Personality disorders in women with severe premenstrual syndrome

Abstract

Premenstrual syndrome (PMS) and its more severe form, premenstrual dysphoric disorder, affect up to 18% of women. Both are commonly associated with other mood-related disorders such as major depression, and cause significant life impairment, but their relationship with personality disorders is less clear. After completing the Structured Clinical Interview for DSM-IV-TR disorders, 33 women with severe PMS and 26 asymptomatic women, counterbalanced for menstrual cycle phase, were administered the Structured Interview for DSM-IV Personality Disorders, a diagnostic interview with low transparency, strong inter-rater reliability, and good diagnostic clarity. Women with severe PMS had a higher prevalence of personality disorders (p=0.003) than asymptomatic women (27% versus 0%), and were more likely to have odd–eccentric, dramatic–erratic, and anxious–fearful personality disorder traits (p< 0.05). Obsessive–compulsive personality disorder (OCPD) was the most common character pathology in the PMS group (n=6, 18%). OCPD, although not necessarily associated with greater severity of premenstrual symptoms, was related to poorer life functioning in women with PMS. The comorbidity of a personality disorder and severe PMS places an additive burden on general life functioning and may have implications for psychiatric treatment or medication given to those with severe premenstrual symptoms.

Introduction

Premenstrual syndrome (PMS) describes a range of affective, behavioral, and physical symptoms which occur during the luteal phase of the ovulatory menstrual cycle and abate following menstruation (Freeman 2003). Upwards of 18% of women have severe PMS, with symptoms that cause significant functional impairment and negatively affect quality of life (Wittchen et al. 2002). Approximately 3–8% of women qualify for a diagnosis of premenstrual dysphoric disorder (PMDD) (Halbreich et al. 2003; Wittchen et al. 2002). Originally known as late luteal phase dysphoric disorder (LLPDD) in the DSM-III-R (American Psychiatric Association 1987), the syndrome was renamed PMDD in the DSM-IV (American Psychiatric Association 1994). Common critical components of a PMS or PMDD diagnosis include symptom severity and timing as well as the impact of symptoms on functioning. Whereas those with PMDD must report experiencing at least 5 of 11 symptoms, a diagnosis of PMS does not require a certain number of symptoms to be present. Women with PMS, and especially those with PMDD, are more likely to have a history of an axis I mood disorder (Halbreich et al. 2003; Yonkers 1997), particularly major depressive disorder or postpartum depression (Critchlow et al. 2001; Pearlstein et al. 1990). The relationship between PMS and axis II personality disorders is less clear, likely attributable in part to nosologic and methodological diagnostic difficulties (Critchlow et al. 2001).

Categorical assessment approaches for studying personality disorder prevalence in women with severe premenstrual problems have generally revealed low estimates. One of the earliest studies found a 10% prevalence of axis II personality disorders in 78 women with LLPDD using the Structured Clinical Interview for DSM-III-R Personality Disorders (SCID-II); this percentage was not elevated in comparison with the limited population prevalence estimations at the time (Pearlstein et al. 1990). In this study, the most common axis II diagnosis was avoidant personality disorder with a prevalence rate of 6%. Another study found that the prevalence of personality disorders diagnosed by SCID-II in women with LLPDD depended on age: personality disorders were not prevalent in younger LLPDD women, whereas LLPDD women over age 30 years were more likely to be diagnosed with avoidant personality disorder than controls (DeRonchi et al. 2000). Finally, Critchlow and colleagues (2001) found no difference in the number of absolute personality disorder diagnoses between women with PMDD and controls using the International Personality Disorders Examination.

Investigations of personality styles in women with PMS or PMDD have produced variable results. An early study found that extreme deviations from the mean on questionnaire-assessed obsessional and hysterical personality types were related to greater premenstrual symptoms and complaints (James and Pollitt 1974). Critchlow et al. (2001) found more obsessive–compulsive personality traits in PMDD women compared with controls. Parry et al. (1996), on the other hand, found that women with LLPDD scored lower than control women on items reflecting a compulsive–conforming personality style on the self-report Millon Clinical Multiaxial Inventory. They also found that women with LLPDD scored higher on borderline–cycloid and passive–aggressive–negativistic personality styles than control women, suggesting that women with LLPDD have an increased vulnerability to mood-related lability and disorder. Measures did not vary between the follicular and luteal phases (Parry et al. 1996), providing support for trait, rather than state, pathology. However, Berlin and colleagues (2001) found that some personality styles in women with PMS varied with menstrual cycle phase whereas others did not. Women with PMS had higher scores on histrionic, obsessive–compulsive, self-defeating, and dependent subscales of the Personality Disorder Questionnaire—Revised than asymptomatic controls in both the follicular and late luteal phases. By contrast, relative to controls, women with PMS had higher scores on the schizoid, schizotypal, borderline, narcissistic, avoidant, and passive–aggressive subscales but only in the late luteal phase. In women with PMS, 33% were found to have personality dysfunction in the luteal phase, but only 13% in the follicular phase. The authors concluded that personality dysfunction in women with PMS may have state and trait-related components (Berlin et al. 2001). Of the limited research described above examining comorbidity of severe premenstrual problems and personality, few studies used assessment instruments that directly assess DSM personality disorder criteria.

The Structured Interview for DSM-IV Personality Disorders (SIDP; Pfohl et al. 1995) assesses presence of 10 DSM-IV personality disorders plus 3 optional disorders: self-defeating, depressive, and negativistic personality disorder. A main strength of the SIDP is the sophistication of its design (Rogers 2001). Unlike other semi-structured personality assessment interviews where symptoms for each DSM-IV character pathology are typically clustered and questioned together (e.g., SCID-II; First et al. 1997), the SIDP diagnostic questions are arranged according to 10 topical facets of a person’s life, e.g., interests, relationships, emotions. The SIDP’s 160 clinical inquiries/ratings proceed from rather benign, non-threatening topics, such as usual activities or interests, to more personally loaded questions, such as those that assess disordered thinking and unlawful or self-injurious behavior. This sophisticated organization fosters a natural progression of inquiry that is well suited to building rapport over the course of the interview, enabling research participants to feel less threatened and be more open in their responses to symptom queries. Thus, transparency is reduced and interviewees are less able to blatantly distort their responses (Jane et al. 2006; Rogers 2001). Moreover, scoring of the SIDP is not unidimensional; unlike the SCID-II, an affirmative answer does not always reflect psychopathology, which further reduces transparency. Additional advantages of the SIDP are its high inter-rater reliability (Jane et al. 2006) and improved diagnostic clarity over other axis II interviews in that discrete (“pure”) disorders are assessed reliably and systematically (Rogers 2001).

The present study identified and compared DSM-IV personality disorders and traits (defined here as pathological behaviors or feelings representative of typical functioning over the past 5 years and not limited to the luteal phase of the menstrual cycle) using the SIDP, in demographically matched samples of women with and without severe PMS. We hypothesized that women with severe PMS would have significantly more personality disorder traits, particularly those involving affective lability, i.e. borderline/histrionic/dramatic (cluster B) and anxious/fearful (cluster C) traits, and have a greater prevalence of personality disorders, than healthy control women. We further hypothesized that avoidant and obsessive–compulsive personality disorders would be the disorders most represented in the PMS group.

Methods

Participants

Potential participants were recruited from communities in the greater San Francisco Bay area through flyers, announcements, or word of mouth from April 2008 to June 2009. About 20% of potential PMS participants were recruited from community outpatient clinics after referral from their physician. A total of 187 women (129 who considered themselves as having severe PMS and 58 who considered themselves as having minimal premenstrual symptoms) underwent an initial telephone screen. This screen included the Premenstrual Symptoms Screening Tool (PSST; Steiner et al. 2003), a self-report tool that assessed severity and impairment of the 11 DSM-IV PMDD symptoms. Women who qualified for a diagnosis of severe PMS (including PMDD) according to the PSST were considered eligible for the severe PMS group. Women who reported mild or no symptoms were considered eligible for the control group. Additional inclusion criteria for both groups were regular menstrual cycles and sleep–wake schedules, good health, and no regular medication over the previous 3 months. Psychiatric exclusion criteria for the PMS group were a history of schizophrenia, eating disorder, or bipolar disorder; substance use disorder within the previous 12 months; or any current axis I psychopathology. Potential controls were excluded if they had any past or current axis I psychopathology.

Following the telephone screen, 54 women eligible for the severe PMS group and 31 eligible for the control group were invited to the laboratory to provide informed consent and undergo clinical interviews. Of these 85 screened women, 14 PMS and 1 control declined to participate. Another seven potential PMS women were excluded during the SCID for premenstrual symptoms not severe enough to interfere with functioning (n=3), starting medications for PMS (n=3), or diagnosis of past eating disorder (n=1). Four potential controls were excluded for past eating disorder (n=1), past major depressive disorder (n=2), or BMI >30 kg/m² (n=1). The characteristics of the 33 women with severe PMS and 26 controls included in the study are shown in Table 1.

Table 1.

Demographic and menstrual characteristics of the participant groups

Descriptor	Severe PMS (n=33)		Control (n=26)		p value
Ethnicity (Caucasian/non-Caucasian)	19/14		14/12		0.76
	Mean	SD	Mean	SD
Age (years)	30.6	6.9	28.9	7.2	0.36
Years of education	16.3	2.0	16.3	1.9	0.95
Body mass index (kg/m2)	23.3	3.7	22.3	2.6	0.21
Age at menarche (years)	12.5	1.6	12.5	1.3	0.83
Menstrual cycle length (days)	29.0	2.5	28.2	1.8	0.17
Length of menses (days)	5.0	1.3	4.8	1.0	0.33
GAF rating	76.6	10.9	87.4	3.4	<0.001

A lower score in GAF reflects poorer functioning (rating reflects current functioning) GAF global assessment of functioning

Procedure

The study was reviewed and approved by SRI International’s institutional review board. All participants provided written informed consent and received a modest stipend for study participation.

Participants were administered the Structured Clinical Interview for DSM-IV-TR Axis I Disorders (First et al. 2002) and the Structured Interview for DSM-IV Personality Disorders (Pfohl et al. 1995) during a 3-h assessment session on one occasion. Of all participants, 15 PMS and 13 control women were in the follicular phase, 18 PMS and 13 control women were in the luteal phase (based on each woman’s typical cycle length, first day of last period, and first day of subsequent period) at the time of their interview, [χ²(1)= 0.12, N=59, p=0.80].

Assessment of severe PMS

In addition to the standard modules of the SCID, participants were questioned about their premenstrual symptoms in a customized module based on the DSM-IV-TR criteria for a PMDD diagnosis. Based on the SCID assessment, all women in the severe PMS group met criteria for diagnosis of current “depressive disorder, not otherwise specified (premenstrual dysphoric disorder, provisional)” by endorsing at least 5 of 11 PMDD symptoms limited to the luteal phase of the menstrual cycle. At least one symptom had to be markedly depressed mood, anxiety/tension, irritability, or affective lability. Additionally, report of premenstrual-related impairment in work, school, usual activities, or in personal relationships was critical to diagnosis. Healthy control women reported having only mild or no PMS symptoms on the SCID. Clinicians conducting the SCID also assigned a Global Assessment of Functioning (GAF) rating score (see DSM-IV-TR; American Psychiatric Association 2000) for each participant. GAF scores were on a scale of 100 to 0 (a hypothetical continuum of mental health-illness with lower scores reflecting poorer functioning) and were based on the participants’ overall reports on the SCID of their current psychological, social, and occupational functioning excluding impairment due to physical or environmental limitations.

