Instruments
The original DRSP contains 21 items pertaining to premenstrual symptoms and 3 items to describe dysfunction in daily life caused by these symptoms, and is based on diagnostic criteria for PMDD from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-Ⅳ). All items are scored daily throughout the menstrual cycle on a scale of 1 (not at all) to 6 (extreme). The J-DRSP was developed according to the formal procedures outlined in the translation guidelines of the latest Patient-Reported Outcomes (PRO) Consortium6.
For the present study, we created the “DRSP-JAPAN” smartphone app, which allows users to log their daily symptoms for the J-DRSP. We released the app at the App Store and on Google Play for free in 2018 (Fig. 1). Following user authentication, participants can rate their condition in the DRSP-JAPAN app and log their symptoms on the server every day. The app offers a visual display of daily condition, the severity of which is indicated by various colors and numbers, enabling participants to look back on the various changes in their symptoms over time.
The app has a blank check function, and a reminder mail is sent automatically if data have not been logged for more than three consecutive days.
Participants
Participant recruitment flyers were posted between September 2018 and March 2019 at Kyoto University, Doshisha Women's College of Liberal Arts, the staff lounge rooms of Kyoto University Hospital, and seven gynecology clinics in Japan where many asymptomatic women comes for medical check. Women were eligible for study participation if they were over 20 years of age, had self-reported regular menstruation, and used a smartphone or tablet. Those with any current mental disorder or who were taking oral contraceptives or ovulation inhibitors were excluded. All participants installed the DRSP-JAPAN app.
Measurements
All data were collected through the DRSP-JAPAN app. At the beginning of the study, we collected demographic data on participant age and self-reported duration of menstruation. In addition to bleeding volume (one item of the J-DSRP), participants were prompted to log the start date of menstruation. Participant J-DSRP data are logged daily until the participants are notified that the research study has come to an end. Data from each cycle comprise those from half of the 7 days before menstruation starts and half of the days spanning day 4 to day 10 after menstruation begins. Once those data are collected, no further data are collected for that cycle.
Statistical analysis
In order to assess factorial validity, we conducted exploratory factor analysis (EFA) using data for the 21 symptom items on the day before menstruation began. To measure the factorability of the correlation matrix, we conducted the Kaiser-Meyer-Olkin (KMO) test and Bartlett's test of sphericity. Eigenvalues and screen plots were used to determine the number of factors, and promax rotation was used. Next, confirmatory factor analysis (CFA) was performed to assess the adoption. We then calculated the comparative fit index (CFI) and root mean square error of approximation (RMSEA) in order to analyze model fitness. CFI values range from 0 to 1, with larger values indicating a better fit. A CFI value of 0.95 or higher is regarded as a good fit7. RMSEA ranges from 0 to 1, with smaller values indicating a better model fit; values less than 0.06 indicate an acceptable model fit.
To confirm criterion validity, we compared the J-DRSP total score obtained the day before menstruation began to that obtained 10 days after the menstruation start date using the Wilcoxon signed-rank test. P < 0.05 was considered statistically significant.
To examine test-retest reliability, J-DRSP scores from each of the two days of the follicular phase and the luteal phase were compared. Specifically, with cycle day 1 (CD1) set to represent the menstruation onset date, we compared data from CD9 to CD10 and CD-1 to CD0.
We determined the intraclass correlation coefficient (ICC), percent agreement, and weighted kappa coefficient. Percent agreement and kappa coefficients were calculated after the total J-DRSP was categorized into six levels. As a sensitivity analysis, ICC was determined for any participant whose symptoms interfered with daily life. Those were selected using J-DRSP items 22, 23, and 24, which concern dysfunction in productivity, social activities, and relationships by the symptom items from item 1 to 21, and those who scored 1 (none) for all of items 22, 23, and 24 were excluded. To measure internal consistency, we also calculated Cronbach's alphas for the total J-DRSP score as well as for each subscale.
Developing the Short-Form J-DRSP (J-DRSP(SF))
The short-form J-DRSP (J-DRSP(SF)) was created using classical test theory. First, items for which the factor load was less than 0.5 were excluded after conducting EFA for the original J-DRSP with promax rotation. Next, any items demonstrating sufficient model fit were explored using RMSEA and CFI. After developing the J-DRSP(SF), we conducted CFA and determined Cronbach's alphas, ICCs, and criterion validity for the J-DRSP(SF) total score and each subscale. ICC values were calculated in both the follicular and luteal phases, i.e., CD9 to CD10 and CD-1 to CD0, respectively. Sensitivity analysis was conducted again using J-DRSP items 22, 23, and 24. Stata 15.3 was used for all statistical analyses.
Ethical considerations
The Kyoto University Ethics Committee approved the study protocol (R1593). Written informed consent was obtained from each participant.