# Adjusted Survival Curves

Has anyone looked into the article, Adjusted Survival Curves, implemented in the survminer package? I cannot find much information regarding it nor many papers that cite this reference. I would primarily like to know how I would explain the "conditional method" using the cox model with covariates? Is this inverse probability weighting or some other procedure - I do not understand the statistics referenced in this paper.

## 1 Answer

This paper is summarizing how to estimate adjusted survival curves. In brief, there are two common approaches :

Marginal estimation. You create a marginal population (using for example inverse probability weighting), and estimate the survival curves of interest in this population. The main limitation here is that the curves will not have any individual interpretation, while the main advantage is that you can use non-parametric methods for estimating the curves (as you don't have to adjust for covariates any longer)

Conditional estimation. You estimate a Cox model (or other regression models for the hazard) and you use the mathematical relationship between hazard and survival function to estimate the survival curve for each covariate pattern (i.e. for each individual in the study). The main problem here is that you will have to decide where/how to fix the several covariates in the model. Moreover, if you fit a proportional hazard model, survival curves will also be affected by this assumption.

Here are two common references for, respectively, marginal and conditional estimation:

• Cole SR, Hernán MA. Adjusted survival curves with inverse probability weights. Computer methods and programs in biomedicine. 2004 Jul 1;75(1):45-9.
• Ghali WA et al. Comparison of 2 methods for calculating adjusted survival curves from proportional hazards models. Jama. 2001 Sep 26;286(12):1494-7.