ABSTRACT
The nested case-control (NCC) study, originally suggested by Thomas (1977) and described in detail in Chapters 16 and 18, is a cost-effective and efficient design. The basic feature of the design is that at each event time, t i $ t_i $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_1.tif"/> , we sample m i $ m_i $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_2.tif"/> controls from those still at risk at that time. We say that the m i $ m_i $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_3.tif"/> controls are matched to the case at t i $ t_i $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_4.tif"/> . In addition, the controls may be matched on potential confounders. Usually equal number of controls, m i = m $ m_i=m $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_5.tif"/> , are sampled for each case and the number is typically small, i.e., less than 10, and often m = 1 $ m=1 $ https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9781315154084/5e5abef9-a934-4c3a-a7bb-f3e6e5296d03/content/inline-math19_6.tif"/> or 2. The sampled data is then traditionally analyzed with Cox regression using a partial likelihood approach which in practical terms can be described as Cox regression stratified on case-control sets or alternatively as conditional logistic regression.
