Semin Thorac Cardiovasc Surg. 2022 Jan 26:S1043-0679(22)00005-3. doi: 10.1053/j.semtcvs.2022.01.001. Online ahead of print.
ABSTRACT
Survival analyses are a group of statistical principles applied to accurately analyze the length of time until a previously defined event occurs. Increasing survival is the underlying goal for most medical interventions, and is particularly critical in oncology-related surgical fields. In order to justify the application of a novel surgical intervention or other cancer therapy, one must first definitively show improvements in patient survival compared to the existing method of treatment. In order to definitively recommend one treatment option over another, it is paramount to design a study that addresses and minimizes sources of bias where possible. This can be challenging due to numerous factors including selecting an appropriate study design, dealing with censored data, obtaining an appropriate sample size, and performing robust statistical analysis. It is critical for surgeons in every stage of training to be able to both understand and apply these methods in order to improve patient care. In this review, we discuss approaches to design survival studies, relevant errors/biases and how to account for them, and cover several field-standard methods to analyze survival data including Kaplan-Meier plots, the log-rank test, and Cox Proportional Hazards Models. Importantly, this is accompanied by easily repurposable examples of how to implement these analyses in both R and GraphPad Prism using a publicly available survival dataset. This review will assist surgeons of all training levels in the design and analysis of survival studies and serve as a starting point for advancing patient care.
PMID:35092846 | DOI:10.1053/j.semtcvs.2022.01.001
