Because we did not have detailed information on serologic tests and histology of the small bowel to confirm CD, we used specific medical read codes instead to identify women with CD in the general

population. The method used to define CD has been validated previously in general practice databases,23 therefore we believe the ascertainment of CD in our study is likely to be good. Other recent studies also have made use of read codes in primary care data to identify cases of CD, reiterating that this method to identify a CD population is valid.36 and 37 When we further increased the specificity of our CD diagnosis by restricting our analysis only to cases with supporting evidence of a gluten-free prescription, our estimates remained broadly unchanged. Approximately selleck 30% of the women with CD did not have any record GSK2118436 molecular weight of a gluten-free prescription in our study. Gluten-free prescriptions are considerably costly when prescribed on the UK National Health Service compared with similar products purchased directly.38

Therefore, women may end up purchasing gluten-free products directly, in which case there will be no primary care data recorded on these purchases. Our study also lacked data on compliance with a gluten-free diet. However, similar to most CD studies, we assumed that all women with diagnosed CD are broadly compliant with a gluten-free diet, which seems reasonable given previous evidence suggesting that complete nonadherence to a gluten-free diet is uncommon among patients with CD.39 We must acknowledge that approximately 1% of women in the

United Kingdom have serologic evidence of CD40 and therefore it is likely that there are women with undiagnosed CD among our general population comparison group. It therefore is possible that the presence of these women could have increased the rate of fertility problems in our comparison group if there was truly an increased risk of infertility PDK4 among women with undiagnosed CD as has been implied previously.10, 11, 14 and 41 However, against that hypothesis, our analysis of the women with undiagnosed CD showed that, if anything, their rates of clinically recorded fertility problems were even lower than in the women with diagnosed CD in almost all of the age groups we studied. Finally, there were communication delays between secondary and primary care.42 Although the exact time for this is unknown, there may be inaccuracies in the recording of the exact date of diagnosis of CD, which may have resulted in the misclassification of some diagnosed cases as being undiagnosed. Nevertheless, the rates of fertility problems in both diagnosed and undiagnosed CD were found to be very similar, and also were comparable with the rates in women without CD. Results from the limited studies assessing CD in women with fertility problems have been inconsistent.