Background

Premenstrual syndrome (PMS) and the more severe premenstrual dysphoric disorder (PMDD) are debilitating illnesses of women in their reproductive years (1, 2) and have even been associated with suicidal behavior (3). Within the United States, studies show that 15% to 40% (1, 4, 5, 6) of women experience premenstrual syndrome. Considerable research shows that the severity of PMS symptoms is significantly impacted by dietary factors (7, 8, 9, 10, 11, 12, 13).

A potentially important dietary factor in PMS is caffeine. This powerful stimulant is the most widely consumed drug in the world (14), and its arousal properties are well-documented (15, 16), as is its toxicity (17). Moreover, caffeine has been reported to exacerbate PMS (7, 18, 19, 20, 21, 22). The authors of a more recent retrospective study also concluded that caffeine may exacerbate PMS symptoms, although they note that caffeine accounted for only a small proportion of the variance (23). Based primarily on the Rossignol studies, caffeine restriction is frequently recommended (24).

These studies, which employed self-administered questionnaires, showed that the risk of PMS was significantly elevated in college women who consumed more than four caffeinated drinks per day and increased further in those ingesting 8-10 drinks (19, 21, 22).

While the Rossignol studies suggested that caffeine intake is associated with PMS, Caan and colleagues failed to support this association (26). They diagnosed PMS retrospectively and then collected caffeine intake data for three consecutive days during both premenstrual and postmenstrual weeks via telephone interviews. The authors suggested that women with PMS were less likely to consume caffeinated coffee or tea, but were more likely to drink decaffeinated beverages (26).

As Halbreich and colleagues have recently pointed out, studies of PMS often suffer from methodological limitations (27), and those mentioned above are no exception to this observation. First, none of these studies diagnosed PMS according to the NIMH Workshop diagnostic guidelines (28). Specifically, data must be gathered for two complete cycles, in each of which must occur a 30% increase in symptoms during the luteal, as compared with the follicular, phase (29). Second, Rossiginol’s studies were cross-sectional and retrospective. While Caan and colleagues addressed the latter concern, their study was also cross-sectional. Third, previous studies did not report any screening of their participants for preexisting psychiatric conditions. And finally, the basis for determining the caffeine content of foods and beverages was not specified.

In order to address the methodological limitations seen in earlier work, the present study first implemented the NIMH diagnostic guidelines. Our diagnostic procedures were also consistent with those proposed by the recent consensus conference for inclusion in the next edition of the International Classification of Diseases (ICD-11) of the World Health Organization (30). Participants were then diagnosed with PMS only if they reported at least a 30% increase on at least one of either physical or emotional symptom scores 5 days before the onset of menses and symptom remission within 4 days of onset of menses. The Patient Health Questionnaire (PHQ, 31) was used to screen for current and previous psychiatric conditions, and participants who reported any psychiatric diagnoses within the past six months were excluded. Finally, caffeine values were adopted from Juliano & Griffiths (2005) and were consistent with those reported elsewhere (32).

The present study is the first to address the diagnostic limitations of prior studies examining the relationship between PMS and caffeine intake. Based on prior literature indicating a positive association between caffeine consumption and experience of PMS symptoms, we hypothesized that women with PMS would consume more caffeine than would controls across the entire menstrual cycle.

Subjects And Methods

The current project was reviewed and approved by the Institutional Review Board at the University of Maryland College Park (IRB #: 04-0583) for recruitment of undergraduates at the institution. Participants were primarily recruited through advertisements in psychology courses, and 110 female undergraduate students initially met with the researcher individually or in small groups to sign informed consent prior to completing online screening questionnaires. Participants who indicated caffeine sensitivity, pregnancy, a diagnosed psychiatric condition within the past six months, or a history of drug abuse were dropped form the study (33, 34). Psychiatric history was assessed using the PHQ. A total of seven participants endorsed suicidal ideation (n = 3), previous or current diagnosis of depression (n = 3), or caffeine sensitivity (n = 1), leaving the sample of 103. Attrition rate was approximately 18% (n = 20). The final sample (N = 83) ranged in age from 18 to 26 (M = 19, SD = 1.98).

