In this cross-sectional study, we retrospectively analyzed the first-visit records of 471 Japanese peri- and postmenopausal women aged 40–65 years who had been enrolled in the Systematic Health and Nutrition Education Program at the menopause clinic of the Tokyo Medical and Dental University Hospital from November 2007 to June 2016. All women enrolled in this program were referred to our clinic for the treatment of menopausal symptoms. Prior to beginning the study, the research protocol was approved by the Tokyo Medical and Dental University’s Review Board (approval number 774), and informed consent was obtained from all participants. All study procedures were performed in accordance with the Declaration of Helsinki.
Women were defined as premenopausal if they had experienced regular menstrual cycles in the past 3 months, as perimenopausal if they had had a menstrual period within the past 12 months but had a missed period or had an irregular cycle in the past 3 months, as postmenopausal if they had no menstrual period in the past 12 months, and as surgically induced menopause if they had had a hysterectomy.
Data regarding the menopausal symptoms and quality of life during the past month were collected using the Menopausal Health-Related Quality of Life Questionnaire (MHR-QOL) and the Hospital Anxiety and Depression Scale (HADS). These data were collected by the physicians and nutritionists who interviewed the women during their initial visit.
The MHR-QOL, which was developed and validated in our clinic [6,7,8,9,10,11,12,13,14,15,16,17,18], is a modification of the Women’s Health Questionnaire [3, 19]. It contains 38 items, which are scored on a four-point or binary scale and cover four major domains of health during menopause (physical health, mental health, life satisfaction, and social involvement). The item scores were used to indicate the symptom frequency (0–1 times per month = 0; 1–2 times per week = 1; 3–4 times per week = 2; almost every day = 3). The physical health domain consists of nine items that assess vasomotor, somatic, and urinary symptoms. The average of the two vasomotor items (“hot flashes” and “night sweats”) was defined as the vasomotor symptom score. The mental health domain consists of 12 items that assess depressed mood, cognitive difficulty, anxiety and fear, sexual function, sleep problems, and low self-esteem. The average of the two sleep problem items (“difficulty initiating sleep” and “non-restorative sleep”) was defined as the insomnia symptom score.
The HADS was developed by Zigmond and Snaith  as a reliable instrument for screening clinically significant anxiety and depression in women visiting a general medical clinic, and the Japanese translation was performed by Kitamura . The HADS has seven items (odd items) that comprise the anxiety subscale and another seven items (even items) that comprise the depression subscale. The women responded to these items using a four-point Likert scale. Women who had a score of 8–10 points were considered likely to be experiencing anxiety or depression, and those who had a score of 11–21 points were considered to be definitely experiencing anxiety or depression.
The height, weight, waist circumference, and hip circumference of the participants were measured to determine their body mass index (BMI) and waist-to-hip ratio. Their body composition, including body fat, muscle mass, water mass, and visceral fat level, was assessed using a bioimpedance analyzer (MC190-EM; Tanita, Tokyo, Japan). The cardiovascular parameters, including systolic blood pressure, diastolic blood pressure, heart rate, cardio-ankle vascular index (arterial stiffness), and ankle-brachial pressure index, were measured using a vascular screening system (VS-1000; Fukuda Denshi, Tokyo, Japan). Additionally, their resting energy expenditure was measured using a portable, indirect calorimeter (Metavine-N VMB-005 N; Vine, Tokyo, Japan).
Physical fitness was assessed with tests for power, reaction time, and flexibility. Hand-grip strength was measured twice for each hand with a hand dynamometer (Yagami, Nagoya, Japan), with the larger value from each hand used to calculate the average hand-grip strength (kgf). Reaction time was measured with the ruler drop test using a wooden ruler that was 60 cm in length and weighed 110 g (Yagami, Nagoya, Japan). Briefly, the tester held the ruler such that the 0 cm line was surrounded by, but not touching, the seated participant’s outstretched fingers and thumb. The tester then dropped the ruler at an arbitrary time and the participant attempted to catch it as quickly as possible, and the distance that the ruler fell before being caught was recorded. The test was repeated seven times. After omitting the largest and smallest values, the remaining five values were pooled to give an average reaction time (cm). The flexibility was measured using a sit and reach box (Yagami, Nagoya, Japan).
The following lifestyle characteristics were also assessed: the habit of regular exercise (yes, no); amount of daily caffeinated beverage consumption (more than 3 cups, 1–3 cups, none); frequency of alcohol consumption (daily, sometimes, never); and the habit of smoking (more than 20 cigarettes per day, 1–20 cigarettes per day, none).
Factors associated with dizziness
Women were defined as experiencing dizziness if they scored 1, 2, or 3 on the second item in the MHR-QOL’s physical health domain, indicating that they suffered from the symptom once a week or more frequently. Women with and without dizziness were then compared for age, menopausal status, body composition, cardiovascular parameters, basal metabolism, physical fitness, physical and psychological symptoms of menopause (i.e., vasomotor, insomnia, depression, and anxiety symptom scores), and lifestyle characteristics. Next, the factors that significantly differed (p < 0.20) between these two groups at the univariate level were selected as the explanatory variables for a multivariate logistic regression analysis that was conducted to identify the factors that are independently associated with the response variable of dizziness using a stepwise variable selection procedure (p < 0.05).
The group comparison was performed using the unpaired t-test, Mann Whitney U-test, or Fisher’s exact test. The variables that significantly differed between the groups were tested for multicollinearity using Pearson’s or Spearman’s correlation analyses. All statistical analyses, including the multivariate logistic regression analysis detailed above, were performed with SAS version 9.2 (SAS Institute, Cary, NC, USA).