Join Dr. Ana Diez-Roux, for SERforum Live!
May 1, 12pm EST


Do you ever find yourself struggling to figure out a question about epidemiologic methods, or other topics in epidemiology, and don’t know who to ask? The SERforum allows for individuals to answer questions that come up in our daily work around substantive and methodological topics in epidemiology.
All topics may be viewed, but to read and post comments, SER membership is required. If you are member, login! Not a member, join us!

You need to log in to create posts and topics.

Case-cohort Analysis

In a case-cohort study, are the logistic and Cox regression analyses for a fixed cohort the same as those for a dynamic cohort? If not, what modifications have to be made in analyzing a dynamic cohort?

Reply from: Lauren Wise

To my knowledge, I don’t think any modifications would need to be used re: fixed vs. dynamic cohorts. Bryan Langholz and Duncan Thomas are the true experts here.


Here are some helpful readings:


Langholz B, Thomas DC. Nested case-control and case-cohort methods of sampling from a cohort: a critical comparison. American Journal of Epidemiology 1990; 131 (1): 169 –176.

Langholz B, Thomas DC. Efficiency of cohort sampling designs: Some surprising results. Biometrics 1991; 47: 1563 –157.

Kim RS. A new comparison of nested case-control and case-cohort designs and methods. Eur J Epidemiol. 2015 Mar;30(3):197-207.

The Kim one might be best...



Adding on to Eric's question -- are we not limited in assessing proportional hazard assumptions of Cox Proportional analysis within case-cohort study designs?

Also read the attached (and references) paper that might answer your queries. Some text from the Sharp et al. 2014 paper:

" Cox proportional hazards (PH) regression models need to be weighted, with cases outside the subcohort only included in the risk set at the time of their event [1]. Different weighting methods have been described in detail [2] and compared by simulation [3]. The usual standard error estimates from the Cox PH model are not valid in the weighted versions, and should be replaced by alternatives such as a robust jack-knife estimator [4]. Weighted Cox regression models can be fit using standard statistical software packages, including Stata [5] and R [6]."

"If some form of Cox regression model has been used, the proportional hazards assumption should be assessed for each covariate in the analysis. Appropriate methods for assessing this assumption include fitting and testing interactions between covariates and the underlying analysis timescale, or using a correlation test based on Schoenfeld residuals [8]; an extended version of the Schoenfeld residuals test has been proposed for weighted Cox models [9]."

Uploaded files: