How to Use SPSS-Kaplan-Meier Survival Curve

[@alexbrown2341]

Very helpful. A paper I am presenting used these plots and this cleared up much of the questions I was having.

[@ThePhantomJack]

Thank you so much! great video.

[@cousinbebop]

Very useful video, thank you

[@solomonosidele4173]

Hi, this is a very good video. I was very confused when the SPSS output could not produce the curve. After trying different methods to produce the curve l became frustrated. Playing this video enlightened me on what l did wrong and help me thoroughly with the comparison test. What do you do when your data is not statistically significant and have mixed results, greater than.05 or less than.05 Thanks

[@tacappaert]

+solomon Osidele You will have to try and figure out why; it could be sample size, measurement bias or that there is no/an unclear relationship there. Is your result similar to other studies or is it contradictory?

[@strawberryswing92]

Hi! Thank you very much for the video. I have a question, can I use the Kaplan-Meier if I have samples that I took in two different times. I want to compare a control and another group but I have several samples from both groups and I'm not sure if I can use this approach or if it's better if I use the chi square test. I would appreciate your help!!

[@tacappaert]

+Sofia Lorda I don't believe you can compare two different samples within a single analysis.

[@PhillipFreeman1]

I have a question about censoring... I have a population and I want to analyse mortality. I know all patients that have died and the dates they died (factor 1). My issue comes when I call all the alive patients 0 (factor 0) it will censure them as lost to follow up. Actually I have no patients lost to follow up whatsoever they are either alive or dead. So am I right in saying their should be no censure marks on the KM? How do I do this practically?

[@tacappaert]

+Phillip Freeman Have you tried changing the code?

[@atay5510]

+Phillip Freeman I was thinking the same thing. Would love to know the answer

[@whitecoat783]

Thank you so much.

[@MrPyrhana25]

Hi, great video, thank you. I have done the analysis on my two patient groups. There was no significant difference between the two survival curves. However, having tested the data to see if they are normal, I have noticed that one of the groups' survival times are not normally distributed but the other one is. You do mention that you will explain what to do later when your data are not normally distributed. Could you give me some advice on this please? Best, Jason.

[@tacappaert]

+Jason Saunders Williams I would suggest trying to transform the non-normal data. That should correct the skew and allow you to run the analysis.

[@massilbenbouriche2932]

Many thanks for all your videos. Can we apply some kind of post hoc analysis in case of differences ? In your example, can we know where the differences are at beginning of the curve (between which groups) ? And more generally, do you know if we can apply Kaplan-Meier (or any survival analysis) with repeated measures ?

[@tacappaert]

I suppose it would be possible to perform a group comparison test at various time points to see if there is statistical significance. I think you could perform this on repeated measures but I think you would end up with a separate curve for each time point.

[@massilbenbouriche2932]

***** Thank you. I finally did 6 cox regressions (one per condition; with a dichotomic DV). Do you know if there is a test to compare the value of my 6 exp(b) ? A kind of Friedman's Test or repeated anova but designed for hazard ratios ?

[@tacappaert]

Massil Benbouriche I am not aware of such a test.

[@alwayessilent4373]

hi , I do KM plot BY variable A, which has to category to show cumulative risk on the graph, But I want to compare this curves with risk at general population and I want to show this comparison on the curve, I only have cumulative risk up to different age (40,50.. .etc) for general population , how can I use this information to plot?

[@tacappaert]

+alwayessilent You could possibly superimpose the two curves using Excel.

[@wei-mingwatson1822]

Hi, Thank you for this video. I am using the K-M curve to show time to dementia by race. I have a group of initially non-demented subjects and about 20% demented over the course of the study, so when using the K-M curve should only those subjects who became demented be included, or do I include all subjects. In this case, my "censored" cases would be quite high (80%) of all cases.

[@wei-mingwatson1822]

Caitlin Wei-Ming Watson I am trying to imitate the figure "Unadjusted Kaplan-Meier curve for time to dementia associated with race" in this paper: www.ncbi.nlm.nih.gov/pmc/articles/PMC3898154/

[@tacappaert]

+Caitlin Wei-Ming Watson Yes, you would include all data.

[@laylashaabi]

thanks Really helpful

[@whitecoat783]

can we get an "at risk" table below the kaplan meier graph in SPSS? is this option available?

[@tacappaert]

+White Coat You will have to perform a separate analysis to the best of my knowledge.

[@rajeshkhanna5161]

very clear.

[@noahbalmaceda937]

i have a question about those treatments .. what if i have 4 treatments labeled 1,2,3,4 .. can i also include the combination of these treatments labeled 5,6,7,8, etc. ??

[@tacappaert]

+Ciandreu Balmaceda Yes, I think that would be possible.

[@goeylovegreen]

hi! is it possible to input the data in percentage because I am looking for cell survival and not patient survival. thank you very much.

[@tacappaert]

magda goey Yes, as long as the data is numeric.

[@leilagingil]

Thank you

[@kevinstha879]

hi how to calculate ... the 1 year , 2 year , and 5 year suvival rate ... if the time are in months.....????? the curve is very difficult to interpret given only in month ...............

[@tacappaert]

+Anup Shrestha The unit of measure you use for time could any level; seconds, days, months or years. You will have to convert your time element to whatever unit you prefer.

[@moezbdour8567]

hi need help please. i want to do a survival curve kaplan meier for 14 patients using spss20 . 6 are died,7 are lost within a period of time and 1 is still alive. the value status will be 1 for those who died,0 for those who are lost and 0 for the one who is alive is is not?thanks

[@tacappaert]

That is correct, except that those who are still alive will not be included in the analysis since they will not have a time to death value.

[@MrSteniorod]

What is the null for these tests?

[@victorp8065]

hey there. thanks for your video! can someone help me with this: for the y-axis (cum survival) I get comma values, i.e 1,0 0,8 0,6 and so on, but I'd like dots like you have. The problem doesnt seem to get solved if I change the variable type to dot... this probably is really easy but I'm kind of desperate with this. can anyone help please???