After the diagnostic interviews, participants were given the Daily Symptom Rating (DSR; Freeman et al. 1996) to rate severity and duration of premenstrual symptoms over two menstrual cycles to confirm presence (for affected women) and absence (for controls) of severe premenstrual symptoms. The DSR lists 17 common PMS symptoms, including the 11 DSM-IV PMDD diagnostic criteria. Participants rated each item on a five-point scale (0=none, to 4=extreme). Severe PMS required a score of 80 or greater on the DSR in the late luteal phase and an increase of at least 50% from the follicular phase score for both screening months. DSM-IV-TR criteria for PMDD required at least five criteria as severe (3 or 4) on at least two premenstrual days, with the same symptoms being rated as absent or minimal (0 or 1) postmenstrually. In the severe PMS group, 23 out of 33 women returned their diaries, of whom 14 met PMDD criteria and 9 met severe PMS criteria based on diary ratings. In the control group, all 26 women returned their diaries, confirming no or mild premenstrual symptoms.

Assessment of axis II disorders and traits with the SIDP

Participants were administered the full SIDP, which probed for symptoms of all 10 DSM-IV-TR personality disorders. Any positive endorsement to a criterion was clarified with additional questions eliciting more information and/or examples. Each criterion was rated as follows: 0—not present, 1—subthreshold or limited evidence, 2—clearly present (i.e., at least 50% of the time, for most of the last 5 years), or 3—strongly present and associated with subjective distress or impairment in social, occupational, or intimate relationships. The SIDP employs a “5-year rule” to diagnose a personality disorder, namely, “the behavior, cognitions, and feelings that have predominated for most of the last 5 years are considered to be representative of the individual’s long-term personality functioning.” (Pfohl et al. 1997) As such, clinicians instructed participants that their responses should reflect their usual self, i.e., what they are like most of the time over the past 5 years. Symptoms were rated as present if they were characteristic of the participant more often than not.

All SCID (axis I) and SIDP (axis II) interviews were conducted by clinicians experienced in administering diagnostic interviews. Interviews were consensed by independent clinicians; discrepancies between the interviewer and reviewing clinician were resolved by a third clinician. Clinicians conducting the SIDP achieved a high degree of inter-rater reliability (κ≥0.85).

Data analysis

Initial analysis was conducted to determine whether women with prospectively confirmed PMDD (n=14), prospectively confirmed severe PMS (n=9), and women without daily ratings (n=10) differed on any characteristic. The three groups had statistically comparable demographic data, PMS-related characteristics such as self-reported severity of premenstrual symptoms, age of PMS onset, average duration of PMS symptoms within each menstrual cycle, and number of threshold DSM-IV-TR PMDD symptoms endorsed in the SCID. Also, there were no significant differences between the three groups on any personality variables. Thus, we combined women into one severe PMS group (n=33) for subsequent analysis.

Between-group analyses on continuous data were conducted with Student’s t tests, and between-group analyses on data with limited ranges or non-normal distributions were analyzed with Mann–Whitney U tests. Fisher’s exact tests were used on categorical variables in instances where the expected count of a cell in the analysis was less than five. Where significant group differences were found on main hypotheses, ANCOVA (with menstrual cycle phase at time of interview as a covariate), Fisher’s exact tests, chi-square tests, and Student’s t tests were used in follow-up analyses. Statistical analyses were conducted with PASW Statistics 18.

Results

Sociodemographic and clinical characteristics of the participant groups

The severe PMS and control groups were comparable in age, years of education, and ethnicity (Table 1). Moreover, there were no significant differences between groups in age of menarche, length of menses, menstrual cycle length, and current body mass index. Parity was similar between groups; 27% in the PMS group and 23% of controls had at least one child. A family history of severe PMS was reported by significantly more PMS (36%) than control women (12%) [χ²(1)=4.73, N=59, p<0.001], and 67% of PMS women had consulted a physician for their PMS symptoms compared with 4% of controls [χ²(1)=24.13, N=59, p<0.001]. As determined by SCID, PMS women endorsed, on average, 7.3 (±1.9) severe premenstrual symptoms (mean rank=43.00) out of 11 total symptoms compared with 0.1 (±0.4) in controls (mean rank=13.50) [Mann–Whitney U=858.0, p<0.001].

