Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Mood Disorders Linked to the Reproductive Cycle in Women

Barbara L. Parry and Patricia Haynes

In both medicine and psychiatry there are gender-differentiated predispositions to certain illnesses.  For example, men are at greater risk to develop cardiovascular disorders, alcoholism, and sociopathy, whereas women have a greater lifetime risk for thyroid disease, eating disorders, late life schizophrenia and depression (91).  Sex differences in the preva­lence of depression begin to appear after puberty and are maintained throughout the reproductive years (126).  More recent evidence indicates that major depressive disorders are increasing with time, that the age of onset is becoming earlier, and that women continue to show an increased incidence of the disorder.  Furthermore, female offspring of depressed patients have earlier onsets of depressive disorders (125).


Women, as compared with men, have a greater lifetime risk for depression.  They predominate with respect to unipolar depression (127), the depressive subtype of bipolar illness and cyclical forms of affective illness such as rapid cycling manic-depressive illness and seasonal affective disorder. In addition, events associated with the reproductive cycle are capable of provoking affective changes in predisposed individuals.  Examples include depression associated with oral contraceptives (79), the luteal phase of the menstrual cycle (22) the postpartum period and menopause.  The fluctuation of gonadal steroids during specific phases of the reproductive cycle may bear some relationship to the particular vulnerability of women for affective changes.  The reproductive hormones could exert their effects on mood directly or indirectly by their effect on neurotransmitter, neuroendocrine, or circadian systems, all of which have been implicated in the pathogenesis of affective illness. In addition, the rate of hormonal change may play an important contributory role in mood disturbances (39).

Several clinical models can be used to examine the role of reproductive hormones on affective illness in women.  As most patients with rapid cycling manic-depressive illness and seasonal affective disorder (SAD) are women, reviewing the factors that predispose women to the development of mood cyclicity may shed light on how reproductive hormones may influence the course of affective illness. Also, the cyclicity inherent in rapid cycling mood disorders is similar to the presentation of other forms of affective illness, such as those associated with the reproductive cycle. An understanding of disorders associated with changes in reproductive hormones, such as premenstrual dysphoric disorder and postpartum depression, can be furthered by the study of these forms of cyclic mood disturbances, in which women predominate.

Rapid Cycling Affective Illness

Rapid cycling manic-depressive illness, defined as four or more affective episodes a year, predominates in women. Kraepelin (55), in describing patients with regular, daily fluctuations of periodic excite­ment noted "in contrast to other forms, in which there was a preponderance of men, two thirds of these patients were women in whom the periodicity of sexual life obviously favors this kind of development." A recent meta-analysis (113) of 10 studies and 2,057 bipolar patients found that of the rapid cycling cases, 72% were women and 28% were men.  Rapid cycling was also more common in bipolar women (30%) than men (17%).  The rates of rapid-cycling varied from 17-67% among women; however, higher rates have also been reported. For instance, Wehr et al (122) indicated that 92 percent of patients with rapid cycling and 64 percent with nonrapid cycling affective disorder were women.

Women's Risk Factors for Rapid-cycling Affective Illness

In addition to being a woman, two other factors appear to be associated with the rapid cycling form of bipolar illness: 1) treatment with tricyclic and other antidepressants and 2) hypothyroidism.  Women, compared with men, show an increased incidence of both drug-induced rapid cycling and hypothyroidism (122).

Drug-Induced Rapid Cycling

Of patients whose rapid cycles of mania and depression have been induced by tricyclics, women predominate.  Wehr and Goodwin (120) found that five patients, in whom rapid cycles developed, represented all but one of the bipolar women who had been maintained on antidepressant drugs.  Rapid cycles did not develop in either of two bipolar male patients who were maintained on tricyclics.  Wehr and Goodwin speculated that the female reproductive neuroendocrine axis, a generator of physiologic rapid cycles, may have been instrumental in the expression of drug-induced cycling.  Reproductive hormonal disturbances and treatments may have been predisposing factors in their patients' illnesses; all but one (four of five) of the women in their sample had irregular menses, amenorrhea, history of estrogen or pro­gesterone treatment, or onset of the illness in the postpartum period.

In Kukopulos' (56) longitudinal study of the  patients who developed rapid cycles after tricyclics, 70 percent were women.  According to Kukopulos, "female sex, middle age, and menopause, along with anti-depressant drugs, contributed to the establishment of rapid cyclicity." Of the patients in his longitudinal study whose course of illness changed to rapid cycling, 87 percent were women, and in a third (25 of 77) of these women, the change in course coincided with menopause.  Kukopulos also noted that the patients with depression and hypomania (bipolar II) were those most prone to rapid cycling.

Wehr and Goodwin (120) concluded that women have a higher risk of drug-induced mania and drug-induced rapid cycling than do men. In contrast, Kupfer et al (57) report that women with recurrent depression are not more likely to switch into hypomania than men, and Coryell et al (18) de-emphasize the effects of drug-induced rapid cycling.

Hypothyroidism and Rapid Cycling

In addition to female gender and antidepressant drugs, thyroid impairment may be associated with rapid cycling (19,122), and women are prone to thyroid disease.  Studies that document sex differences indicate that almost all patients on lithium who develop hypothyroidism are female.  In studies by Transbol et al (114), and Cho et al (15), 90 to 100 percent of bipolar patients with lithium-induced hypothyroidism were women.

As shown by Cowdry et al (19), abnormalities of the thyroid axis, some of which may become apparent only during treatment with LiCO3, are associated with rapid cycling. Overt hypothyroidism was found in 12 (51%) of 24 rapid cycling patients and in none of the nonrapid cycling patients.  Elevated TSH levels were present (and higher) in 92% of the rapid cycling group as compared with 32% of the nonrapid cycling group.  In this sample, women represented 83% of the rapid cycling group and 53% of the nonrapid cyclers.

In an mood disorder clinic population, Cho et al (15) found postlithium thyroid medication use was significantly higher among rapid cycling women (32.2 %) than nonrapid cycling women (2.1 percent).  Furthermore, thyroid dysfunction was primarily limited to women.  Of those patients taking thyroid medication in addition to lithium, 92% were women.  Five out of seven women with hypothyroidism had rapid cycling.

In a study by Wehr et al (122), 47% of women with rapid-cycling and 39% of women with nonrapid cycling bipolar illness developed thyroid disease.  Of those patients with thyroid disease, the thyroid disease emerged after the onset of affective illness in 90% of the patients and in the majority of cases it emerged during lithium treatment.

Transbol et al (114) evaluated the prevalence of hypothyroidism in lithium-treated manic-depressive outpatients and found elevated TSH levels in 25% of the patients.  Of these patients, 95% were women over 40 years of age.  None of the men had elevated TSH levels.  Rapid cycling and nonrapid cycling subgroups were not reported separately.

In a medication-free sample, abnormalities of the thyroid axis may not be as apparent. Post et al (81) found no significant differences in thyroid hormone levels for rapid cyclers and non-rapid cyclers across gender.

Similar to rapid cycling, mixed states for bipolar disorder predominate in women. As compared with manic or depressive states, a mixed state indicates a simultaneous occurrence of manic and depressive symptoms. Chang et al (14) found elevated TSH levels and low thyroxine levels in patients with mixed mania compared with pure mania for both men and women; there was a trend toward a greater proportion of women in the mixed group. There were no significant differences in previous lithium exposure.

Reproductive status may affect the appearance of thyroid disease.  There is a common appearance of goiter during puberty, pregnancy, and menopause.  Women are particularly prone to develop hypothy­roidism during the postpartum period (4).  This type of hypothyroidism may represent an autoimmune phenomenon, as the extent of postpartum hypothyroidism correlates with traces of microsomal antibodies early in pregnancy (48, 59).  Women with isolated gonadotropin deficiency have blunted basal TSH, and TSH responses to TRH (106).  Administration of estrogen restores the TSH response to that of normal controls, and cessation of estrogen treatment reduces the amount of releasable TSH.  In hypogonadal men, TSH response to TRH is similar to normal men but increases with estrogen treatment.  Oral contraceptives increase the TSH response to TRH (86).  Thus, estrogens seem to be required to maintain, and may enhance, the normal TSH response to TRH in the female.

