Seventeen women in their 20s volunteered to participate in this study. The women, all college students, responded to a campus advertisement. The study protocol was approved in advance by the Institutional Review Board of Shitennoji University and was performed in accordance with the Declaration of Helsinki of the World Medical Association. All subjects received an explanation of the nature and purpose of the study: to investigate soothing effects of plant fragrance on emotional symptoms in the premenstrual phase. We did not, however, inform subjects of which fragrance we would use for the experiments. Prior to receiving any data about the experiments, all subjects gave their written informed consent to participate in the study.
The subjects underwent medical examinations and interviews and completed a standardized health questionnaire regarding medical history, medications, current health condition, regularity of menstrual cycle, premenstrual discomfort, and lifestyle. While referring to subjects’ self-reported regular menstrual cycles, we determined the cycle phase during the experiments by the onset of menstruation and oral temperature verified by concentrations of ovarian hormones, estrone (E1), and pregnanediol-3-glucuronide (PdG), in a urine sample taken early in the morning. Both E1 and PdG were indexed to creatinine (Cr) excretion [21–24].
As to premenstrual discomfort, responses of participants to the health questionnaire indicated that all women had some subjective psychophysiological complaints of varying degrees, from mild to moderate. However, none of the women reported that premenstrual symptoms markedly interfered with work, school, usual activities or relationships with others. To determine the severity of premenstrual symptoms, we asked subjects to record a daily symptom diary based on the menstrual distress questionnaire  for at least two menstrual cycles. The average value of the increase on the total scores of the diary from the follicular (day 5 to day 11 from the first day of menstruation) to the late-luteal phase (within seven days before the next menstruation) was 10.3 ± 2.4%. None of the subjects reported greater than a 30% increase, a standard for diagnosing PMS [3, 26]. Notably, according to a series of studies by the authors of this paper [21–24], a less than 20% increase of subjective symptoms from the follicular to the late-luteal phase did not influence autonomic nervous system activity during the menstrual cycle. Considering this information, no subjects in the present study suffered from severe PMS or premenstrual dysphoric disorder (PMDD), but the subjects’ symptoms did fall within the sphere of premenstrual molimina (subclinical levels of premenstrual symptomatology), signaling impending normal menstruation, which a majority of reproductive-age women experience [3, 26].
None of the subjects had been clinically diagnosed with diabetes mellitus, hypertension, hyperlipidemia, or cardiovascular or any other endocrine or systemic disorders that could affect the autonomic nervous system. The subjects were non-obese and non-smokers. None of the women reported taking oral contraceptives to control the menstrual cycle. We did not perform a pregnancy test for this study; however, regular menstrual cycles subsequently resumed in all subjects after the completion of the study. Thus, results showed that none of the subjects were pregnant during the study period.
Referring to a study of Kiecolt-Glaser et al. , we performed the olfactory function test on all subjects to assure that none had anosmia. Briefly, subjects were given two sets of three bottles—two held distilled water; the third contained essential oils (lavender or orange)—and were asked to choose the one that differed from the other two. To be eligible for the study, subjects had to choose the correct response in both trials.
All subjects were asked not to consume any food or beverages containing alcohol or caffeine after 21:00 of the day preceding the experiment. The subjects were also instructed to abstain from alcohol use and excessive physical activity for 24 hours before testing .
All subjects were examined on two separate occasions (aroma and control trials) in the late-luteal phase (within seven days before the next menstruation). The order of testing was randomized so that equal numbers of subjects were studied first in each trial. All measurements were taken between 11:00 and 15:00 and were performed in a temperature-controlled (25°C), quiet, comfortable room with a minimization of arousal stimuli. Height and body weight of each subject were measured to calculate body mass index (BMI) as body weight divided by height squared. Subjects then rested for at least 10 minutes before the start of the experiment.
This experiment used two kinds of aroma stimulation: lavender (Lavandula angustifolia, Lot No. BLAH10, Kensoigakusha Co. Tokyo, Japan) and water as a control. Major components of the lavender oil used in this study comprised linalyl acetate (37.18%), linalool (36.83%), trans-β-ocimene (4.25%), β-caryophyllene (3.55%), cis-β-ocimene (2.88%), and lavendulyl acetate (2.06%). Referring to previous studies [28, 29], we pipetted 10 μl of lavender essential oil or water onto a small cotton pad designed for a diffuser (Aroma breeze NOVA T, ALTA Corporation, Nagoya, Japan). Airflow from the diffuser was set at 1.3 m per min and placed near the subject’s nostrils using the diffuser’s 30 cm long circular cylinder fitted with a perforated funnel (diameter 5 cm).
