Is exercise associated with primary dysmenorrhoea in young women?


Dr A Daley, Primary Care Clinical Sciences, Clinical Sciences Building, School of Health and Populations Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT UK. Email


Please cite this paper as: Blakey H, Chisholm C, Dear F, Harris B, Hartwell R, Daley A, Jolly K. Is exercise associated with primary dysmenorrhoea in young women? BJOG 2010;117:222–224.

Anecdotal beliefs that exercise is an effective treatment for primary dysmenorrhoea have prevailed for many years although evidence is contradictory. Previous studies have also contained a number of methodological inadequacies. A questionnaire that assessed menstrual pain and levels of exercise was administered to 654 university students. Attempts were made to blind the purpose of the study. A response rate of 91.3% (597/654) was obtained. Analyses showed no association between participation in exercise and primary dysmenorrhoea. Prospective studies would be useful in further research.


The epidemiology of primary dysmenorrhoea is difficult to establish because diverse diagnostic criteria are often used but estimates range from 25% (all women) up to 90% (adolescents),1 with 10% of women describing their symptoms as debilitating.2 Few women consult their general practitioner about their symptoms, preferring to self medicate.3 Several factors have been associated with primary dysmenorrhoea including smoking, depression, heavy menstrual flow, age, parity and body mass index (BMI). It is also popularly thought that exercise reduces the frequency and/or severity of primary dysmenorrhoea. Exercise has been linked with reduced prevalence of primary dysmenorrhoea and associated symptomatology in some studies but not in others1 and a recent meta-analysis4 of observational studies for chronic pelvic pain found that exercise was associated with a small reduced risk of dysmenorrhoea (OR:0.89, 95% CI 0.80–0.99).

The methodological quality of previous observational studies of exercise and primary dysmenorrhoea may be questioned because they have often included small sample sizes, potentially leading to underpowered studies. Past studies have tended to use single item unvalidated questions to assess exercise which do not take account of different intensities of exercise that individuals might engage in (i.e. mild, moderate and vigorous). Many previous studies have largely ignored confounding factors such as age, parity, smoking, stress, mood and use of the oral contraceptive pill (OCP) and failed to rule out the possibility of secondary dysmenorrhoea as the cause of symptoms. Furthermore, authors have frequently made no attempt to blind the study purpose; this is critical because women may over estimate pain and/or symptoms if they are aware of the research question(s). Taken collectively, these issues may, at least in part, explain the inconsistent nature of research in this field to date and there is a need for studies that address some, or all, of the methodological issues raised here.


Women aged 18–25 years were recruited from a university located in the West Midlands. Thirteen heads of schools within the university were approached for permission to distribute the study questionnaire to female students at the end of a lecture; 12 schools agreed to this request. Data were collected between February and March 2008. A total of 654 questionnaires were distributed. The questionnaire was entitled ‘Women’s Health and Lifestyle Questionnaire’ so as to blind the specific purpose of the study. Participants were asked to indicate their age, ethnicity, height, weight and current smoking behaviour. Items relating to age at menarche, length of menstrual cycle, whether currently menstruating (menstrual status), use of OCP or intrauterine device, parity and history of gynaecological diseases were also included. Demographic items were piloted for meaning and presentation.

A modified version of the Godin Leisure-Time Exercise Questionnaire5 was used (Appendix S1). This questionnaire asks ‘during a typical 7-day period (a week), how many times on average do you do the following kinds of exercise (i.e. strenuous, moderate and mild intensity) for more than 15 minutes during your free time?’ For the purposes of this study we amended this statement to ‘30 minutes or more’, in line with the current government recommendations for exercise. Thus, participants were asked to report how many times per week they participate in strenuous (e.g. running, vigorous swimming, netball and circuits), moderate (e.g. fast walking, cycling, easy swimming, dancing) and mild (e.g. archery, bowling, golf, easy walking) intensity exercise for more than 30 minutes during their free time. Total weekly leisure exercise score is calculated in arbitrary units by summing the products of the separate components by 9, 5 and 3 METs, respectively, as shown in the following formula: Weekly leisure exercise score = (9 × Strenuous) + (5 × Moderate) + (3 × Light). The amended items were piloted for meaning with University students prior to data collection.

Women rated menstrual pain using a visual analogue scale (VAS) between 0 (no pain) and 10 cm (extremely severe pain). Pain was also assessed using a verbal multidimensional pain score (VMPS),6 which grades pain as none, mild, moderate or severe according to the impact on daily activity, systemic symptoms and analgesic requirements. Women who rated their pain as greater than zero were considered to have primary dysmenorrhoea. Participants were also asked to specify any analgesic medication and alternative treatments that they used to treat dysmenorrhoea. The mental component score subscale of the SF-12 questionnaire (herein referred to as mood) was included in the study questionnaire in an attempt to further blind the purpose of the study.