In summary, cyclic affective disorder in the form of rapid cycling manic-depressive illness is more prevalent in women.  Treatment with antidepressant drugs often precipates rapid mood cycles, particularly in women with bipolar II illness.  Thyroid impairment, more prevalent in women, also is associated with the rapid cycling form of the illness. The predominance of these disorders in women suggests that abnormalities in reproductive hormones, and other neuroendocrine systems, may exacerbate mood cyclicity. There is very little information available on the course of rapid cycling bipolar illness in men. 

Clinical Psychopharmacological Treatment

Thyroid Treatment

Thyroid hormone has been used to treat both cyclic and noncyclic forms of affective disorders.  Gjessing first used Hypermetabolic thyroid treatment in the 1930s to treat periodic catatonia and rapid cycling affective disorder.  Stancer and Persad (107) reported treatment of intractable rapid cycling manic-depressive illness with levothyroxine.  Five of the seven women, who developed rapid cycling bipolar disorder with an onset during the postpartum or involutional period, responded to hypermetabolic doses of thyroid hormone.  The effect of thyroid hormone had differential effects in men and women: The treatment was unsuccessful in two men and one adolescent girl, which further indicate that female reproductive hormones may be important in the development of mood cyclicity.

Prange et al (83) reported triiodothyronine enhanced the antidepressant effect of imipramine in women but not in men.  Men responded to initial doses of imipramine in a shorter period of time than women.  Women treated with imipramine and thyroid supplement responded as rapidly as men treated with imipramine alone.  Later, Goodwin et al (34) demonstrated that among tricyclic nonresponders, men benefited from the addition of T3 as often as women.

The mechanisms for thyroid-enhancement of responses to antidepressants may be as Whybrow and Prange (131) suggest, relating to the capacity of thyroid hormone to alter the ratio of alpha to beta adrenergic receptors and their sensitivity to noradrenergic neurotransmitters.  Depressed women appear to be uniquely responsive to thyroid hormone.  Women also are uniquely vulnerable to thyroid impairment. Therefore, changes in hormonal axes are likely to play an important role in altering the course of mood disorders.


Women also are sensitive to other hormonal treatments.  Estrogen also has been used as a treatment in refractory depression. Klaiber et al (54) used 5 to 25 mg of oral conjugated estrogen in cyclic doses to treat pre- and postmenopausal women.  With estrogen, as compared with placebo, there was a significant drop in Hamilton scores, which correlated with reduction in previously elevated monoamine oxidase (MAO) levels.  With inconsistent results, Prange (83) gave 25 mg and 50 mg estradiol to depressed patients already treated with imipramine.  The higher estrogen dose was toxic.  The lower dose was associated with reduction of Hamilton scores and improved sleep. A meta-analysis of 26 studies on the effect of hormone replacement therapy on depressed mood in menopausal women found that estrogen significantly reduced depression, progesterone blunted the antidepressant effect of estrogen, and that androgens increased the antidepressant effect of estrogen (143).

Estrogen may also induce rapid cycling or at least predispose to tricyclic-induced rapid cycling as reported by Oppenheim (70).  Interestingly, Parry and colleagues (unpublished observations) observed one male rapid cycling patient who had low testosterone secondary to mumps orchitis.  Progesterone, on the other hand, may suppress rapid cycles of mood (45).

The mechanism by which estrogen exerts its possible antidepressant effect is unknown, but work by Kendall et al (52) showed estrogen is needed for reduction of serotonin receptor binding during imipramine treatment.  Ovariectomy blocked the effect of imipramine on serotonin receptors, and estrogen treatment reinstituted it. In postmenopausal women, estrogen can augment reduced serotonergic activity and some norepinephrine-related processes (38). In depressed, perimenopausal women, there is a greater magnitude of treatment effect when estrogen is used as an adjunct to a selective serotonin reuptake inhibitor (SSRI) (fluoxetine) than using fluoxetine alone or estrogen alone (111).

Estrogen also appears to play an important role in the maintenance of internal circadian rhythm synchronization. Wise et al. (134) suggested that menopause may be associated with an age-related deterioration or fragmentation of multiple circadian rhythms. In ovariectomized rats, estrogen implants reduce the degree of rhythm desynchrony by enhancing the coupling between component oscillators (112).  Evidence linking disrupted circadian rhythms to affective disorders is extensive and will be discussed in more detail below.

Besides improvement in mood, estrogen has been posited to improve memory and cognitive functioning in postmenopausal women and in women with Alzheimer’s Disease (44). However, these findings are equivocal and have been a source of controversy. Conflicting results may be due to methodological issues, such as: failure to define criteria for menopausal status, failure to assess for mood or physical disturbance as a confounding factor, wide variation in the type of neuropsychological instruments used, and considerable variability in the type, dosage, and route of hormone replacement therapy.  For these reasons, the authors (44) concluded that until randomized, controlled trials are conducted, there is inadequate evidence to support the conclusion that estrogen replacement therapy improves cognitive functioning.

Preliminary findings in imaging may provide additional information on the effects of estrogen on the brain.  Resnick et al. (88) studied fifteen women on estrogen replacement therapy (ERT) compared with seventeen nonusers on positron emission tomography (PET)-regional cerebral blood flow (rCBF) activation patterns.  Activation patterns differed significantly across groups on figural and verbal memory tasks. More specifically, task-associated changes in rCBF pertaining to memory components occurred in the parahippocampal region and right frontal regions. ERT users also performed better on neuropsychological figural and verbal memory tasks. These findings lend further support to hypotheses by Sherwin and colleagues (102) that ERT may preserve some aspects of verbal memory.

In a PET activation study by Berman et al. (9), they also demonstrated that reproductive hormones are likely to have a modulating role on executive functioning.  The authors found virtually no activation of the prefrontal cortex while performing an executive functioning task in the absence of gonadal steroid hormones (ovarian suppression induced by a gonadotropin-releasing hormone agonist). With estrogen or progesterone, the rCBF pattern normalized.

There also is considerable evidence from the animal literature suggesting that estrogen significantly affects physiologic systems involved in learning and memory. For example, estrogen induces the formation of excitatory synapses in the CA1 region of the hippocampus and has activational effects on cholinergic function in the central nervous system (62). For example, estrogen protects against an age-related decline of cholinergic function in the basal forebrain projections to the frontal cortex (103).

Seasonal Affective Disorder

Like rapid cycling affective disorder, seasonal affective disorder (SAD) is a cyclic mood disorder that occurs predominately (80%) in women.  Thus, it also may serve as a model to understand the contribution of reproductive hormones to affective illness in women.

Patients with SAD who have recurrent winter depressions, have symptoms of major affective disorder with characteristic "atypical" features (such as hyperphagia, hypersomnia, and lethargy), which begin to develop each year in association with shortening of the daylength.  A majority of these patients respond to high intensity (2500 lux) light treatments, which artificially extend the daily photoperiod. More recent work suggests that using higher intensity (10,000 lux) light treatments for a shorter time period (30-45 minutes) (58; see also Biological Rhythms in Mood Disorders) also may be safe and effective.  Initially, bright light was thought to act in seasonal depression by suppressing melatonin, a hormone that is centrally involved in seasonal reproductive cycles in animals.  This hypothesis since has been brought into question and a multitude of other hypotheses proposed (92,121). Increasing evidence has indicated that the serotonin system may be disturbed in SAD (101, 50) and is likely to play a role in the mechanism of action of light therapy (68).

A large proportion (70%) of women with SAD also has mood changes in association with the menstrual cycle. In a recent study, 38% of women diagnosed with late luteal phase dysphoric disorder also met criteria for SAD (60). Some women with seasonal affective disorder report improvement in their premenstrual symptoms with light therapy.  Parry and colleagues (78) identified a woman with a family history of bipolar illness who developed severe premenstrual depression with suicidal ideation only during the fall and winter and was relieved of premenstrual symptoms during the spring and summer.  Light was an effective treatment for this patient with seasonal PMS and that its therapeutic effect could be blocked by the simultaneous administration of melatonin.  Light also increased this patient's TSH.  Propranolol and atenolol, beta blockers that inhibit the synthesis of melatonin, had a therapeutic effect similar to light.  Light therapy also may benefit women with non-seasonal premenstrual syndrome (74, 77), which will be discussed in more detail below.

