0506822 - ÚI 2020 RIV BE eng A - Abstract
Klaschka, Jan - Malý, Marek - Šípek, A.
Interval estimation of congenital anomaly risk in twins.
40th Annual Conference of the International Society for Clinical Biostatistics. Book of abstracts. Leuven: ISCB, 2019. s. 366-367. ISBN 978-94-6165-287-4.
[Annual Conference of the International Society for Clinical Biostatistics /40./. 14.07.2019-18.07.2019, Leuven]
R&D Projects: GA MZd NV17-29622A
Institutional support: RVO:67985807
Keywords : interval estimate * congenital anomalies * twins
OECD category: Statistics and probability

IN: 40th Annual Conference of the International Society for Clinical Biostatistics. Book of abstracts. Leuven: ISCB, 2019. s. 364-365. ISBN 978-94-6165-287-4. [Annual Conference of the International Society for Clinical Biostatistics /40./. 14.07.2019-18.07.2019, Leuven]. ABSTRACT: In a medical research project, we analyze data of several Czech national health registries in order to assess prevalence of different kinds of congenital anomalies (birth defects). The risk of birth defect in twins is known to differ from that in the children from single pregnancies, and dependence between the twins should be considered, especially in the field of testing and interval estimation. Objectives: Under the assumptions of equal birth defect risk for all children born as twins, and equal phi coefficient (as a measure of dependence) for all twin pairs, we have derived and programmed interval estimates of the risk from twin data. The confidence interval types dealt with are asymptotic Wald (normal approximation), and exact Clopper-Pearson (twice the smaller tail), and Blaker (combined tails). Two different situations are studied: Either the information which pairs of twins belong together is available, or not (which is the case in our project). In the former situation, the phi coefficient may be estimated, while in the latter one we only have to rely on external estimates, or the conclusions drawn must be conditional on the phi value. Methods: While the Wald type confidence bounds are given as closed-form expressions, the exact confidence intervals are calculated numerically. One of the key steps of the numerical procedure is the calculation of the probability mass function of the number of events by a numerical inversion of the characteristic function

Permanent Link: http://hdl.handle.net/11104/0297973