Other limitations of this study include possible bias
due to cohort exclusions based on loss to follow-up
(5.2%) or missing data on baseline risk factors (3.9%).
However, because these exclusions were relatively
small, we conclude that they should result in only
minimal bias. In addition, inaccuracy of the baseline
history of stroke could result in cases being identified
during follow-up that were actually diagnosed before
baseline. Another possible source of bias is misclassification
due to the inaccurate diagnoses coded on medical
records and death certificates. Finally, the stroke
identification approach taken in this study, which required
that a self-report of hospitalization for stroke be
confirmed by a medical record listing stroke as a discharge
diagnosis, could lead to incomplete case ascertainment.
Self-reported stroke cases may exist without
a matching medical record because no medical record
was obtained or because the medical record contains a
discharge diagnosis other than stroke. However, we
found that the results were not affected by treating the
unconfirmed stroke cases as true cases
Other limitations of this study include possible biasdue to cohort exclusions based on loss to follow-up(5.2%) or missing data on baseline risk factors (3.9%).However, because these exclusions were relativelysmall, we conclude that they should result in onlyminimal bias. In addition, inaccuracy of the baselinehistory of stroke could result in cases being identifiedduring follow-up that were actually diagnosed beforebaseline. Another possible source of bias is misclassificationdue to the inaccurate diagnoses coded on medicalrecords and death certificates. Finally, the strokeidentification approach taken in this study, which requiredthat a self-report of hospitalization for stroke beconfirmed by a medical record listing stroke as a dischargediagnosis, could lead to incomplete case ascertainment.Self-reported stroke cases may exist withouta matching medical record because no medical recordwas obtained or because the medical record contains adischarge diagnosis other than stroke. However, wefound that the results were not affected by treating theunconfirmed stroke cases as true cases
การแปล กรุณารอสักครู่..