Self-controlled case series with multiple event types
Ghebremichael-Weldeselassie, Yonas;
Whitaker, Heather J;
Douglas, Ian J;
Smeeth, Liam;
Farrington, C Paddy;
(2017)
Self-controlled case series with multiple event types.
Computational statistics & data analysis, 113.
pp. 64-72.
ISSN 0167-9473
DOI: https://doi.org/10.1016/j.csda.2016.10.010
Permanent Identifier
Use this Digital Object Identifier when citing or linking to this resource.
Self-controlled case series methods for events that may be classified as one of several types are described. When the event is non-recurrent, the different types correspond to competing risks. It is shown that, under circumstances that are likely to arise in practical applications, the SCCS multi-type likelihood reduces to the product of the type-specific likelihoods. For recurrent events, this applies whether or not the marginal type-specific counts are dependent. As for the standard SCCS method, a rare disease assumption is required for non-recurrent events. Several forms of this assumption are investigated by simulation. The methods are applied to data on MMR vaccine and convulsions (febrile and non-febrile), and to data on thiazolidinediones and fractures (at different sites).