Greetings,
I would like to ask a dumb question if I may. Suppose in SPSS, I have data on the outcome for mortality (dead versus alive) and raw data for age. Should I just leave the data for age as continuous variable or should I dichotomize it ( = 65 years) ?
I have heard that we are allowed to use continuous data for Kaplan Meier analysis such as for age, weight, leukocyte county, etc. I wonder why so many papers analyze the Kaplan Meier survival with categorical data instead?
Many thanks
Don't catagorize data whenever possible. You usually lose useful information. Best, David Booth
Yeah, perhaps they(researchers) found it "easier and and trendy", just joking. Researchers trying to analyze continuous variables with Kaplan Meier Analysis usually encounter problems of how to convert continuous to group-like variable, the log-rank test only compares survival between groups. As warned Prof. @David Eugene Booth, "Don't catagorize data whenever [you think it's] possible. You usually lose useful information." ie., just leave the data for age (and similar on continuous scale, etc) as continuous variable!
You might want this information: There is a variable called the stratification variable(which represents the grouping effect) in Kaplan-Meier survival analysis (KMSA). And as the name suggests, the stratification variable should be a categorical type of variable. StatisticsSolution.com thinks, researcher[s] should note that "Kaplan-Meier survival analysis (KMSA) provides incorrect results when covariates other than the time are considered as the prominent aspect in obtaining the extent of a certain consequence." https://www.statisticssolutions.com/kaplan-meier-survival-analysis-kmsa/
Researchers trying to analyze continuous variables with Kaplan Meier Analysis usually encounter problems of how to convert continuous to group-like variable, the log-rank test only compares survival between groups. Alternatively, "methods that don’t dichotomize (grouping), like Cox regression, should be used instead". Tomas Bencomo (
Oct 17, 2019, towardsdatascience.com accessed on 20/12/2020 at https://towardsdatascience.com/kaplan-meier-mistakes-48cd9e168b09 )
You can use the Cox proportional hazards (CPH) model for analyzing categorical and continuous data for survival analysis. Cox proportional hazards (CPH) model considers multiple independent variables for estimating differences. Read more here https://www.reneshbedre.com/blog/survival-analysis.html
As others have said, you can use a suitable model such as the Cox proportional hazards model to analyze the effect of a continuous variable on a time-to-event outcome. You can then use this model to perform g-computation and create graphics such as the "survival area plot". This is basically a Kaplan-Meier plot but displays the survival probability at t as a function of the continuous variable as well. This is documented in a preprint here: Preprint Visualizing the Causal Effect of a Continuous Variable on a ...
The contsurvplot R-package makes it easy to create these types of plots as well: https://github.com/RobinDenz1/contsurvplot
- Badges
- Science topic
- Similar topics
- Psychology