The SCID interview identified 10 women (30%) in the PMS group with a history of at least one DSM-IV-TR axis I disorder. Five had a past history of a single episode of major depressive disorder with time since remission ranging from 9 to 1,233 weeks (median=343 weeks); three had a past history of an anxiety disorder; and three had a past history of a substance use disorder. Clinician-rated global assessment of functioning scores were significantly lower in PMS (76.6±10.9) than in control women (87.4±3.4) [t(39.5)=5.38, p<0.001].

Comparison of DSM-IV-TR personality traits and disorders between women with and without severe PMS

Severe PMS women had significantly more SIDP personality disorder traits than control women in each of the three personality disorder trait clusters: cluster A (odd, eccentric) traits, cluster B (dramatic, erratic) traits, and cluster C (anxious, fearful) traits (Table 2). A higher number of personality disorder traits was associated with a poorer GAF score which indicated a poorer level of functioning (ρ=−0.50, p< 0.001).

Table 2.

SIDP personality disorder prevalence and differences between groups in number of traits endorsed within each personality cluster

Personality disorders	Severe PMS (n=33)		Control (n=26)		Fisher’s exact p value
	n (%)		n (%)
A (odd/eccentric)
None
B (dramatic/erratic):
Narcissistic	1 (3)		0 (0)
C (anxious/fearful)
Avoidant	2 (6)		0 (0)
Obsessive–compulsive	6 (18)		0 (0)		0.030
Mixed personality disorder	1 (3)		0 (0)
Optional diagnoses
Depressive	1 (3)		0 (0)
Total with disorders	9 (27)		0 (0)		0.003
Personality clusters	Severe PMS (n=33)	Control (n= 26)	Mann–Whitney U	Standardized test statistic	p value
	Mean rank for no. of traits	Mean rank for no. of traits
A (odd/eccentric)	35.5	23.0	610.0	3.03	0.002
B (dramatic/erratic)	35.2	24.0	599.5	2.78	0.005
C (anxious/fearful)	35.9	22.5	625.0	3.07	0.002

A significantly greater proportion of severe PMS women (27%) met criteria for any personality disorder diagnosis on the SIDP relative to controls (0%) [N=59, p=0.003, Fisher’s exact test] (Table 2). Two of the nine women with personality disorders in the PMS group met criteria for two separate personality disorders. Among the women with PMS, a diagnosis of personality disorder was unrelated to the menstrual cycle phase that they were in at the time of assessment [N=33, p=1.0, Fisher’s exact test]. Moreover, presence of comorbid PMS and a personality disorder was not necessarily related to history of an axis I disorder [N=33, p=1.0, Fisher’s exact test].

There was a significantly higher proportion of women with obsessive–compulsive personality disorder in the PMS group (18%) compared to the control group (0%) [N=59, p=0.030, Fisher’s exact test]. Contrary to our hypothesis, there was not a significantly greater prevalence of avoidant personality disorder in the PMS group compared with the control group; two women with severe PMS met criteria for this disorder compared to 0 control women. Reanalysis of the data including only women with prospectively confirmed PMS (n=23) in the PMS group produced similar results: women with prospectively confirmed PMS were more likely than controls to have any personality disorder [N=49, p=0.007, Fisher’s exact test] and particularly OCPD [N=49, p=0.042, Fisher’s exact test].

Factors related to obsessive–compulsive personality disorder in women with severe PMS

There were no significant differences in any demographic variable (i.e., age, years of education, ethnicity, BMI) between PMS women with and without OCPD. PMS women with comorbid OCPD were as likely to be diagnosed with any axis I disorder, including major depressive disorder, as PMS women without OCPD. There were no differences between PMS women with and without OCPD in severity of premenstrual symptoms on both the PSST questionnaire and clinical interview. Moreover, PMS women with and without OCPD were similar in menstrual-related variables, including age of menarche, menstrual cycle and menses length, seeking medical attention for menstrual symptoms, and family history of severe menstrual problems.

