Abstract
A
The accuracy of a history of “yeast infections” in women is fraught with many limitations, due to both recall bias and bias based on the presence of other genital symptoms, in addition to the well-documented inaccuracies associated with making the diagnosis of CVV. Data clearly show that the diagnosis of CVV is no better than a coin toss—by patients themselves (even when they have a history of yeast infections), 4 –6 but also by clinicians making the diagnosis after evaluating women with symptoms. 5,7 Few clinicians use microscopy to diagnose CVV, 8 and even when used, the accuracy of CVV diagnoses compared with that of a yeast culture or DNA-based test is poor. 4,9 Thus, the value of historical data to accurately document past CVV has traditionally been poor.
In this issue, Harlow et al. have performed a much needed analysis of the relationship between the history of CVV and current vulvodynia, not only when past diagnoses are taken at face value, or when cases and controls have similar inaccuracies in their reporting (“nondifferential misclassification”), but also assessing how that relationship estimate changes when the accuracy of the CVV diagnosis varies between cases and controls. The accuracies between these two groups may differ in sensitivity (likelihood of diagnosing yeast when it is actually present) or, alternatively, in specificity (likelihood of diagnosis of no yeast when it is actually not present), or both. They found, based on history alone, that a vulvodynia diagnosis was five times as likely (OR = 5.5) among those who reported a history of 10 or more yeast infections. Furthermore, those with vulvodynia were twice as likely to report a history of 10 or more infections since their symptoms of vulvodynia started than were the controls. At face value, although causality cannot be determined, it appears that these two conditions are clearly associated. However, the accuracies of the recollection of yeast infections are likely to vary in cases and controls. If the controls were more likely to report yeast infections when they actually did not have them (more false positives—and lower specificity), that association was even stronger (greater OR). However, if those with vulvodynia were more likely to have false positive CVV diagnoses (lower specificity), the OR weakened or reversed, suggesting no statistical relationship between yeast infections and vulvodynia. Clearly, the accuracy of the CVV diagnoses and the recall of these diagnoses impact the interpretation of the nondifferential findings—as demonstrated by Harlow et al.
Although this approach does not stipulate the direction or degree of these inaccuracies in the real world, it does allow one to hypothesize how the relationship might differ based on one's perception of the likelihood of over diagnosis (lower specificity) and underdiagnosis (lower sensitivity) in women with vulvodynia, in women without vulvodynia, and even in women who have symptoms of vulvodynia and are unaware of their diagnosis. Women with vulvodynia, by definition, experience vulvovaginal symptoms. These symptoms are known to wax and wane. 10 Women with vulvodynia may repeatedly present with new or recurring symptoms that may be interpreted as CVV—decreasing specificity by increasing the rate of false positive diagnoses. Controls are less likely to report vulvovaginal symptoms at the frequency seen in women with vulvodynia—thereby lessening their chances of overdiagnosis and hence resulting in higher specificity. Hence, vulvodynia patients would be expected to have a lower specificity than controls—which, as shown in Tables 3 and 5 in their article, results in a lesser or nonexistent association between CVV infections and vulvodynia. Where cases and controls exist on this continuum of CVV accuracy at any one time remains unclear.
This article clarifies that the diagnosis of vaginal yeast infections is critical in the interpretation of the relationship between CVV and vulvodynia, and that improving on this, and recognizing the probability of misdiagnoses, is critical if we are to better understand the relationship between vaginal yeast infections and vulvodynia. Harlow et al. have advanced our understanding of the complex relationship between CVV and vulvodynia by demonstrating the impact that recall and other biases have on interpretation of results. Their findings highlight the need to better understand how these inaccuracies differ among women with and without vulvovaginal symptoms. Future studies are needed that include culture or DNA verification of diagnosis, to allow us to better understand these biases, thereby allowing us to better interpret the relationship between yeast infections and vulvodynia. In the meantime, the work by Harlow et al. helps clarify the limitations of the misclassification arising from the nonverified history alone—with or without clinician diagnosis—and the need for improved diagnosis, educating patients, and providers regarding the degree of inaccuracy of CVV diagnosis based on symptoms alone, and further research on the relative sensitivity and specificity of the diagnosis based on symptoms among those with vulvodynia and those without chronic vulvar pain.