Stability of Kaplan-Meier estimates

save Figure 1

save Figure 2

save Figure 3

save Table 1a

save Table 2a

save Table 1b

option 1: a start and stop date variable, indicating the start date of the observation period and the date of the last observation, or
option 2: a time interval variable, indicating the time interval of the observation period and
an event variable, indicating if an event (e.g. death; coded as 1) or censoring (coded as 0) took place
in case of group comparison: a group variable, indicating the group belonging to 0 or 1

Upload this .csv file to the R shiny application by browsing your directory.

Indicate whether you are using option 1 (start and stop date variable) or option 2 (time interval variable), compare step 1.
Specify the name of the start and the stop date variable or the time interval variable, as well as the event variable in your data set, respectively.
Choose the number of groups in your data: 1 or 2. If 2, choose the name of the group variable in your data set.
Choose which figure format (png, jpeg or pdf) you wish.
Click on “Go” and wait for the results to appear in the “plots” and “derived metrics” tab, respectively.

In the “plots” tab, the following figures are shown and can be downloaded by clicking on “save Figure”

Figure 1: Kaplan-Meier estimate with 95% confidence intervals and numbers at risk.
Figure 2: Upper and lower limits of the stability interval (SIU, SIL) for the Kaplan–Meier estimate (KM) as proposed by Betensky (2015).
Figure 3: Kaplan-Meier estimates of time to censoring, C, observation time, T, and time to censoring among those who are censored, C|C < X.
Figure 4: Difference curve between upper and lower limits of the stability interval (SIU, SIL) and partial difference curves between Kaplan–Meier estimate (KM) and SIU and SIL, respectively.

In the “derived metrics” tab, the following tables are shown and can be downloaded by clicking on “save Table”

Table 1a: Quantile summaries of KM estimate and proposed upper and lower bounds with associated 95% confidence intervals (lower CI, upper CI)
Table 2a: Quantile summaries of C, C|C < X and T=min(X, c) with associated 95% confidence intervals (lower CI, upper CI)

Table 1b: Quantile summaries of KM estimate and proposed upper and lower bounds with associated 95% confidence intervals (lower CI, upper CI)
Table 2b: Quantile summaries of C, C|C < X and T=min(X, c) with associated 95% confidence intervals (lower CI, upper CI)

For adequately reporting time-to-event analysis, the following is recommended:

Including the opening date of accrual, closing date of accrual, and cutoff date for analysis (Shuster, 1991).
Graphical presentation by Kaplan-Meier plots with detailed life tables (Shuster, 1991): see Figure 1.
Calculation of the stability intervals (and derived metrics) as proposed by Betensky (2015): see Figures 2, 4 and Table 1.
If median follow-up is required: calculate the median of C|C < X (Shuster, 1991; Betensky, 2015): see Table 2, middle column.