Data analysis

Data were analysed using SPSS (version 15.0; SPSS Inc., Chicago, IL, USA). Multiple regression analyses between level of pain (as measured by VAS) and leisure time exercises score was conducted, controlling for BMI, ethnicity, OCP use, menstrual and smoking status. VMPS scores were coded as a binary outcome (VMPS scores of 0 or 1 = no/minimal pain, and VMPS 2 or 3 = moderate/severe pain) for use in logistical regression analysis of leisure time exercise category (high and low categories according to a median split of scores) and VMPS category, controlling for ethnicity, BMI, OCP use, smoking and menstrual status was performed.


A response rate of 91.3% (597/654) was obtained. The following participants (total n = 27; 4.1%) were excluded (age >25 years, n = 5; endometriosis, n = 1; pelvic inflammatory disease, n = 1; fibroids, n = 2; ovarian cysts, n = 18), resulting in a final sample size of 570 participants. See Table 1 for participant demographics. Using VMPS, 21.6% of participants reported no pain and 78.4% experienced varying levels of pain (mild = 50.5%; moderate = 21.4%; severe = 5.8%). This corresponds to 72.1% experiencing no/minimal pain and 27.9% experiencing moderate/severe pain.

Table 1.   Participant characteristics
 n (%)
Age (years)
18–19357 (63.1)
20–21195 (34.5)
22–239 (1.6)
24–255 (0.9)
Mean (SD)19.3 (1.1)
BMI (kg/m2)
<20156 (29.9)
20 < 25319 (61.1)
25 < 3044 (8.4)
≥303 (0.6)
Mean (SD)  21.6 (2.8)
White441 (77.9)
Non-white125 (22.1)
Smoker25 (4.4)
Non-smoker540 (95.6)
Taken226 (40.1)
Not taken338 (59.9)
Menstrual status
Menstruating112 (19.6)
Not menstruating458 (80.4)

The multiple regression analysis between pain VAS and leisure time exercise scores, after controlling for mood, ethnicity, BMI, OCP use, smoking and menstrual status was non-significant (P = 0.75). Logistic regression between VMPS category and leisure time exercise category, controlling for mood, ethnicity, BMI, OCP use, smoking and menstrual status was non-significant (P = 0.34).


In line with many other studies, but dissimilar to others, participation in exercise was not associated with dysmenorrhoea as measured by VAS score or VMPS, after controlling for confounding variables. A number of factors may account for the discrepancy between studies. A recent review1 showed that in studies where more than 500 participants were included there tended to be no association between dysmenorrhoea and exercise/physical activity patterns but smaller studies (<500 participants) were more likely to report positive associations, where bias is also more likely to be present. In addition, this review1 showed that small studies were less likely to have blinded the study purpose or controlled for possible confounders, making their findings uncertain. Anecdotal beliefs that exercise is an effective treatment for primary dysmenorrhoea have prevailed for many years and while it might seem intuitively appealing to promote exercise as a treatment for menstrual disorders such as primary dysmenorrhoea, the findings from this study along with many others, would not support such a view. Of course, there are many other important health reasons for encouraging women to be physically active and exercise performed in moderation is unlikely be harmful.

The findings from this study should be interpreted in light of several limitations and strengths. Retrospective reporting of menstrual symptoms may lead to inaccurate reporting of symptoms. Having said this, we reported a high prevalence of dysmenorrhoea (78.4%), yet the proportion of women reporting severe menstrual pain was low even compared with prospective studies; suggesting that pain scores were unlikely to have been over reported through the use of a retrospective design. Exercise was assessed by self-report and the possibility that participants overestimated this because of social desirability cannot be excluded.

The sample comprised university students who may be more educated and affluent than women not attending university; this may reduce the generalisability of the findings. Particular strengths of the study are the high response rate of (91.3%) and the fact we were able to control for variables known to influence both symptoms and exercise participation. We also included a large proportion of women from non-white ethnic backgrounds, which many previous studies have neglected to do; this will serve to increase the generalisability of our findings.

To conclude, dysmenorrhoea is a common gynaecological complaint and thus an important public health issue. While some small observational studies suggest that participation in regular exercise decreases primary dysmenorrhoea and associated symptomatology, we found no association between these outcomes. Clearly more prospective research is required before women are informed that exercise is an evidence-based intervention that is likely to reduce or alleviate menstrual pain.

Disclosure of interests


Contribution to authorship

AD devised the study. HB, CC, FD, BH and RH collected the data. HB and KJ conducted the statistical analyses. All authors contributed to the writing of the manuscript.

Details of ethics approval

University of Birmingham BMedSci Internal Ethics Committee.




We would like to thank the participants for completing the study questionnaire.