In patients with seasonal affective disorder who have summer depression and winter hypomania, Wehr et al (123) reported that eight of 12 patients were women and suggested that temperature may influence these patients' clinical state.

Women with major depression and bipolar illness may also experience cyclic, seasonal recurrences. Suhail and Cochrane (110) measured seasonal variations in psychiatric admissions over a period of one year for a psychiatric hospital in Birmingham. They found that admissions for depression and mania showed a statistically significant seasonal pattern only among women (depression highest in the winter and mania in the summer), which suggests that women may have a specific vulnerability to variations in light and temperature. Interestingly, none of the subjects in this study (402 women) were diagnosed with SAD.

It appears that certain women with genetic vulnerability for mood disorders may be at risk for developing other cyclic affective disorders such as seasonal premenstrual disorders.

Applications of Rapid Cycling and Seasonal Affective Disorder to Postpartum and Premenstrual Affective Illness

Rapid cycling bipolar illness and SAD are examples of cyclic forms of mood disorders that predominate in women.  The interaction of reproductive hormones, particularly with thyroid hormones in rapid cycling affective disorder and with melatonin in seasonal affective disorder, may provide clues to the pathogenesis of reproductive-related depressions.  The increased incidence of hypothyroidism occurring postpartum (4) or in winter may in part account for rapid cycling mood disorders and depression that may occur at these times,  respectively. Another cyclic mood disorder that can be viewed as a form of rapid cycling is recurrent premenstrual depression in which thyroid and melatonin disturbances also have been implicated (72, 78).  Melatonin may play a role not only in the pathogenesis of seasonal premenstrual syndrome (78), but also in nonseasonal forms of this disorder (72, 73).  Recent studies demonstrate that patients with premenstrual depression have circadian disturbances of melatonin secretion compared with normal controls and may respond to light treatment (72, 74, 77, 73).  Similar studies of melatonin in postpartum mood disturbances are currently underway. 


Pregnancy and Postpartum Mood Disorders

The relative risk for developing a major psychiatric illness or psychosis requiring hospitalization is highest during the postpartum period (53), and may recur during pregnancy. First episodes of manic-depressive illness in women often have their onset in the postpartum period (31, 87).  Reich and Winokur (87) observe that not only is there a special risk for female manic-depressive patients to develop mania or depression in the postpartum period, but having a postpartum affective episode appears to predispose to subsequent postpartum affective episodes (an increased risk of 50%). Furthermore, a woman's initial risk for a postpartum psychosis is 1 in 500 (34) to 1 in 1000 (117).  Once a woman has had a postpartum psychosis, her risk for psychosis following a subsequent pregnancy is one in three (53).  Thus, in postpartum depression and psychosis, a previous episode sensitizes a woman to the development of future episodes with subsequent pregnancies.

Clinical Psychopharmacologic Treatment

Before discussing specific treatment modalities, it needs to be emphasized that at this early stage of recognition of postpartum psychiatric syndromes by American Psychiatry, a rational treatment plan for postpartum depression and psychosis cannot be developed from double-blind, placebo-controlled crossover trials of pharmacologic or psychotherapeutic interventions.  Because these illnesses can be so devastating to the individual and her family, treatment approaches have utilized whatever interventions have been immediately useful and available.  In the literature, the majority of the more scientifically rigorous treatment studies are confined to studies that utilize patients with maternity blues, i.e., those individuals without severe disorders, as their study subjects.  Therefore, by necessity, suggested treatment approaches discussed in this chapter reflect clinical experience more than information derived from research investigations.

There are several important principles of treating postpartum depression and psychoses.  The first principle of treatment is that organic illnesses must be ruled out.  An initial presentation of postpartum psychiatric illness may be due to an underlying Sheehan's syndrome, thyrotoxicosis (if presenting as an acute psychoses in the first month after delivery), or as hypothyroidism (if presenting as major depression in the 4-5th month postpartum).  All too often these medical emergencies are overlooked, with disastrous consequences.  Thus, one of the first crucial steps in the initial evaluation and treatment of postpartum disorders, as in other medical and psychiatric disorders, is a thorough history, physical and laboratory examination.

The other important principle guiding treatment is that the earlier the symptoms are recognized and treated, the better.  For example, postpartum psychosis may initially present with symptoms of depersonalization: the patient may feel distant from her child and from the situation at hand.  She may feel as though she is just being an onlooker (as portrayed in the film "Rosemary's Baby").  This phenomenon may be interpreted as a "failure to bond" but more likely represents the initial presentation of an emerging psychosis.  Patients then may develop strange and bizarre sensations described as though their, or that of their child's head, is separate from their body.  If treatment is instituted with small doses of an antipsychotic medication, these symptoms may resolve within a few days to a week.  However, if not recognized and treated in its initial stages these symptoms may rapidly progress to paranoid delusions and a frank agitated psychosis which may become more severe, more refractory to treatment, and more likely to recur over the next 6 months to a year. Without aggressive management and early detection, the symptoms may extend into the second and third year postpartum.

It also must be stressed that because of the changing nature of postpartum psychiatric illness, different treatments at different stages of the illness are indicated.  For example, an early presentation of psychosis would best be treated with neuroleptic medication.  However, this psychosis may resolve and the patient may develop symptoms of major depression later in the course of the illness that may require antidepressant medication.  Furthermore, the initial presentation of the depression may appear in an agitated form with many anxious features and insomnia.  Thus, treatment with an antidepressant that may have some anticholinergic features (e.g., imipramine, nortriptyline) that induce more sedative side effects or affect specific receptor subtypes that do not tend to elicit anxious, agitated side effects would be indicated.  Later, the patient may present with symptoms of a retarded, anergic depression with possible obsessive compulsive features; a compound that may have activating qualities (e.g., desimpramine, fluoxetine) may be indicated.  As Post (82) suggests in the treatment of affective illness and psychosis, different treatment modalities may have differential effects and efficacies depending upon when in the course of the illness these treatments are administered.

• See Table 1

Specific issues related to the treatment of first postpartum psychosis and then postpartum depression are discussed below.

Postpartum Psychoses

One of the most important aspects of management of the postpartum psychoses are that the earlier they are recognized and treated, the more likely they are to respond to treatment and be associated with a more positive outcome and prognosis.  Since most postpartum psychoses have an acute onset within the first two weeks postpartum (generally not until after the 3rd postpartum day), and 80% of them occur within one month postpartum, clinicians should be on the alert for early signs of depersonalization, delusional thinking, mania or bizarre behavior.  Women with a previous history of postpartum psychiatric illness or affective illness are at particularly high risk.  The clinician should have a low threshold for hospitalization in any patient with symptoms of an impending postpartum psychosis.  Early hospitalization can prevent infanticide or suicide that may occur when mothers at risk are left at home alone to care for their infants, a frequent occurrence in modern culture.  Often small (2-5 mg) doses of neuroleptics such as haloperidol (or if it contributes to extra-pyramidal symptoms, perphenazine or loxapine) may decrease the symptomatology and prevent the development of a more severe psychosis. Further research is needed to investigate the efficacy of newer atypical antipsychotics for postpartum psychosis; however, clinical observations and anecdotal reports (B. Parry, personal communication) suggest that atypical antipsychotics, though effective for some other types of schizophreniform psychoses, do not appear to be as effective as the typical neuroleptics for the affective psychoses manifested in postpartum psychiatric illness, perhaps due to relative selectivity of dopamine receptor subtypes.

Neuroleptic medication also is efficacious in treating the symptoms of a postpartum psychotic depression (see below), without incurring the risks of tricyclic antidepressants.  In the potentially hypothyroid postpartum state, antidepressants may induce rapid cycling and are not recommended in breast feeding women. Risk factors for antidepressant-induced rapid cycling include being female, being in the postpartum state, and being hypothyroid. 

