Selection of studies
The search strategy identified 542 references. After removing duplicates, the titles and abstracts of 314 articles were screened for inclusion. Forty-one potentially relevant papers were identified for full-text evaluation, but 31 papers were subsequently excluded (see Fig. 1). Ten studies met inclusion criteria and were included in this review.
Study and participant characteristics
Included articles were published between 2002 and 2018 and originated from nine different countries. No papers pertaining to LLPDD or “severe” or “extreme” PMS and suicidality were identified; thus, this review pertains solely to the current classification of PMDD. Of the ten papers included, three studies utilized a case control design, four were epidemiological cohort studies, and three were cross-sectional studies of identified patient groups (see Table 3).
Participants comprised of 10,951 women aged between 13 and 53 years, including 281 control participants (2.6%). From these women, 725 (6.6%) were identified to have PMDD (see Table 3); however, Yonkers et al. (2003) did not report the prevalence data for PMDD within their sample.
QATSDD scores for the included articles ranged between 38.5 and 79.5%, with a mean average of 60.3%. Yonkers et al. (2003) obtained the lowest score, due to limited methodological descriptions and a lack of clarity regarding several key criteria. In total, one paper was rated as having “excellent” quality, seven as “good” and two as “moderate”. As this is the first review to evaluate suicidality in women with PMDD, all studies were included in the final synthesis in order to provide a comprehensive overview of the available research.
Nine studies reported that participants were assessed for PMDD based upon the diagnostic criteria outlined in the DSM, using variable methods (see Table 3 for an overview). Yonkers et al. (2003) reported using a questionnaire that was designed specifically for their study. However, as prospective charting of two symptomatic cycles is required for full DSM diagnosis, PMDD diagnoses should be considered provisional for all studies included in this review.
Assessment of suicidality
Suicidality was assessed using structured diagnostic interviews (n = 5), standardized questionnaires (n = 1) and statements about suicide for which affirmative responses were scored (n = 1). In addition, three studies recruited women from a hospital following a suicide attempt (see Table 3).
Summary of suicidality findings
Findings were grouped according to the type of suicidality being reported, including suicide cognitions (n = 5) and suicide attempts (n = 7). Of the ten studies included in this review, two studies reported data pertaining to both cognitions and attempts and therefore featured in both groupings. Supplementary data relating to PMS comparison groups, suicide risk profiles (in terms of frequency, impulsivity and severity), the influence of Axis I disorders and menstrual cycle phase were also summarized.
In 2018 de Carvalho et al. reported findings from 727 Brazilian women, identifying that women with PMDD (n = 128) were two or three times more likely to report “current suicide risk” (p.45), compared with women without PMDD. Whilst the authors did not further specify the suicide features that constituted “risk”, several studies have reported data showing that women with PMDD described increased suicide cognitions, including suicidal thoughts, ideation and plans. First, Yonkers et al. (2003) assessed 907 women in Connecticut, USA. The researchers found that 24% of women with PMDD endorsed suicidal thoughts at any level (several days, more than half of the days or every day) and that 20% of these women endorsed suicidal thoughts for several days.
Data pertaining to suicidal ideation and plans were reported by three further studies. Both Hong et al. (2012) and Pilver et al. (2013) analysed data from large epidemiological studies based in South Korea and the USA, respectively. Together, these studies included data from 6464 women aged between 18 and 49 years, from which 227 women (3.5%) were identified to experience PMDD. The frequency of suicidal ideation and plans was analysed among PMDD and non-PMDD cases, and data were reported for both 12-month prevalence (Hong et al. 2012) and lifetime prevalence (Hong et al. 2012; Pilver et al. 2013).
In both studies, the researchers identified that women with PMDD were significantly more likely to report suicidal ideation and plans, compared with women without PMDD. Prevalence rates for lifetime suicidal ideation in women with PMDD were 45.8% and 37.4%, compared with 17.3% and 13.3% for women without PMDD, respectively (Hong et al. 2012; Pilver et al. 2013). Likewise, prevalence rates for suicide plans in women with PMDD were 16.9% and 19.1%, compared with 4.2% and 4.6% for women without PMDD, respectively (Hong et al. 2012; Pilver et al. 2013). In addition to a significant lifetime prevalence, Hong et al. (2012) also identified a significant 12-month prevalence of both suicidal ideation and plans, findings that were independent of social desirability and demographic covariates.
Furthermore, Ogebe et al. (2011) reported findings for adolescents aged between 13 and 21 years. In this study, the researchers identified that 18% of adolescents with PMDD reported suicidal ideation, in comparison with 10.9% of adolescents with no or mild PMS symptoms. In sum, women with PMDD are significantly more likely than women without PMDD to endorse suicide cognitions and therefore present an increased risk for suicidality. Importantly, these findings were not limited by age, social desirability or demographic covariates and were reported across a mixture of America, African and Asian cultures.
Data pertaining to suicide attempts were reported in seven studies and were further defined into self-report data (n = 4) and hospital admissions data (n = 3).