Screening Instruments

Daily Dairies

Procedures

The present study included two phases, Electronic Screening and Daily Diaries. Participants were required to attend an information session to receive instructions on how to proceed and also to sign informed consents. The screening session was online and available at a website provided on the Information Sheet for Participants. The screening session included a caffeine consumption survey, a demographic questionnaire, and the entire Patient Health Questionnaire (PHQ). Participants who qualified to continue the second phase of the study (i.e. did not meet any exclusion criteria), completed the daily diaries described above. In order to control for possible time-of-day-effects, daily entries were undertaken at the same time every night for two consecutive menstrual cycles, and the experimenter was able to determine for each entry the exact date and time. Data obtained from participants who missed more than five entries in either month were excluded from the final analysis. Participants who missed one day of entry were allowed to complete the entry the following day by submitting two entries at the same time. Data submitted more than 24 hours late were excluded from analysis.

Results

Statistical Procedures

Demographic Variables

Caffeine Consumption and PMS

Discussion

Physicians frequently recommend that PMS patients restrict their intake of caffeine, which has been thought to potentially exacerbate symptoms (24, 25). The principal evidence supporting this assumption comes from a series of investigations conducted by Rossignol and colleagues (19-22). However, as detailed above, significant methodological questions have been raised concerning these studies. In particular, these cross-sectional studies (a) used retrospective self-reports rather than prospective dairies for diagnostic purposes (e.g. 1983 NIMH Workshop guidelines on DSM criteria for PMDD (40); American Psychological Association, 2000 (41), (b) failed to screen for psychiatric conditions, and (c) did not report the basis for determining the caffeine content of foods and beverages (42).

Two previous studies provide results contrary to those of the Rossignol group. In the first of these, PMS was diagnosed on a retrospective basis, and caffeine consumption was based on three days of daily caffeine intake data. Results showed no evidence that caffeine contributes to PMS symptoms (26). In the second, Gold and colleagues (43) employed retrospective, questionnaire data for both PMS and caffeine consumption. They also reported that caffeine had no significant effect on PMS symptomatology. Unfortunately, both studies also suffered from some of the same methodological concerns associated with the Rossignol studies.

The present study employed a design and methodology that addressed the issues raised concerning the prior work. In particular, this study used a prospective approach both for PMS diagnosis and for the assessment of caffeine intake. Specifically, data on both PMS symptoms and caffeine consumption were based on daily monitoring over the course of two complete menstrual cycles. In addition, psychiatric history was screened, and caffeine content was based on guidelines derived from prior research (38). This methodology provided for both a more accurate diagnosis of PMS and a more accurate assessment of caffeine consumption.

There were two principal findings in the present study. First, that caffeine intake is not associated with PMS symptoms (e.g. women with PMS were not more likely to consume caffeine than non-PMS sufferers). As noted above, this finding is consistent with those of two retrospective studies (26, 43). Present findings suggest that it is unlikely that caffeine consumption exacerbates PMS symptoms. The second major finding in the present study was that women, across groups, consumed more caffeine during the follicular than during the luteal phase. Thus the menstrual cycle phase difference in level of consumption appeared independent of diagnosis. In interpreting the findings of the present study, it is important to note that the college student sample, although they met inclusion criteria, may not consume as much caffeine as do older populations (44) and it was also unclear whether any participant(s) restricted caffeine use during the course of this study. A final consideration is that the PMS diagnostic criterion used here was a 30% increase in symptomatology, as recommended by NIMH guidelines, and thus level of impairment could not be determined (45). Nonetheless, self-ratings of impairment have been shown to be highly correlated with symptom self-ratings in women with PMS (32).

Conclusion

Media and internet sites often suggest that caffeine exacerbates PMS symptoms (25, 46-48). The current data does not provide evidence to support these recommendations. Data suggests that caffeine intake decreased during the luteal phase across women with and without PMS. The present study has demonstrated the value of longitudinal methods of data collection, in order to facilitate our understanding of how the menstrual cycle interacts with caffeine consumption, and it recommended that similar methods be applied to future research. In addition, future research might address the effects of caffeine withdrawal symptoms in relation to PMS, since there are some overlapping symptoms (49). Finally, further research should be more clearly guided by knowledge of underlying mechanisms and by theoretical models of caffeine-adenosine effects that integrate those mechanisms.