If the symptoms of an emerging postpartum psychosis are recognized and treated early, they may resolve within a week.  Cases in which the symptoms are not recognized and treated in their initial stages may become much more refractory to treatment and take much longer to resolve.  In general however, the postpartum psychoses have a good prognosis, resolve in 2-3 weeks, and are amenable to treatment.  Unfortunately postpartum psychosis is the condition under which women are most likely to commit infanticide.  Of those women with postpartum psychoses, 4% may commit infanticide (31).  This consequence generally does not occur unless the patient is psychotic.  The tragedy of this occurrence is made all the more poignant by the recognition that this disorder is otherwise so amenable to treatment and thereby preventable.

Although dosages of neuroleptics can be reduced after the initial episode of psychosis is resolved, this process should be done gradually and cautiously.  Women remain at risk for recurrences, particularly in those women with a previous history of psychiatric illness, for at least 6 months, and often up to 12 months postpartum.  Data from a large scale epidemiological study in Edinburgh (53) suggest that there is an increased risk for psychiatric admissions for up to two years postpartum. Videbech and Gouliaev (117) found that 60% of women with postpartum psychosis had recurrences after 7-14 years; most were of the nonpuerperal type (42%).  Although it is not necessary to leave a patient on neuroleptic medication for this length of time, it is wise for the clinician to be on the alert for early signs of recurrence and to bear in mind that a patient who initially presents with symptoms of a postpartum psychosis within the first few weeks after delivery may develop symptoms of a postpartum depression later in the course of her illness (i.e., four-five months postpartum).  For an early onset of psychoses, patients should probably remain on neuroleptics for at least six weeks postpartum.  The concentrations of different antipsychotic drugs vary greatly in the breast milk. Overall, given the amount of data available, the typical neuroleptics are relatively safe (compared with the risks of untreated psychiatric illness) during pregnancy and lactation.  In contrast, risperidone is not recommended in breast-feeding women (115).

As in other psychiatric illnesses, psychopharmacologic intervention is most effective when combined with psychotherapeutic interventions.  Particularly with regard to postpartum psychoses, pharmacologic intervention is urgent to prevent the mother from becoming increasingly psychotic and potentially committing infanticide.  At this point the patient cannot be cognitively and emotionally available to participate in a psychotherapeutic interaction until the medications help to reduce the hallucinations, delusions, and agitated behavior.  However, as most clinicians and even psychotic patients appreciate, medications are most likely to be received and taken willingly when some sense of rapport, trust, and support is perceived by the patient and her family as coming from the physician.

Postpartum Depression

In contrast to postpartum psychosis in which there is an acute onset occurring early in the postpartum period, postpartum depression generally has a more insidious onset that occurs later, i.e., 4-5 months postpartum.  Its severity may range from mild to moderate dysthymia and anxiety disorders to major melancholia.  As with the postpartum psychoses, organic abnormalities need to be ruled out, particularly hypothyroidism, which occurs in 10% of women postpartum and has a peak incidence at 4-5 months postpartum (48).  Transient hyperthyroidism may actually appear earlier in the postpartum course.  Indications for the use of antidepressant medication are similar to those for other affective illnesses, and include the presence of neurovegetative signs.  It is necessary to bear in mind that the course of affective illness over time, and that of postpartum illness being no exception, is that untreated episodes tend to become more severe, more frequent and often more refractory to treatment.  These depressive episodes should be treated aggressively with both pharmacologic and psychotherapeutic strategies early in the course to prevent untoward biological and psychological consequences, which can include a disruption of maternal bonding with the newborn child. 

Since many of the depressions may appear with obsessive-compulsive features, implicating serotonergic mechanisms, recent clinical experience suggests the efficacy of the serotonergic antidepressants, such as fluoxetine, sertraline and paroxetine.  However, side effects particularly of agitation from fluoxetine (24) or the potential induction of mania in bipolar patients need to be monitored closely. Although some evidence has shown that fluoxetine is effective for postpartum depression (3), it may not be the best first line of treatment in the anxious depressions often seen early in the postpartum state.

If agitated depressive symptoms occur early in the postpartum state, small doses of neuroleptics can be beneficial.  Anxiolytics are best avoided because of their risk for the development of physiologic dependence, withdrawal, paradoxical exacerbation of agitation, and their inadvisability of use in breast feeding women.  With the use of antidepressant medication, it is best to collaboratively consider the risk-benefit analysis of breast-feeding, as some studies indicate that small amounts may be excreted into the breast milk.  If the mother is reluctant to give up breast feeding, and her depressive symptoms are severe enough to warrant pharmacologic treatment, research suggests that antidepressants such as amitriptyline, nortriptyline, desipramine, clomipramine, dothiepin, and sertraline are at low risk for adverse effects—especially with infants older than 10 weeks. In nursing infants, higher serum levels and adverse effects were reported with doxepin and fluoxetine (136).  The clinician is advised if administering antidepressant drugs to a postpartum patient, to rule out hypothyroidism, to follow closely the course and timing of the mood changes of the patient, and to discontinue antidepressant medication if there is evidence of drug-induced rapid cycling. 

Recently, estrogen skin patches have been reported to be beneficial in severe postpartum depression.  For postpartum dysphoria, Dalton (21) has recommended progesterone treatment (100 mg IM for the first postpartum week and then 400 mg bid by suppository for two or more months postpartum).  However, some clinicians and investigators find that progesterone may actually exacerbate depression.  For severe or psychotic postpartum depression or mania refractory to pharmacotherapy, electroconvulsive therapy (ECT) remains the treatment of choice.  Sleep deprivation has therapeutic efficacy in a majority of patients with major depressive disorders.  The efficacy of sleep deprivation in postpartum non-psychotic depression currently is under experimental investigation (Parry et al, unpublished observations).  The relapse that may occur with recovery sleep after postpartum sleep deprivation potentially may be averted with lithium, in non-breast feeding women.

Since the experience of a postpartum depression or psychoses can be very disruptive cognitively and emotionally for the woman and her family, these disorders, like other psychiatric disorders, are best treated with a combination of pharmacologic and psychotherapeutic management with the aim of providing education, support, and cognitive structuring whereby these patients and their families can attempt to gain "some method out of the madness" which stems from this very confusing, disorienting, and emotionally traumatic cataclysm in their lives. Preliminary studies have indicated significant decreases in depressive symptoms using interpersonal psychotherapy during pregnancy (105), cognitive-behavioral counseling or fluoxetine for postpartum depression (3), and group treatment with educational, supportive, and cognitive-behavioral components for postnatal depression (61). Further research with standardized, empirically-validated therapeutic approaches is necessary in this area.

Anthropological studies indicate that other cultures have rituals allowing for 40-day rest periods for the mother after the birth of a baby in which to "mother the mother." During this time period, the focus is on allowing the mother time to rest, recuperate, eat and sleep.  Female relatives come to the home to prepare meals, do housework and care for the infant.  Thus, social support, education, child care services, and social recognition of the new motherhood status is ensured.  Previously, a week hospital stay for the mother after delivery was required.  Now, in modern cultures, the mother usually goes home a day after delivery, and often without extended family or neighbors to help with infant care.  In this isolated environment, there are not the supportive therapeutic factors available that would otherwise help to mitigate against the development or exacerbation of the spectrum of non-psychotic depressions. In addition, demographic risk factors, such as an unplanned pregnancy, not breast-feeding, unemployment after childbirth, and being head of the household may place additional strain upon the mother and increase the likelihood of postnatal depression (118).

Maternity Blues

The maternity blues is not considered a disorder since it occurs in 50-80% of women and because of the absence of major symptomatology.  It is best treated with reassurance that the symptoms occur in a majority of women, and that they generally improve spontaneously within a week to ten days.  In rare instances the symptoms may progress to a more severe postpartum disorder, which stresses the necessity of making frequent follow-up visits.  However, this progression is the exception rather than the general rule.  In contrast to postpartum psychosis, pharmacologic intervention generally is not warranted for the maternity blues.  Instead, psychotherapeutic intervention in the form of education, support and reassurance, has more import.


