There is ongoing debate in the scientific literature regarding what constitutes female sexual dysfunction (FSD) and dyspareunia [9, 10]. In fact, there are differences between the definitions of dyspareunia provided by professional organizations [11–13]. These inconsistencies have the potential to affect prevalence estimates reported. For example, the American Psychiatric Association’s DSM-IV [12] and the World Health Organization’s ICD-10 [13] define dyspareunia as pain associated with sexual intercourse. In the last decade, Basson et al. have revised definitions of FSD [11]. They broaden the definition of dyspareunia to also include pain with attempted or completed vaginal entry. As a result of this broader definition, studies using Basson et al.’s more encompassing definition are likely to report relatively higher prevalence estimates.

Painful intercourse can be associated with a range of conditions such as endometriosis, interstitial cystitis (IC), and vaginismus [6, 14]. DSM-IV and ICD-10 stipulate that sexual pain resulting from “general medical conditions” or “local pathology” should not be classified as dyspareunia [12, 13]. In addition, diagnostic systems usually classify dyspareunia and vaginismus as separate, mutually exclusive sexual dysfunctions [11–13]. Some authors have debated whether this distinction is appropriate, pointing to the lack of evidence related to the clinical presentation of vaginismus and superficial dyspareunia [6, 14]. ICD-10 and DSM-IV also exclude sexual pain due to lubrication problems. Many studies, however, do not exclude women who suffer from these conditions when reporting the prevalence of dyspareunia [8, 15]. Consequently, these studies may overestimate the prevalence of dyspareunia.

There is increasing evidence that aspects of study design and conduct have a substantial impact on the prevalence of dyspareunia reported in published studies. A recent meta-analysis investigated associations between the prevalence of sexual difficulties reported in 55 published studies and the design features of those studies [2]. Data collection procedures, inclusion criteria, duration of sexual difficulty recorded, sample size, and response rate were all significantly associated with the reported prevalence of at least one type of sexual difficulty.

Reported prevalence estimates of sexual pain were lower in studies that conducted interviews in person compared with studies that used self-administered questionnaires. It is possible that interviewing in person is a more accurate way of gathering data on dyspareunia. Alternatively, when women are interviewed in person, embarrassment and social desirability may bias responses.

Studies in which the duration of sexual pain recorded was longer (3–6 months or more) also reported lower prevalence estimates [2]. These associations between prevalence and study design were independent of likely predictors of true variation in prevalence such as study location, study year, and age range of participants. There is further evidence that dyspareunia can persist for varying durations [3, 16]. It is therefore plausible that investigations that only record longer-lasting difficulties will report lower prevalence estimates. In addition, some investigations recruit participants from clinical settings [17, 18], which may limit the generalizability of the results obtained. Women recruited in this way may be different from the general population in a range of ways that could affect their sexual function, such as being less healthy generally and belonging to a demographic that has better access to health care.

A wide variety of instruments have been used to assess dyspareunia. Studies that have employed simple questions are common in the literature [16, 19] and include well-cited and influential studies such as that by Laumann et al. [15]. These simple questions often ask respondents to report if they have experienced sexual pain for a month or more during the previous year [16, 19]. By contrast, a range of validated multi-item instruments have been developed [20, 21] and are being increasingly used in prevalence studies [8, 18].