Abstract
Providing information about the risk of disease and clinical factors that may increase or decrease a patient's risk of disease is standard medical practice. Although case-control studies can provide evidence of strong associations between diseases and risk factors, clinicians need to be able to communicate to patients the age-specific risks of disease over a defined time interval for a set of risk factors.
An estimate of absolute risk cannot be determined from case-control studies because cases are generally chosen from a population whose size is not known (necessary for calculation of absolute risk) and where duration of follow-up is not known (necessary for calculation of incidence). This problem can sometimes be overcome by using a nested case-control design.
We have collected data on a National Cancer Institute funded population-based cohort study. This study contains a matched set of cases and controls within the cohort. This design is more cost-efficient than a full cohort study since expensive predictor variables (genomic measures, sex hormone levels, mammographic breast density) are measured on all of the cases, but on only a sample of the cohort who did not develop the outcome of interest (the controls). In addition, this design avoids the potential biases of conventional case-control studies that draw cases and controls from different populations. Importantly, the presence or absence of the outcome of interest has been established for the entire cohort within the same time period.
The specifics of the sampling in our study do not adhere to the assumptions for absolute risk estimation methods previously developed in the literature. Here we introduce a novel method which provides locally efficient estimators to predict the absolute risk of a cohort from measures only taken on the matched case-control participants. The proposed method is evaluated using simulation studies and survival data from women with ductal carcinoma in situ, a non-invasive form of breast cancer. A generalization of the proposed method is related to other similar sampling designs such as nested case-control, case-cohort, and two-stage case-control.
Disciplines
Statistical Models | Survival Analysis
Suggested Citation
Molinaro, Annette M.; van der Laan, Mark J.; Moore, Dan H.; and Kerlikowske, Karla, "Survival Point Estimate Prediction in Matched and Non-Matched Case-Control Subsample Designed Studies" (August 2005). U.C. Berkeley Division of Biostatistics Working Paper Series. Working Paper 149.
https://biostats.bepress.com/ucbbiostat/paper149
Comments
Address correspondence to annette.molinaro@yale.edu