Given that there is a high recurrence rate for both postpartum psychosis (initial risk 1/500; subsequent risk 1/3) and postpartum depression (initial risk 1/10; subsequent risk 1/2), prophylactic treatment for women, particularly for those who have a previous history of affective illness, is an integral part of the management of these disorders. Patients with a previous history of nonpuerperal affective illness are three times more likely to develop postpartum mood disorders, particularly mania.  Thus, one of the most effective prophylactic interventions in this group is lithium.  Although lithium dosage should be halved about one week before delivery because of marked fluid and electrolyte changes occurring then, it can be restarted shortly thereafter.  Lithium is contraindicated in breast-feeding women.  Clinicians particularly should be on alert for lithium induced hypothyroidism in postpartum women, since 90% of patients who develop hypothyroidism on lithium are women (15) and the postpartum period presents a particular risk factor for the development of hypothyroidism (4).  Furthermore, this condition, i.e., postpartum hypothyroidism, can induce rapid mood cycling.

Patients with a previous history of affective disorders may have an exacerbation of their illness during pregnancy.  Although lithium is contraindicated during the first trimester because of the infant's risk for Ebsteins anomaly of the heart, in severe cases lithium may be administered cautiously, checking particularly for fluid and electrolyte changes, during the third trimester.  For mania occurring during pregnancy, neuroleptics (with careful monitoring) or ECT can be given without undue risk to the fetus. To date, there have been few published reports assessing the efficacy of the anticonvulsants on manic symptoms in the puerperium (17, 135); the majority of work on teratogenicity of anticonvulsant drugs are based on women with epilepsy. Carbamazepine and valproate may expose the fetus to a substantial risk of congenital malformation and/or developmental delay (100, see also Lithium and the Anticonvulsants in Bipolar Disorder).  Because of the risk of neural tube defects with anticonvulsants, lithium treatment is preferable.  More data needs to be gathered to determine whether anticonvulsants are safe for breastfeeding; sodium valproate appears to be relatively safe. Valproate concentrations are smaller as compared with carbamazepine; at the same, valproate has been associated with an increased risk of hepatotoxicity in young children (5, 135).  The World Health Organization and American Academy of Pediatricians suggest that valproate may be safer than carbamazepine.

Another prophylactic treatment which has received attention, although is controversial, is progesterone (100 IM after labor, daily for 7 days, then progesterone suppositories for two months or until the return of menstruation) (21).  Since progesterone is essentially an anesthetic in animals, its use in humans is probably more effective for the agitated rather than the depressive symptomatology of postpartum psychiatric syndromes.  It also may exacerbate depressive symptomatology.

Additional information contributes to the hypothesis that postpartum illnesses are unique and organic in etiology.  There has been successful use of three different substances for prophylaxis in high-risk patients, i.e., patients who have had previous postpartum psychosis or depression.  Administration of long-acting parenteral estrogen or progesterone has seemed to ward off recurrences.

In summary, the severity and recurrence of postpartum psychiatric disorders deserve early, aggressive and innovative treatment approaches.  Their presentation is often episodic and fluctuating, with different presentations of symptoms being more related to different stages of the illness than to different categories of the illness.  Treatment requires longitudinal follow-up care.  Because postpartum psychiatric disorders appear to exhibit a pattern consistent with a model of kindling and behavioral sensitization (as opposed to a model of tolerance), it is crucial in treatment strategies to interrupt this cycle using early and aggressive treatment and prophylactic management whether it is by ECT, lithium, hormonal or chronobiologic interventions such as sleep deprivation or phototherapy.

• See Table 2 and Table 3

Premenstrual Depression

One clinical model for studying the relationship of gonadal hormones to affective illness is the affective changes associated with the menstrual cycle.  Historically referred to premenstrual syndrome (PMS), this condition has been more rigorously defined as late luteal phase dysphoric disorder (LLPDD) in the DSM-III-R and as premenstrual dysphoric disorder (PMDD) in the DSM-IV under mood disorders.  For purposes of familiarity, the term PMS will be used in this chapter.  In PMS, the mood and behavioral changes are recurrent and predictable and thus can be studied prospectively and longitudinally.  Similar to winter in SAD, and the postpartum period in affective illness, the late luteal phase of the menstrual cycle is a vulnerable time for the development of depressive mood changes. Therefore, medication discontinuation and/or modification should consider changes in therapeutic levels as a fluctuation of menstrual cycle phase; this may prevent mood relapse at a vulnerable time.

Studies indicate that PMS may be related to major depressive disorders.  In support of this hypothesis, patients with PMS and affective disorders, in contrast to patients with anxiety disorders (93), respond to sleep deprivation: Total and late-night partial sleep deprivation temporarily alleviate symptoms in a majority of patients with major affective disorders (95).  Eighty percent of patients with premenstrual depression responded to a night of total sleep deprivation and that late-night partial sleep deprivation (in the second half of the night) was more effective than early-night partial sleep deprivation (in the first half of the night) (80).  In a follow-up study, PMS subjects responded equally well to both early and late night partial sleep deprivations, but only after a night of recovery sleep (75).  Sleep deprivation lowers prolactin (71) and increases TSH (96) although at least in one of the studies (51) these hormonal changes did not correlate with clinical response.  An effective intervention with total or partial sleep deprivation in patients with PMS would be consistent with current theories that implicate prolactin and thyroid disturbances in the pathogenesis of PMS.  Sleep reduction also may serve as a final common pathway in the genesis of mania in postpartum psychiatric illness (124).  Thus, the interaction of sleep with a sensitive circadian phase of thyroid or prolactin secretion may be a common predisposing factor for the development of affective illness, premenstrual depression, and possibly postpartum mood disorders.

Clinical psychopharmacologic treatment

Nutritional Supplements

Vitamin B6 (pyridoxine) has been used to treat premenstrual mood symptoms because of its purported efficacy in treating oral contraceptive induced-depressions (79).  This effect is related to observations that estrogen may increase tryptophan metabolism via the kynurenine-niacin pathway and thus increase the requirements for B6.  A slight decrease in the excretion of tryptophan metabolites in women suffering from PMS symptoms has been found, although plasma levels of pyridoxal phosphate generally are not different in women with and without premenstrual symptoms.  Controlled studies suggest mixed results for pyridoxine; some find B6 to be better than placebo whereas other controlled studies find it no better than placebo.  Some of the studies affirming efficacy are confounded by women taking concomitant hormones, psychotropics, or diuretics, or by reported improvements not being specific to the premenstrual phase.  Pyridoxine may have weak effects on global ratings, behavior, and social activities (132), although a significant amount of physical and mood symptomatology may remain.  High and prolonged doses of pyridoxine may be associated with neurological toxicity.  In PMS studies very few of these symptoms have been noted in controlled studies.  There has been no dose response relationship evident for B6 in reported studies; significant improvements have been reported at doses as low as 50 mg.

The evidence supporting vitamin E's efficacy over placebo is not presently convincing.  Vitamin E has been studied for the treatment of premenstrual symptoms, with improvements in motor coordination reported.  Significant improvements in premenstrual anxiety and depression found in earlier studies of women with fibrocystic breast disease, have not been replicated.  For vitamin A, early uncontrolled studies are encouraging but no placebo controlled studies are available to support its use.

One recent controlled study indicates that oral magnesium is effective in relieving premenstrual mood changes (27).  Magnesium deficiency may activate premenstrual symptoms through various means.  Significant magnesium deficiency in red blood cells of PMS patients initially reported warrant further follow-up.  Although magnesium supplements have been recommended, consistent reports linking treatment of low blood magnesium levels with reduction of premenstrual symptoms have not been found.


Theories of PMS often attribute symptoms to fluctuations of ovarian steroid hormones.  Thus, several studies have tested hormonal treatments.  Although some studies support low luteal progesterone, high estrogen, or low progesterone-estrogen ratios being etiologic, other studies do not implicate high estrogen levels relative to progesterone.  Some studies support the possibility of an asynchrony between declining rates of progesterone and estrogen, early ovulation, or decreases in progesterone over time relative to estrogen.