There was no significant difference between the mean age of onset of premenstrual symptoms between PMS women with (17.3±4.1 years) and without (20.0±8.4 years) OCPD, and no difference between the number of years from menarche to onset of premenstrual symptoms between those with (5.2±2.8 years) and those without (7.1±8.3 years) OCPD. Women with PMS and OCPD had similar premenstrual symptom profiles to those without OCPD. However, no PMS woman with OCPD reported impairment in concentration in the late luteal phase, whereas half of the PMS women without OCPD reported problems concentrating [N=30, p=0.057, Fisher’s exact test]. Women with PMS and OCPD had significantly poorer overall level of functioning, reflected in lower GAF scores (mean=68.8±9.2), than PMS women without OCPD (mean=78.8±10.3), though both groups had significantly poorer functioning than control women (mean=87.4±3.4), irrespective of what menstrual cycle phase they were in at the time of assessment [F(2, 53)=17.3, p<0.001].

Discussion

The Structured Interview for DSM-IV Personality Disorders was used to determine and compare categorical axis II traits and personality disorders in women with severe PMS and demographically matched healthy controls. We hypothesized that personality disorders, most notably obsessive-compulsive personality disorder (cf., Berlin et al. 2001, James and Pollitt 1974) and avoidant personality disorder (cf., DeRonchi et al. 2000, Pearlstein et al. 1990) would be more prevalent in women with severe PMS than in control women. Our results show that personality disorder traits of all three axis II clusters (odd/eccentric, dramatic/emotional, and anxious/fearful) are more commonly found among women with severe PMS than among healthy women with few premenstrual symptoms. Also, a significant proportion (18%) of women with severe PMS had obsessive–compulsive personality disorder, which was associated with poorer global life functioning. These findings suggest that character psychopathology is evident in a substantial proportion of women with severe PMS and may contribute to poorer overall functioning.

A main finding of this study was that women with severe PMS were more likely to be diagnosed with personality disorders in general (27%) compared with control women (0%); this prevalence was higher than has been documented in some other studies. Pearlstein et al. (1990), for example, found a 10% prevalence of personality disorders in women with LLPDD. Our finding is nearly twice the general population prevalence of 14.8% for DSM-IV personality disorders reported by Grant et al. (2004) and suggests a significant relationship between character pathology and severe premenstrual symptoms. Contrary to our prediction, we did not find a higher occurrence of avoidant personality disorder in our PMS sample. However, we did find a preponderance of obsessive–compulsive personality disorder (18% of the PMS group). Similarly, Critchlow et al. (2001), using the International Personality Disorders Examination, found that almost all PMDD women diagnosed with personality disorders had anancastic (obsessive–compulsive) personality disorder (18%). The finding of high co-occurrence of severe PMS and obsessive–compulsive personality disorder is clinically germane given the well-established association of mood and anxiety-related symptoms.

PMS women with and without OCPD did not differ on severity of premenstrual symptoms or on menstrual history-related variables. However, PMS women with OCPD tended to report less of a problem with concentration during the late luteal phase than PMS women without OCPD. One explanation is that women with PMS and OCPD already have trait difficulties with concentration (e.g., preoccupation with minor details to the detriment of the major point of a task, or sustained focus on perfectionism that interferes with task completion) such that no greater impairment is recognized during the premenstrual phase.

In the current study, the SIDP was administered in either the follicular or luteal phase of the menstrual cycle but we found no effect of menstrual cycle phase on personality disorder diagnosis. Berlin et al. (2001) reported that some personality styles are more prevalent in the symptomatic late luteal phase in women with PMS, but OCPD traits were stable across menstrual cycle phase. Parry et al. (1996) found no variation of personality traits across menstrual phase. Our finding suggests that increased OCPD is a trait rather than a state-specific phenomenon. Since menstrual phase was estimated based on menstrual cycle characteristics and since repeated measurements of personality were not made at different phases of the menstrual cycle in study participants, further studies are needed to confirm these findings.