Before measurements were taken, the subjects were instructed to relax quietly and comfortably for at least 10 min in a seated position while equipped with electrocardiograph (ECG) electrodes. They then filled out the Profile of Mood States (POMS) explained in detail below. The ECG was recorded 5 min before inhalation of the scent. Each subject inhaled the scent for 10 min. We then measured the ECG for 5 min at 0, 10, 20, and 30 min after inhalation. During ECG recording, all subjects breathed in synchrony to a metronome at 15 beats per minute to ensure that the respiratory-linked variations in heart rate did not overlap with low-frequency heart-rate fluctuations (below 0.15 Hz) from other sources [21, 23]. After the ECG was recorded, the subjects repeated the POMS test. The ECG signals were later analyzed by means of HRV power spectral analysis, as described below, to evaluate whether aroma stimulation changed autonomic nervous system activity.
R-R interval power spectral analysis procedure
The autonomic nervous system activity was noninvasively measured by HRV power spectral analysis, which decomposes the series of sequential R-R intervals into a sum of sinusoidal functions of different amplitudes and frequencies by the Fourier transform algorithm. The technique of the analysis for the present investigation has been applied in basic physiological and clinical research fields, and its validity and reliability has been previously confirmed [21, 23, 30–32]. Researchers have elsewhere described the procedure of R-R interval power spectral analysis used in the present study in great detail [30, 31]. Briefly, the ECG signal was amplified (MEG-6108, Nihon Kohden, Tokyo, Japan) and digitized via a 16-bit analog-to-digital converter (Model PS-2032GP, TEAC, Tokyo, Japan) at a sampling rate of 1000 Hz. The digitized ECG signal was differentiated, and the resultant QRS spikes and the intervals of the impulses (R-R intervals) were stored sequentially on a hard disk for later analyses.
Before the R-R spectral analysis was performed, the stored R-R interval data were displayed and aligned sequentially to obtain equally spaced samples with an effective sampling frequency of 2 Hz and displayed on a computer screen for visual inspection. Then, the direct current component and linear trend were completely eliminated by digital filtering for the band-pass between 0.03 and 0.5 Hz. The root mean square value of the R-R interval was calculated as representing the average amplitude. After passing through the Hamming window, power spectral analysis by means of a fast Fourier transform was performed on a consecutive 256-sec time series of R-R interval data obtained during the test. Spectral powers were calculated for the following respective frequency band: low frequency (LF) power (0.03–0.15 Hz), an indicator of both sympathetic and parasympathetic nervous system activity; high frequency (HF) power (0.15–0.5 Hz), which solely reflects parasympathetic nerve activity; and Total power (0.03–0.5 Hz) representing overall autonomic nervous system activity.
Basal heart rates and autonomic nervous system activities differ from individual to individual. Thus, the mean values for heart rates before inhalation of scent were set as the baseline values and the mean values for autonomic nervous system activity before the inhalation were standardized to 100%. The rate of change after the inhalation was compared between aroma and control trials .
Assessment of emotional symptoms
We administered the Japanese version of the POMS test (Kaneko Shobo Co., Tokyo, Japan), a globally standardized, self-administered, 65-item questionnaire (including 7 dummy items) to assess premenstrual mood states before and after inhalation of the lavender aroma and water. Each item was rated on a five-point Likert-type scale of zero to four, ranging from “not at all” to “extremely.” We added these raw scores to generate six subscales of emotional state: tension–anxiety, depression–dejection, anger–hostility, vigor, fatigue, and confusion. These added raw scores were then converted into T-scores according to the POMS manual . We should mention that, according to our recent study , five negative POMS variables—tension–anxiety, depression–dejection, anger–hostility, fatigue, and confusion—reflect the cluster of premenstrual psychoemotional symptoms. In addition, those variables significantly and positively correlated in the late-luteal phase. Referring to a study by Kuroda et al. , to investigate the effect of lavender aroma on mood states, we compared changes in the POMS scores of the lavender and control trials before and after the ECG measurements.
To investigate the acute influence of inhalation of the lavender aroma on HRV spectral power, the effects of aroma and time and their interaction (aroma x time) were evaluated using two-way ANOVA with repeated measures. When significant interactions were found, we conducted paired t tests between lavender and control trials and one-way ANOVA with repeated measures during each trial. When Mauchly’s test of sphericity showed significance, probability values were adjusted using the Huynh-Feldt correction. Paired t tests were performed to compare changes in scores of the POMS test before and after the ECG measurements between aroma and control trials. Values are reported as mean ± standard errors (SE). P values < 0.05 were considered statistically significant. All statistical analysis was performed using a commercial software package (IBM SPSS Statistics Version 20).