY
Spotting/breakthrough Nausea Abdominal cramps Breast pain/swelling Acne Dizziness
bleeding
N
%
16 7 7 4 4 4
69.6 30.4 30.4 17.4 17.4 17.4
There were no significant differences between the O.C. and placebo groups in educational background, occupational status, age, menstrual cycle length, or in the proportion of women in each group who met criteria for prospective confirmation. Eighty-three percent of the overall sample confirmed on at least two DRF items in the baseline cycle. Of the 82 subjects studied, 13 failed to meet criteria for prospective confirmation, 1 had progesterone levels which indicated an anovulatory cycle, and 23 did not complete the study. The remaining 45 women (20 o.c., 25 placebo) with prospectively confirmed premenstrual symptoms who completed the trial were included in the treatment analyses. Both O.C. and placebo groups showed a significant clinical improvement on every symptom except headache. Post-hoc analyses of cycle x phase interactions showed a significant reduction in premenstrual symptom scores between the baseline cycle and the third treatment cycle. There was also a decrease in menstrual phase scores for four variables (‘mood swings’, ‘more sleep’, ‘unhappy’ and ‘tense’) in the second
C. A. GRAHAM and B. B. SHERWIN
262
treatment cycle compared to the first month on treatment, irrespective of treatment group. However, scores in postmenstrual and intermenstrual phases did not change across the trial. Significant group x cycle x phase effects occurred for ratings of breast pain on both the DRF @ < 0.03) and VAS @J < 0.04). Post-hoc analyses yielded very similar results for this symptom on both measures. In the O.C. group, ratings of premenstrual breast pain were significantly lower during the third treatment cycle compared both to baseline @ < 0.05) and to the first treatment cycle (p < 0.01). No significant changes in premenstrual breast pain ratings occurred across the trial in the placebo group (Fig. 1). W Baseline
T
0
3rd. treatment cycle
TT
tV&S
O.C.
F&T
INI&
i
F&T
lNliF7
Placebo
FIG. 1. Mean VAS breast pain ratings (f SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, premenstrual scores lower in third treatment cycle compared to baseline @ < 0.05); other comparisons reported in text.
There was also a differential treatment response for DRF edema scores, although this finding is qualified by the fact that there was a pretreatment difference on this variable, with the O.C. group having higher premenstrual scores than the placebo group. Analyses of the significant group x cycle x phase effect @ < 0.02) revealed a significant decrease in edema premenstrually during the second @ < 0.01) and third @ < 0.01) treatment cycles for the O.C. group compared to their scores at baseline. In contrast, premenstrual edema scores of the placebo group remained stable across the trial (Fig. 2). A comparison of the four phase scores within each cycle also showed a greater reduction in premenstrual edema and breast pain in the O.C. group. Although at baseline premenstrual edema and breast pain ratings were significantly higher than postmenstrual scores for both groups @ < O.Ol), by the third treatment cycle only the placebo group showed any between-phase differences on these variables. On one variable, sexual interest, the O.C. group showed a significant deterioration during treatment. For the VAS variable ‘sexual interest’ there was a significant group X cycle X phase effect @ < 0.007); in the O.C. group, sexual interest scores were lower during both the menstrual (p < 0.01) and postmenstrual 0, < 0.01) phases of the third treatment cycle compared to baseline (Fig. 3). There were no significant
Oral
contraceptives and PMS
263
changes in sexual interest scores during treatment in the placebo group. Once again, there was one pretreatment group difference, with the O.C. group showing higher sexual interest than the placebo group during the menstrual phase @ < O.Ol), but no other significant group differences during the baseline cycle. 5
I
4
1
n Baseline
T
0
h&S
INI&
F&T
i
PA
3rd. treatment cycle
bit43
P&T
ltiR
Placebo
O.C.
FIG. 2. Mean premenstrual edema ratings (+ SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, premenstrual scores lower in third treatment cycle compared to baseline @ < 0.01); at baseline, premenstrual scores higher in O.C. group than in placebo @ < 0.05); other comparisons reported in text
n Baseline 0
3rd. treatment cycle
11 dlL _TT
l
TT
i
PRE
MENS
O.C.
fa3T
INTER
I
PRE
MENS
POST
INTER
Placebo
FIG. 3. Mean VAS sexual interest ratings (+ SEM) of the oral contraceptive (o.c.) (N = 20) and placebo (N = 25) groups. In O.C. group, menstrual and postmenstrual scores lower in third treatment cycle compared to baseline @ < 0.01); other comparisons reported in text.