Consistent with the studies suggesting that estrogen is high relative to low progesterone, several investigators have treated PMS symptoms with progesterone or other progestin compounds.  However, efficacy claims for the progestins are based primarily on uncontrolled studies. Dalton (20) reported good results with open progesterone administration, and Dennerstein et al (23) found beneficial trends during the first but not second month in a controlled study of oral progesterone. In a double-blind, placebo controlled trial, Baker et al. (7) found that administration of vaginal progesterone suppositories may alleviate some symptoms of anxiety and irritability in women with moderate to severe premenstrual syndrome. Symptom improvement was noted only in a subcategory of questions relating to guilt and self-image. No differences were found on several instruments measuring dysphoria (e.g. Hamilton depression and anxiety, Beck depression inventory, etc.), and the researchers did not use standardized diagnostic criteria to screen subjects.

Other studies have not found progesterone to be superior to placebo (28). Furthermore, administration of the menses-inducing, progesterone antagonist mifepristone altered neither the severity nor the timing of symptoms in women with prospectively diagnosed PMS (99).  These findings suggest that hormonal events during the late luteal phase are not the direct cause of mood disturbance.  The authors hypothesized that mood disturbances may be synchronized, or linked, with the reproductive cycle, but not caused by the hormonal changes of the cycle itself.

With synthetic progestins, several open or single blind trials report improvement rates of 50-82%.  When compared to placebo, however, the synthetic progestins are not found to be effective over placebo, including ethisterone, norethisterone and, with some exceptions, medroxyprogesterone acetate, when given short of inhibiting ovulation.  Ylostalo and colleagues (138) found norethisterone effective for breast tenderness. 

Most of the evidence does not support either natural or synthetic progestins as effective treatments of premenstrual mood symptoms.  It appears that the improvements attested to this treatment are likely attributable to a placebo effect, except perhaps where given in a manner sufficient to induce anovulation.

Ovarian steroid treatment studies using oral contraceptives also have yielded inconsistent results.  Several studies suggest that premenstrual moodiness, irritability, fatigue, and depressed mood may be less commonly reported in women using various oral contraceptives than in women not using them.  However, other studies have found little difference in premenstrual symptoms between oral contraceptive users and non-users.  The results suggest that suppression of ovulation is not curative of PMS, when gonadal steroids are yet present from exogenous administration.  Given that steroids have psychological effects, it can be difficult to predict individual responses in different women.  Some data, for example, suggest that the estrogen-dominated pills may adversely affect women with premenstrual irritability, whereas progesterone-dominated pills adversely affect women with premenstrual depression.  More research is necessary to determine whether specific identified subgroups would reliably benefit from oral contraceptive treatment of premenstrual mood symptoms.

Gonadotropin Releasing Hormone

At a more central level, interruption of hypothalamic-pituitary-ovarian cyclicity may relieve premenstrual symptoms, when amenorrhea is induced.  Gonadotropin releasing hormone agonists (GnRH), which with chronic administration down-regulate pituitary gonadotropin secretion, have been used with some success in several studies. Muse and colleagues reported that a GnRH agonist robustly attenuated both behavioral and physical symptoms in 7 women over a 3 month period compared to placebo (67).  In this study, GnRH effectively interrupted the menstrual cycle (reversibly) during a three month administration, with amenorrhea induced during the second and third cycles.  There were no significant differences in behavioral symptoms between GnRH and placebo during the first cycle, although physical symptoms improved during the first cycle. 

Two subsequent studies have since used the intranasally administered form of GnRH agonist, buserelin.  Bancroft et al (8) found a total daily dose of 600 mg of buserelin prompted improvement in bloating and breast tenderness with less clear effects on mood in 10 of 20 women studied openly.  The other 10 women suffered adverse effects such as a worsening of their symptoms.  Hammarback and Backstrom (42) found GnRH to be better than placebo for mood, swelling and breast tenderness, although 3 of 26 women experienced worsening of their symptoms at 400 mg per day.

In a double-blind, placebo-controlled study, West and Hillier (129) administered goserelin, a GnRH agonist analogue, to 27 women diagnosed with premenstrual tension. Over the course of three months, they found that the GnRH analogue significantly reduced physical symptoms, such as breast discomfort and swelling. Statistical trends also suggested that goserelin may dampen cyclical fluctuations of anxiety and reduce irritability. Less consistent improvement was noted with depressive symptomatology. The authors also noted a substantial placebo effect.

The GnRH agonist, leuprolide acetate, has also been found to be effective in treating premenstrual mood disturbances.  Brown et al. (10) administered the depot leuprolide intramuscularly for three consecutively cycles in 25 women with late luteal dysphoric disorder. Depot leuprolide was ineffective for women with severe premenstrual depression and associated with adverse events. For women with less-severe symptoms, depot leuprolide was well-tolerated and effective in reducing symptoms. The authors speculated that severity of symptoms may be indicative of diagnostic subtypes; women with severe premenstrual depression may have a more sensitive hypothalamic-pituitary-ovarian feedback system.

Freeman et al. (30) compared women with PMS, as documented by daily mood ratings, and women with dysphoric symptoms throughout the cycle who experience premenstrual symptoms. They found that depot leuprolide was significantly more effective than placebo for subjects with PMS but not for subjects with premenstrual exacerbations of dysphoric disorders.  The reduction in dysphoric symptoms for women with PMS did not occur until the third cycle. The authors suggested that the differential responses to the GnRH agonist may indicate that premenstrual depression has mechanisms different from those of other dysphoric mood disorders.

For eight weeks, Schmidt and colleagues (98) administered leuprolide or placebo to 20 women with premenstrual syndrome in a double-blind, crossover design. Leuprolide treatment in women with prospectively documented PMS was associated with a significant reduction in symptoms as compared with placebo and baseline values. When given leuprolide plus estradiol or progesterone, the women with PMS had a significant recurrence of symptoms; normal women given the same regimin experienced no changes in mood.  According to Schmidt and colleagues (98), these findings suggest that normal plasma concentrations of gonadal steroids may trigger a disruption of mood in susceptible women.

This form of treatment may be considered a medical ovariectomy (67) and the potential hazards for long term use, such as increased cardiovascular morbidity and mortality, and osteoporosis, have not been clarified.  Estrogen supplementation will not reverse the effects (66).


Danazol is another antigonadotropin which causes hypothalamic pituitary-gonadotropin suppression, as a synthetic androgenic derivative of ethisterone.  Preliminary reports indicate that this steroid is better than placebo for negative affect, pain, and behavioral change at doses of 100-400 mg per day.  Double-blind studies (97, 119, 37) indicate that danazol compared with placebo significantly lowers premenstrual symptoms of lethargy, irritability and anxiety in the premenstrual phase.  Side effects of nausea, giddiness, skin rash, flushing, vaginitis, musculoskeletal and breast pain, decreased breast size, weight gain, mild hirsutism, and depression have been reported.  Thus treatment with danazol should be considered only after other treatments fail, and when symptoms are severe.  It should not be considered in women contemplating pregnancy or nursing.  As such, danazol's usefulness may be limited because of its androgenic properties in women of childbearing age.  As a trend in these studies, the 200 mg/day dose appears to be better tolerated.  Its efficacy is comparable to the 400 mg/day dose in most patients.

Prolactin Inhibition with Bromocriptine

Several studies have tested bromocriptine for treatment of premenstrual symptoms because of the observed increases in prolactin levels during the luteal phase.  Although most studies do not find abnormalities in prolactin levels in women with premenstrual symptoms, most, but not all studies indicate bromocriptine to be effective primarily for the treatment of mastodynia.

Significant improvements in mood, depressive symptoms (25), and irritability (138) have been sporadically reported, in double-blind, and single-blind placebo-controlled studies, respectively.  Limiting side effects have been noted in approximately 20% of women treated with bromocriptine; side effects can include nausea, headache, vomiting, dizziness, fatigue and paroxysmal tachycardia.