Wittchen et al. (2002) explored the prevalence and co-morbidities of PMDD in a community sample of 1488 young women, living in Munich, Germany. During their analysis, the authors identified a “remarkably high risk” (p.128) of suicide attempts among PMDD cases (n = 112), with 15.8% reporting at least one suicide attempt during their lifetime, compared with a frequency of 3.2% for women without PMDD. Since these findings, three further studies have reported similar observations.
In their epidemiological studies, Hong et al. (2012) and Pilver et al. (2013) also assessed participants for a history of suicide attempts and identified a significantly higher prevalence of previous suicide attempts of 13.6% and 16.2% respectively in women with PMDD, compared with women with no premenstrual symptoms (Hong et al. 2012: 3.9%; Pilver et al. 2013: 4.9%). Soydas et al. (2014) compared 70 Turkish women with PMDD to a healthy control group and identified a significantly higher frequency of historical suicide attempts in the PMDD group (7.1%) compared with the control group (0%). Thus, findings consistently show that women with PMDD are significantly more likely than women without PMDD to self-report previous suicide attempts.
Hospital admissions data
Three studies assessed the prevalence of PMDD among women who had been admitted to the hospital following a suicide attempt. In the first study, Baca-Garcia et al. (2004) assessed 125 women who were admitted to the emergency room of a Spanish general hospital following a suicide attempt and compared them with 83 female blood donor controls: 54% of the women who had attempted suicide met the diagnostic criteria for PMDD (67/125), in contrast to just 6% (5/83) of the control group. Ducasse et al. (2016) collected data from 232 women consecutively hospitalized in a French psychiatric unit following a suicide attempt and found that 23% retrospectively met the diagnostic criteria for PMDD (51/232). Finally, Shams-Alizadeh et al. (2018) studied 120 women admitted to a general hospital in Iran following a suicide attempt compared with a matched control group. In accordance with the previous findings, the researchers identified that PMDD was significantly more frequent in women who had attempted suicide (30.8%; 37/120), compared with the control group (5%; 6/120).
In summary, women with PMDD are significantly more likely to self-report previous suicide attempts, compared with women without PMDD. In addition, hospital admission studies show that a remarkably high proportion of female suicide attempters retrospectively met the criteria for PMDD.
Comparison with PMS
Pilver et al. (2013) identified that women with PMDD were significantly more likely to report suicidal ideation (37.4%), plans (19.1%) and attempts (16.2%) compared with women with PMS (22%, 7.6% and 7.4%, respectively). Likewise, Shams-Alizadeh et al. (2018) found that PMDD was significantly more frequent in a group of female suicide attempters compared with a control group, whilst no differences were identified in the prevalence of PMS between the two groups. However, Ogebe et al. (2011) identified that 24% of adolescents with PMS reported suicidal ideation, compared with just 18% of adolescents with PMDD. Thus, whilst there is some evidence to support an increased risk profile for women with PMDD, findings to date are contradictory and require further research.
Suicide attempt risk profiles
Data pertaining to the frequency, impulsivity and lethality of suicide attempts were reported by both Baca-Garcia et al. (2004) and Ducasse et al. (2016). Their findings showed that women with PMDD who had attempted suicide did not exhibit a more severe risk profile compared with suicide attempters without PMDD. Although Baca-Garcia et al. (2004) identified that 59% of women with PMDD had made a previous suicide attempt and that 32% had made more than two previous attempts, these findings were not significantly different to the frequencies reported by non-PMDD cases. Likewise, no significant differences were identified regarding the impulsivity of suicide attempts between women with and without PMDD in either study, or regarding the lethality of suicide attempts.
Influence of Axis I disorders
Psychiatric co-morbidities are commonly experienced by women with PMDD; therefore, it is important to establish whether the relationship identified between PMDD and suicidality exists independently. Several studies included in this review reported data that controlled for the presence of co-morbid Axis I psychiatric disorders, with findings showing that the relationships between PMDD and suicidality largely remained significant after these adjustments were made. Specifically, Pilver et al. (2013) identified that the strong associations between PMDD and suicidal ideation, plans and attempts were independent of psychiatric co-morbidity, Wittchen et al. (2002) identified that suicide attempts in women with PMDD remained significant after controlling for major depressive disorders, and Hong et al. (2012) identified that both lifetime and 12-month suicidal ideation remained significant, when adjustments for psychiatric disorders were made, although significant associations with suicide plans and attempts disappeared.
Relationship to menstrual cycle phase
With PMDD symptoms commencing during the week(s) prior to the onset of menses, it would be anticipated that there would be an increased frequency of suicide attempts during the luteal menstrual phase for women with PMDD. However, Baca-Garcia et al. (2004), Ducasse et al. (2016) and Shams-Alizadeh et al. (2018) all reported finding no differences in cycle phase between suicide attempters with and without a retrospective diagnosis of PMDD. Thus, despite the significant associations identified between PMDD and suicidality, women with PMDD did not seem to be more likely to attempt suicide during the luteal menstrual phase, when symptoms are expected to reach peak severity.