Not only did we find that global functioning is worse in women with severe PMS compared to controls, we also found that women with comorbid PMS and OCPD had lower global functioning than those with PMS alone, supporting research findings that those with personality disorders experience greater functional impairment than those without (Nakao et al. 1992). Our finding emphasizes the potential impact that character pathology has on role impairment and general life functioning above and beyond that associated with severe premenstrual symptoms alone. A recent study found that even after controlling for chronic medical disorders in treatment-seeking participants, obsessive–compulsive personality disorder was one of two personality disorders (borderline personality disorder was the other) uniquely associated with increased mean total costs to society, with the most relevant cost drivers being absences from work, inpatient health care, and outpatient mental health care (Soeteman et al. 2008). Thus, given these costly consequences, our findings point to the importance of elucidating whether there is differential treatment effectiveness, e.g., with medication or therapy, between PMS women with or without comorbid OCPD.

Results supported our hypothesis that on the SIDP, women with severe PMS would endorse more cluster B (dramatic) traits, and more cluster C (anxious/fearful) traits than controls. However, we found that PMS women had more cluster A (odd/eccentric) personality traits as well. An earlier study also found that LLPDD women had a higher crude odds ratio for schizotypal personality disorder than asymptomatic women although presence of schizotypal personality disorder in this study could not significantly predict who had LLPDD or not (DeRonchi et al. 2000). Moreover, evidence suggestive of a higher possibility of Cluster A personality disorders in women with PMS was reported by Berlin et al. (2001), although results were limited to the late luteal phase. Our finding suggests that women with PMS have several personality disorder traits that are not specific to one cluster. Similarly, others have reported several personality disorder traits being more common in women with PMS than asymptomatic women (Berlin et al. 2001) even though the prevalence of absolute personality disorders may be low (Pearlstein et al. 1990).

The mixed composition of the PMS group in this study may potentially affect generalizability of the findings. While all PMS women in our sample met DSM-IV-TR criteria for provisional PMDD, 10 of these women (30%) did not complete prospective diary ratings of symptoms, 14 women (42%) had confirmed PMDD, and the remaining 9 (27%) had confirmed severe PMS. Given that prospective ratings of PMS symptoms are the gold standard for diagnosis and that retrospective and prospective ratings of PMS symptoms can be discordant (Halbreich and Endicott 1985), confirmation of the full PMS sample would have been optimal for diagnostic validation. However, a comparison of the prospectively rated PMS, PMDD groups, and those who did not submit diaries revealed no differences in either background demographics, menstrual-related variables, or in likelihood of personality traits or diagnoses. All of the women reported significant premenstrual symptoms that interfered with their functioning, making their symptoms clinically relevant, and it is important to note that women with prospectively confirmed severe PMS missed the diagnostic threshold for PMDD often by just one or two symptoms on a diary rating. Our findings extend others showing that personality disorders are evident in women with significant premenstrual symptoms. Limitations of this study include small samples and inability to draw conclusions about the causal relationship between severe PMS and OCPD. Further, it is unknown if personality pathology is a factor in the onset of PMS or the reverse. In this study, PMS reportedly started several years after menarche but around the time in young adulthood when personality disorders typically become recognizable. Thus, future investigations with larger sample sizes may seek to elucidate factors leading to etiology of both disease pathologies, perhaps by prospective longitudinal rather than cross-sectional retrospective design.

In conclusion, our findings that women with severe PMS had a greater endorsement of personality disorder traits in all three clusters evaluated in the SIDP suggest that pathological character traits in general may be associated with severe PMS. Moreover, an axis II diagnosis of obsessive–compulsive personality disorder on the SIDP was found to be most prevalent in severe PMS. OCPD, while not necessarily associated with greater severity of premenstrual symptoms, was related to poorer life functioning in women with severe PMS, beyond the level of functional impairment associated with PMS alone. This observation highlights the impact of the burden of the comorbidity of these diseases on general life functioning and may have implications for treatment or medication given to those with severe premenstrual symptoms.