When women were divided into subgroups based on whether they had prospectively confirmed on specific DRF items (depression, irritability, breast pain, edema) only one symptom-depression-showed any relationship to treatment response. For the sake of clarity, women who had prospectively confirmed on depression at
264
C. A. GRAHAM and B. B. SHERWIN
baseline were designated ‘depressed’ and those who failed to meet confirmation criteria were called ‘non-depressed’. Thus, there were four subgroups: ‘depressed’ O.C. (N= 12), ‘non-depressed’ O.C. (N= 8), ‘depressed’ placebo (N= lo), and ‘non-depressed’ placebo (N = 15). The ‘depressed’ O.C. group reported greater improvement in three behavioral symptoms; specifically, these women were less impaired at work @ < 0.02), required less sleep @ < 0.05), and had more energy (p < 0.04) premenstrually whilst taking O.C.S. In contrast, women with premenstrual depression who received placbeo, and those without a tendency to cyclical depression in both treatment groups, either showed no change in these symptoms or less consistent improvement. On the Post-treatment Questionnaire subjects in the O.C. group were more likely to identify which treatment they had received compared to the placebo subjects @ < 0.005). Fourteen (38.9%) placebo subjects believed that they had received both active and inactive tablets and seven (19.4%) incorrectly guessed that they had taken active tablets throughout the study. In comparison, four women (17.4%) in the O.C. group incorrectly guessed that they had received both active and inactive tablets and one (4.3 %) thought that she had taken a placebo throughout the study. Only 24 % of the overall sample correctly guessed that the active tablets contained a synthetic estrogen and progesterone; 45 % stated that they did not know what type of hormone(s) were contained in the active tablets and 3 1% gave an incorrect response.
DISCUSSION
To our knowledge, this is the first double-blind, placebo-controlled trial of o.c.s in women with prospectively-confirmed PMS. Several other features of the experimental design increased the likelihood that any observed differences in the effects of the O.C. and placebo could be attributed to direct hormonal effects on the behavior. Prior studies on the relationship between o.c.s and PMS have looked at premenstrual symptoms in women already taking o.c.s and have not assessed symptom change before and after starting O.C.S. Moreover, because our subjects were not aware that the treatment was an o.c., factors such as attitudes toward the pill, personality characteristics which affect choice of contraception, etc. were controlled. These data do not support the commonly-held theory that premenstrual mood symptoms are a consequence of hormonal fluctuations of the ovarian cycle. However, the fact that breast pain and edema improved to a greater extent on the active pill than placebo suggests that these two symptoms may be related in some way to ovulation and luteal function. In agreement with this, a recent comparison of women using and not using different types of O.C.S. found breast tenderness was the only symptom to show a clear difference between groups [ 1 ] . A specific aim of this study was to relate response to O.C. administration with different types of premenstrual changes in the hope that guidelines for treatment decisions might emerge. Our findings suggest that a triphasic O.C. may provide relief for premenstrual breast pain and edema. However, most women who seek treatment for PMS do so because of an inability to cope with the mood-related changes rather than the physical symptoms [ 81 and cyclical mood changes improved equally with O.C. and placebo treatment. These findings are consistent with those of Cullberg [9] who reported that
Oral contraceptives
and PMS
265
premenstrual irritability and depression were not significantly relieved by o.c.s compared to placebo. In agreement with previous reports [lo], premenstrual mood symptoms seem to be more responsive to placebo than physical symptoms. Previous studies have suggested that the response afforded by placebo is transient and lasts only two or three cycles [ 111. In the present study there was no suggestion of any abatement of the response to placebo by the third treatment cycle, although it is possible that with a longer treatment period, this may have occurred. In a recent cross-over study of estradiol patches in the treatment of PMS [ 121, there was a response to both placebo and active treatment during the first three months of use. However, after three months, patients who switched from placebo to active treatment maintained their initial gains while the scores of those who changed from active treatment to placebo generally deteriorated. In addition to the positive effects of the active treatment on breast pain and edema, there are also possible negative effects of the O.C. which must be considered. It has previously been suggested that women who are prone to PMS are more likely to experience side effects using o.c.s [ 131 . As there is very little information available on what proportion of women taking o.c.s experience side effects, it is difficult to determine whether the present sample had more difficulties tolerating 0.c.s than women without premenstrual complaints. Side effects (particularly