Dysmenorrhea, in definition, consists of two types being primary and secondary. As to the causes of secondary dysmenorrhea, in our pre-matched sample with 1,325 individuals in Fig. 1, the univariable analysis showed that women with history of diagnosed endometriosis had a higher occurrence rate of dysmenorrhea than women without it (P < 0.001) and this was in line with a study showing that women with endometriosis had 2.2-times more incidents of menstrual pelvic pain/cramping than women without it 20. Indeed, endometriosis is the major leading cause for secondary dysmenorrhea.
As to age, it is a non-modifiable and well-established influential factor of primary dysmenorrhea revealed in previous studies 2,21, that correlated with our pre-matched sample showing that women with dysmenorrhea were younger than women without it (P < 0.001). Nevertheless, after PSM, the powerful influential factors were no longer statistically significant (all P > 0.05) and provided a clearer picture for primary dysmenorrhea.
After PSM, regarding demographic characteristics, BMI (body mass index) showed that there were no differences between the two groups (P = 0.661), and it was inconsistent with a prospective population-based cohort study suggesting that a U-shaped association was found between dysmenorrhea and BMI 22. One explanation is the clear difference of age between studies that included women over 30 in this study and under 30 in the quoted study. Moreover, speaking of marital status, Chiu et al. suggested that unmarried women had 2.59-times more incidents of dysmenorrhea than married women 23; however, our study showed no statistical significance between dysmenorrhea and marital status of women.
In terms of lifestyles and dietary patterns, according to the univariable analysis, our study revealed that exposure to tobacco smoke (P = 0.007), current home-cooking habit (P = 0.01) and tea consumption habit (P = 0.036) were statistically significantly influential factors for dysmenorrhea. In addition, the result of exposure to tobacco smoke was consistent with a previously prospective cohort study that showed a significant relationship between an increased occurrence of dysmenorrhea and exposure to tobacco smoke 11. However, concerning the tea-drinking habit, there was a discrepancy between our study and the previous study, which revealed that tea-drinking was associated with a lower incidence rate of dysmenorrhea 24, as our study reflected the opposite result. Indeed, the relationship of tea-drinking habit to dysmenorrhea remains unclear and needs to be further analyzed. Notably, our study revealed that women having the habit of home-cooking had lower incidences rate of dysmenorrhea than women without this habit and to the best of our knowledge, few studies have mentioned this relationship. This trend might be due to the different nutritious status between women with and without the home-cooking habit 25; specifically, the home-cooking habit could provide more comprehensive nutrition and healthy food intake than the eating-out habit 26.
Furthermore, in this study, we found no association between vegetarian food consumption and risk of dysmenorrhea. This result is in line with the nested case control study conducted by Najafi et al. 12.
Regarding multivariable analysis, our study suggested that women with exposure to tobacco smoke had 1.42-times more chances to develop dysmenorrhea than women without exposure. This seemed to be a risk factor of dysmenorrhea and correlated with the previous studies 11,27. Hopefully, this result could raise public awareness of anti-passive smoking and reduce the occurrence of dysmenorrhea; besides, women with the home-cooking habit could reduce the 26% incidence rate of dysmenorrhea in women without this habit. Clearly, it poses a protective factor for dysmenorrhea in women and probably implies that by home-cooking, women tend to get more nutritious and healthier diet intake that could lessen the menstrual pain, as supported by previous studies 25,26,28. As to the tea-drinking habit, in multivariable analysis, there was no longer a statistical significance between the two groups, and the influential effect of tea-drinking needs to be further analyzed.
There are some limitations in this study. Firstly, this study was a questionnaire-based retrospective design and recall bias for data collection should be considered. Secondly, although powerfully statistical adjustments were performed by propensity score matching, numerous possible confounders may have affected the results of this study. Besides, the use of PSM means that there are no missing values in the covariates used in the analysis, so it reduces somewhat the required sample size. Thirdly, since this study was a retrospective design, it was difficult to pinpoint the causal relationships of dysmenorrhea perfectly, and lastly, most studies have focused on adolescents or young women, but in this study, mature women aged over 30 were targeted; hence, this mature population lacked other supportive and comparison studies.
In conclusion, dysmenorrhea is a common illness among reproductive women and the primary type accounts for most cases of this pain. In this study, PSM was used to mitigate the effects of secondary type being female-related diseases, so specifically influential factors for primary dysmenorrhea could be clearly visualized. Hence, after multivariable analysis, exposure to tobacco smoke and no habit of home-cooking were found to be the potential risk factors for developing dysmenorrhea in women.