Prospective Study of

e thank Tuttolomondo et al. (1)fortheirinterestinourstudy(2).In response, we would like toclarify several points. First, coding ofstrokes has been ongoing in the Nurses’Health Study since 1976. We have nothad funding to implement coding byTOAST (Trial of Org 10172 in AcuteStroke Treatment) classification (3) buthope to do so in the future. In an internalcomparisonofcodingbyPerthcriteria(4)versus TOAST, the concordance rate forlacunar infarction was very high (91%);however, a percentage of large artery in-farctions as classified by Perth coding cri-teria was classified as “unknown type” byTOAST criteria due to inconclusive ca-rotid Doppler findings. Thus, it is possi-ble that differences in stroke classificationmay have lead to slightly higher risk esti-mates for large artery infarction in ourpopulation.Second, our results are consistentwithdiabetesbeingastrongriskfactorforlacunar infarction. As shown in Table 2 ofthe article, incidence rates for lacunar in-farction were higher than for large arteryinfarction among women with type 2 di-abetes (50 and 36 per 100,000 person-years, respectively), and risk estimateswere also slightly higher for lacunar thanlarge artery infarction (3.6 vs. 2.7 in age-adjusted analyses) compared with womenwithout diabetes.Third, both fatal and nonfatal infarc-tions were included, but only first eventswere considered. Thus, our methods dodifferfromsomehospital-basedstudiesinthat only first stroke events were in-cluded. As stated in the article, resultswere similar for confirmed (medicalrecords)andprobable(letterortelephonecorroboration) cases, thus the combinedresults were presented (see p. 1731, col-umn 2).Finally, as suggested by Tuttolomondoet al., it is possible that there are sex differ-ences in the strength of association be-tweendiabetesandlargearteryinfarction.Our results are limited to women, and weencourage further sex-specific evalua-tions of these associations.