spotting) were also fairly commonly reported by completers, especially in the first cycle on O.C.S. A clinical impression was that the women who dropped out, often after a very short time on the tablets, may have been more ambivalent about taking hormones. On the other hand, it may be the case that the women who withdrew from the study did constitute a subgroup who were more likely to experience side effects. The changes in sexual interest induced by the pill were unexpected and of particular interest, given the dearth of evidence on the effects of the triphasic 0.c.s on sexuality. After starting on the triphasic o.c., women showed less sexual interest during the menstrual and postmenstrual phases (corresponding to the pill-free week and the first few days of the next pill cycle). These findings are in contrast with two other studies [ 14, 151 showing that O.C. users report highest sexual interest in the postmenstrual week. A caveat with regard to these data is related to the higher baseline levels of sexual interest in the O.C. group. Nevertheless, the findings provide interesting information with regard to the relationship between sexual interest and mood. These negative effects of the O.C. on sexual interest could not be attributed to a negative influence on mood and suggest a direct hormonal effect of the O.C. on sexuality, although the mechanism by which this may have occurred is unclear. The finding that sexual interest and mood are dissociable is supported by results of a previous placebo-controlled study in which women who had undergone sterilization experienced decreased sexual interest on the O.C. but no alteration in depression scores [ 161. Given the continued uncertainty in the literature on whether o.c.s are associated with depressive mood change [ 21 , it is interesting that none of the women receiving 0.c.s reported any negative effects on mood. In fact, a subgroup of women with a tendency to premenstrual depression responded better to 0.c.s than women without such cyclical depression. Because the number of subjects in the subgroup analyses were small, these findings should be regarded as tentative. However, given the fact
C. A. GRAHAM and 8. B. SHERWIN
266
that there is currently no basis on which to predict a woman’s response findings, if replicated, would have important clinical implications. .4c/tnowledgemenrs-Support
Council
of Canada
for this research
(No. MA-8707)
was provided
by grants
from
to o.c.s,
the Medical
these
Research
and from Fonds pour la Formation
de Chercheurs et 1’Aide a la Graham was supported by a Medical Canada provided the O.C. and placebo
Recherche (No. EQ3632) awarded to B. B. Sherwin. C. A. Research Council of Canada Studentship. Syntex Laboratories, tablets. The authors would like to thank Natalie Rubin for her help with data collection
and analyses.
REFERENCES
1. WALKER
A, BANCROFT J. Relationship between premenstrual symptoms and oral contraceptive use: A controlled study. Psychosom Med 1990; 52: 86-96. 2. BANCROFT J, SAR~ORIUS N. The effects of oral contraceptives on well-being and sexuality. Oxford
Rev Reprod Biol 1990; 12: 57-92. 3. FREEMAN EW, SONDHEIMER S, WE~NBAUM PJ, RICKELS K. Evaluating 4.
5. 6. 7.
8. 9.
premenstrual symptoms in medical practice. Obsfer Gynecol 1985; 65: 500-505. HEALTH AND WELFARE CANADA. Oral Contraceptives. Government Report, 1985. HALBREICH U, ENDICOTT J, SCHACHT S, NEE J. The diversity of premenstrual changes as reflected in the Premenstrual Assessment Form. Acra Psychiat Stand 1982; 65: 46-65. ENDICOTT J, NEE J, COHEN J, HALBREICH U. Premenstrual changes: Patterns and correlates of daily ratings. J Affect Disord 1986; 10: 27-35. SAPERSTEIN S, EDCREN RA, ELLIS DJ, LEE GJL, KUSHINSKY S, OLMXED A, MROSZCZAK E. Bioequivalence of norethindrone and ethinyl estradiol for two different weight tablets with the same hormonal content. Conrracepption 1986; 33: 547-557. CORNEY RH, STANTON R. A survey of 658 women who report symptoms of premenstrual syndrome. J Psychosom Res 1991; 35: 71-82. CULLBERG J. Mood changes and menstrual symptoms with different gestagen/estrogen combinations.
Acra Psychiar Stand (Suppl) 1972; 236: 1-86. IO. METCALF MG, HUDSON SM. The premens!rual syndrome: Selection of women for treatment trials. J P.~pchosom Res 1985; 29: 631-638. Il. MAGOS AL, BRINCAT M, STUDD JWW. Treatment of the premenstrual syndrome by subcutaneous oestradiol implants and cyclical oral norethisterone: Placebo controlled study. Br Med J 1986; 292: 1620-1633. 12. WATSON NR, Sruoo JWW, SAVVAS M, GARNET.~ T, BABER RJ. Treatment of severe premenstrual syndrome with oestradiol patches and cyclical oral norethisterone. Lancer 1989; iii: 730-?32. 13. DENNERSTEIN L, SPENCER-GARDNER C, BROWN JB, SMITH MA, BURROWS GD. Premenstrual tension: Hormonal profiles. J Psychosom Obsret Gynecol 1984; 3: 37-51. 14. ADAMS DB, GOI.D AR, BURT AD. Rise in female-Initiated sexual activity and its suppression by
oral contraceptives.
N Errgl J Med 1978; 299:
15. WARNER P, BANCROF~ J. Mood,
sexuality,
1145-I
150.
oral contraceptives,
and the menstrual cycle. J Psyrhosom
Rrs 1988; 32: 417-427. 16. LEETON J, MCMASTER R, WORSLEY A. The effects on sexual response and mood after of women taking long-term oral contraception: Results of a double-blind cross-over N Z Obstet Gynaecol 1978; 18: 194-197.
sterilization study. Aust