Salt and Water Balance

Diuretics have been studied related to the observations that fluctuations in capillary filtration rate and permeability to plasma proteins occurs premenstrually.  Women with premenstrual symptoms also have been found to have a higher luteal body water/potassium ratio than controls.  Early uncontrolled studies of diuretics indicated satisfactory results with various compounds, including ammonium chloride, chlorothiazide, chlorthalidone, and quinethazone.  However, neither chlorthalidone nor potassium chloride has been found to be better than placebo in double-blind placebo studies.  For bloatedness, two studies (69, 116) suggest the superior efficacy of spironolactone over placebo.

The sulfuramide diuretic, metolazone, 1-5 mg daily, was found to be effective for irritability, tension, depression, headache and water retention symptoms (128) in a placebo-controlled diuretic study that included only women with premenstrual weight gain.  Excessive diuresis and weakness occurred occasionally, with the 5 mg dose. Thus, some evidence supports diuretic or antimineralocorticoid efficacy.  The efficacy of such interventions, however, may be limited to subgroups of women with premenstrual weight gain.

Prostaglandin System

Women with premenstrual symptoms have also been treated experimentally with compounds affecting prostaglandin metabolism.  Prostaglandins are known to mediate dysmenorrheic somatic complaints, and also have been investigated in women with premenstrual mood complaints.  Linoleic acid, the main precursor for PGE-1, was elevated in women with premenstrual symptoms in one study (11) whereas the PGE-1 product was lower in these women in another study (47).  One interpretation is that the conversion of linoleic acid to PGE-1 is impaired in symptomatic women.

Mefenamic acid, a prostaglandin synthesis inhibitor, has been found more

effective than placebo for mood symptoms in some studies (64, 137), although Gunston (35) found no significant benefit on mood for mefenamic acid over placebo.  In this study, there were improvements in gastrointestinal symptoms.  Budoff (12) suggests that mefenamic acid improves breast tenderness, ankle swelling and abdominal bloating with little effect on mood symptoms of tension, lethargy and depression.  Overall, the effects of mefenamic acid on mood have been inconsistent in comparison to its effects on pain.  Many of its effects on affective symptoms have been obscured by the inclusion of dysmenorrheic women in study populations.  One methodologically sound study (64) however, that excluded dysmenorrheic subjects, did find positive effects on mood of active drug over placebo.  Significant decreases in premenstrual water retention and "arousal" symptoms have been reported with a diet low in fat that reduces the availability of prostaglandin precursors (49).

Efamol (evening primrose oil), a prostaglandin synthesis precursor (containing linoleic acid precursors, gamma linoleic acid, and vitamin E) also has been tried as a strategy to augment PGE-1 from administration of these essential fatty acid precursors.  Although more effective than placebo for depressive symptoms in one study (85), another study was unable to find significant differences between placebo and Efamol (13).  Again, the results of the treatment effects on premenstrual mood symptoms, distinct from dysmenorrhea, are complicated by inclusion of women with dysmenorrheic symptoms.


Because of the complex effects of the hypothalamic-pituitary-gonadal axis and CNS neurochemistry, some treatment studies have aimed to influence neurotransmitters, or neuromodulators more directly.  The central alpha-2 adrenergic presynaptic autoreceptor agonist properties of clonidine have been theorized to compensate for a putative excess of central noradrenergic activity via presynaptic noradrenergic inhibition.  One placebo comparison of clonidine (32) indicated efficacy over placebo in reducing premenstrual psychiatric rating (BPRS) scores.  Clonidine can decrease plasma renin activity and promote aldosterone secretion, and has been used for the treatment of opiate withdrawal.  The promising results from this initial controlled trial deserve further study in order to determine which specific symptoms may respond best to stimulation of this inhibitory autoreceptor that governs the release of serotonin as well as norepinephrine.

In a double-blind crossover study (16),  naltrexone was recently employed in an attempt to discover whether an opiate antagonist, given before an expected midcycle rise and fall of B-endorphin, would inhibit the effects of a putative premenstrual endogenous opioid withdrawal.  The naltrexone treatment was modestly but significantly better than placebo, with the largest improvements found in concentration difficulties, behavioral changes, and negative affect.  It was given only on days 9-18 of the cycle, such that patients may have been in an opiate antagonist withdrawal state during the late luteal phase.  These results suggest that either an increased luteal opiate receptor sensitivity or a blockade of midcycle agonist occupancy may be therapeutic for premenstrual dysphoria.  More work is needed to determine replicability and if the dose or schedule could be optimized to augment the benefit or decrease the side effects of nausea, decreased appetite and dizziness.

Lithium has been tried for premenstrual mood symptoms because of the thymoleptic action of the compound observed in patients with affective disorders.  Rubinow and Roy-Byrne (94) reviewed lithium treatment in premenstrual tension, citing three open trials with positive results which were not confirmed in two subsequent double-blind, placebo-crossover studies.  For recurrent suicidal depressions, or rapid cycling mood disorders linked to the menstrual cycle, lithium may well be the treatment of choice.  Other authors have suggested that patients who respond to lithium may be those who meet diagnostic criteria for cyclothymic disorders.  Glick and Stewart (33) reported lithium to be effective in open trials of three schizophrenic patients with premenstrual exacerbations of schizophrenia.  However, even at low doses (600-900 mg/day), most patients may detect significant lithium-related side effects.  Lithium's teratogenecity, of course, indicates caution in women of childbearing potential.

Alprazolam also has been found in two published reports to be more effective than placebo for mood symptoms and global improvement (43, 104). The agreement between studies is encouraging and suggests that alprazolam may indeed be a useful treatment for premenstrual mood symptoms, although, some questions remain as to whether there is attribution secondary to the subjective effects of alprazolam.  Longer term studies now are needed to determine whether dependence or withdrawal symptoms would be problematic over a longitudinal intermittent treatment regimen of luteal alprazolam and follicular drug withdrawal.

Given the overlap of premenstrual mood symptoms with that of depression or anxiety, recent studies have focused on the use of antidepressants.  Open studies have indicated encouraging results with nortriptyline, clomipramine, fluoxetine, nefazodone, and paroxetine (2, 139).  Double blind trials indicate that fluoxetine, sertraline, paroxetine, clomipramine, imipramine, and phenelzine may be effective (2, 26). Currently, SSRIs, such as fluoxetine and sertraline are the treatment of choice for premenstrual dysphoria. In large-scale, double-blind studies (109, 140), both have been found to be effective, compared with placebo, when given continuously throughout the cycle. Preliminary studies (40, 108, 142, 29) have indicated that women with PMDD who respond to continuous sertraline or fluoxetine treatment also respond to treatment restricted to the luteal phase. Larger controlled trials are necessary to confirm these findings.  Intermittent treatment may have advantages in reducing side effects and increasing medication compliance; at the same time, an irregular dosing schedule may increase the likelihood of forgetting to take medications as prescribed.  The efficacy of the SSRIs in treating this disorder is predicated upon the documented differences in serotonergic activity between women with PMDD and normal controls (41). Although several lines of evidence suggest luteal fluctuations of MAO activity, there is little information available on MAOI trials.  Antidepressants may be indicated for the more severe PMS mood symptoms, although premenstrual dysphoric changes and irritability may continue despite effective antidepressant treatment (141).

The 5-HT1A partial agonist, buspirone, has also been used in the treatment of premenstrual mood symptoms.  Rickels and colleagues (90) have shown that buspirone was significantly more effective than placebo for irritability, fatigue, pain and social functioning in 34 patients treated with a mean daily dose of 25 mg 12 days prior to menstruation.  Buspirone can cause lightheadedness, headache, or gastrointestinal distress primarily when the dose is first being adjusted, but lacks apparent dependence potential.  It is not known whether the therapeutic effects reported for fenfluramine, nortriptyline, clomipramine, fluoxetine and buspirone might all be mediated through 5-HT1A receptor mechanisms.

Chronobiologic Manipulations

Some nonpharmacologic experimental treatments for PMDD include sleep deprivation and light therapy. The efficacy of these types of treatment are based on a chronobiologic model of depression; this model posits that affective disorders may arise due to disruptions in circadian rhythms or a desynchronization of the oscillator regulating these rhythms with the sleep-wake cycle, each other, or the environment.

One night of total sleep deprivation during the symptomatic premenstrual phase of the menstrual cycle was found to alleviate PMS symptoms.  Follow-up studies suggest only one night of partial sleep deprivation (sleep 9 PM - 1 AM or 3 AM to 7 AM) may be beneficial (80). Light therapy involves sitting in front of a box of lights 2 hours a day for one week. Although early studies suggested superior efficacy of evening (7-9 PM) bright light (>2500 lux, 5 times brighter than room light) (74), subsequent trials showed equal efficacy of bright evening, bright morning (6:30-8:30 AM) and dim (10< lux) red evening light administered in the premenstrual phase to PMS patients (77).  Although these treatments may show promise, further trials are needed before these interventions can be recommended for general clinical usage.  These findings are interesting in comparison with recent evidence suggesting a seasonal variation in PMS with approximately 70% of the women experiencing fewer symptoms during the summer when the photoperiod is naturally longer.

• See Chart 1


A depression occurring in association with the reproductive cycle may sensitize a woman to future depressions.  A previous history of psychiatric illness or of affective changes during pregnancy may predispose to oral contraceptive-induced depressions (79). Severe premenstrual depression may predispose a woman to postpartum depression and, like affective illness, premenstrual depression may have its onset or be exacerbated after a postpartum depression.  Alternatively, a major affective disorder may be exacerbated or precipitated during a premenstrual period.  There are anecdotal reports that after treating cycles of bipolar illness with lithium, premenstrual and seasonal mood cycles persist or become more prominent.  Price et al (84) report patients with rapid cycling disorders have an increased tendency to have more severe forms of PMS, although Wehr et al (122) found no convincing relationship between manic-depressive cycles and menstrual cycles in their patients with rapid cycling disorders.

Thus, the cyclicity of affective disorders in the form of rapid cycling bipolar illness or SAD may be compounded by periodic affective change occurring in association with the premenstruum, with pregnancy and the postpartum period, and with altered reproductive hormonal milieus induced by oral contraceptives or gonadal hormone treatments.


With the kindling model of depression in mind (82), one wonders whether such periodic reproductive-related depressions may sensitize women to future affective episodes.  Most longitudinal studies to date have not specifically focused on gender-related differences in course of the illness.  Sex differences in depression begin to appear after the onset of puberty in adolescence (126).  There is a marked increase in major depression in female children at approximately 16 years of age, a time of major change in the neuroendocrine reproductive axis.  In contrast, male children exhibit a gradual increase in depression across all ages, with considerably lower absolute rates than in female children.  Furthermore, the onset of major depression is earlier (12 to 13 years of age) in both male and female offspring of depressed probands (125), and the risk for depression appears to be increasing with time (36, 125, 127).

How cyclic depressions related to reproductive events may affect other forms of cyclic affective disorders is unknown, but a relationship does seem to exist.  Work in animals may provide models and possibly shed light on the mechanisms involved.  For example, the predisposition of women with thyroid impairment to cyclic forms of depression has a parallel in an animal model. Richter (89) produced abnormal cycles of motor activity experimentally in female animals, but not male animals, by partial thyroidectomy.  As in rapid cycling patients, treatment with thyroid extract abolished the abnormal behavioral cycles, which returned after cessation of treatment.  Richter hypothesized that the abnormal, regular cycles of activity were produced by the effect of thyroid deficiency on homeostatic mechanisms controlling luteotropin (prolactin) release (possibly related to the effect of TRH on prolactin).  He induced similar cycles by daily subcutaneous injections of prolactin.  Such cycles were also produced by inducing pseudopregnancy.  This condition stimulates pituitary secretion of prolactin, which acts on the ovary to produce persistent corpora lutea and the secretion of progesterone.  Ovariectomy abolished running activity; estrogens increased it.  Giving the rats anhydrohydroxy progesterone produced longer abnormal activity cycles.  As occurs in affective illness, the abnormal activity cycles become shorter with time.  Of relevance here is our clinical work demonstrating higher baseline prolactin levels (Parry et al, unpublished observations) and increased prolactin response to TRH in women with rapid cycles of mood related to the menstrual cycle (PMS) (76).

Reproductive hormones modulate hormonal, neurotransmitter, and biologic clock mechanisms that have each been the focus of hypotheses about the pathophysiology of affective disorders.  Estrogen and progesterone can alter the biosynthesis, release, uptake, degradation and receptor density of norepinephrine, dopamine, serotonin, and acetyleholine (63).  The gonadal steroids also modulate other hormonal mechanisms (thyroid, cortisol, prolactin, and opiates) that also affect neurotransmitter systems.

Gonadal hormones affect biologic clock mechanisms, which also have been implicated in the pathophysiology of affective disorders.  Estrogen shortens the period of circadian activity in ovariectomized hamsters and rats (1, 65).  The onset of activity occurs earlier on days of the estrous cycle when endogenous titers of estradiol are high in intact hamsters.  Progesterone delays the onset of activity in intact rats by antagonizing the effect of estrogen (6).  Estrogen, in addition to shortening the free-running period and altering the phase relationship of the activity rhythm to the light-dark cycle, increases the total amount of activity and decreases the variability of day-to-day onsets of activity.  These findings parallel clinical work by Wever (130) who found that the mean free-running period of the sleep-wake cycle is significantly shorter in women than in men (28 minutes).  The wake episode is shorter (1 hour 49 minutes) and the sleep episode longer (1 hour, 21 minutes) in women compared with men.  Thus, the fraction of sleep is longer for women than men.  The circadian temperature rhythm was similar in both sexes.  According to Wirz-Justice et al (133), women sleep longer than men at all times of the year.  When in free-running conditions, women, compared with men, tend to become internally desynchronized, particularly in the summer, by shortening the period of their sleep-wake cycle. 

Estrogen also serves to enhance coupling between different circadian pacemaker components (1).  Ovariectomized female rodents develop rhythm desynchronies.  Replacement with estrogen restores the normal coupling relationship between these disparate rhythms.  This factor may be involved in some of its therapeutic effects.

Thus, the decline of ovarian hormones common to the premenstrual, postpartum and menopausal periods, and their inherent cyclicity, may destabilize or sensitize neurotransmitter, neuroendocrine, and biologic clock mechanisms, and thereby set the stage for the development of cyclic affective disorders.

Although the cyclicity of the endocrine milieu may increase the vulnerability to episodic depressions in women, it may protect against the development of many chronic illnesses, which are more characteristic in men. Mood cyclicity may contribute to longevity by enhancing homeostatic mechanisms. Indeed, work on hamsters has shown that longevity increases with the restoration of synchronous high amplitude circadian rhythms (46). The investigation of hormonal contributions to affective illness in women and the examination of the way in which the course of these illnesses is affected by reproductive events of the life cycle may increase our understanding of affective illness and potentially provide alternative treatment strategies.


Women are at higher risk than men to develop depressive episodes during the reproductive years.  Furthermore, women are vulnerable to depressions associated with oral contraceptives, abortion, the premenstrual period, the puerperium, and menopause.  The phenomenology and the biologic mechanisms involved in these illnesses perhaps should be viewed in the context of other manifestations of the link between depression and female reproductive functions.  For example, women are especially vulnerable to a rapid cycling form of affective illness and to hypothyroidism, an associated factor for this form of affective disorder.  The postpartum period also is associated with impaired thyroid function, and there are reports of the induction of rapid cycles of mood following the termination of pregnancy.  Thus, alterations in thyroid hormones may be a feature of both postpartum and rapid cycling forms of affective disorder in women.

A previous history of a postpartum depression places a woman at a high risk for the development of a subsequent puerperal episode.  Also, difficulties during pregnancy may predispose a woman to the development of other reproductive-related depressions.  The role reproductive hormones play in this possible sensitization phenomenon needs to be examined in order to understand the relationship of depression to the female reproductive cycle.  Appropriately timed clinical psychopharmacologic interventions may serve to inhibit this sensitization phenomenon and benefit long term prognosis.

published 2000