Survivor Analysis Assignment
M. AGUNG PRASETYA ADNYANA YOGA
21/475855/PKU/19330
UNIVERSITAS GADJAH MADA
2021-2022
Question:
1. Create Kaplan-Meier survival curve for the overall survival of this glioma patients. Examine
95% confidence interval of survival function!
2. Did combine treatment perform better? Please make Kaplan-Meier survival curve.
3. Using log-rank, is the difference statistically significant? If not do you think practically
significant?
4. Using Cox regression or proportional hazard model, please adjust the effect of combine
treatment with age. Consider age group into two groups only (below 40 and 40+)
5. How are you going to balance between practically and statistically significant in this results?
Why?
Anwers :
1. a. Survival probability toward age group
the first things that I did before analyse the Kaplan meire curve, I shall to write a
particular command in stata to prepare my data that further can be used to do statistical
analyse, this is the command as follow “ stset id, failure (event).
. stset id,failure(event)
failure event: event != 0 & event < .
obs. time interval: (0, id]
exit on or before: failure
40 total observations
0 exclusions
40 observations remaining, representing
27 failures in single-record/single-failure data
820 total analysis time at risk and under observation
at risk from t = 0
earliest observed entry t = 0
last observed exit t = 40
Survival probability among age group <40 years old and age group 40-59 years old are
more higher compared with age group more than 60 years old. During the admission in
the darmais hospital, the survival probability of participants in age group 60 years old is
very short, its only around less than 33 weeks and the survival probability slowly
decreasing dramatically as long the time increasing of hospitalization.
1.00
0.20 0.40 0.60 0.80
Survival Probability
0.00
0 10 20 30 40
analysis time
Observed: agegroup = <40 Observed: agegroup = 40-59
Observed: agegroup = 60+ Predicted: agegroup = <40
Predicted: agegroup = 40-59 Predicted: agegroup = 60+
b. Survival probability toward event (status at the end of follow up)
Participant in age group < 40 years old have a good survival probability or have a
probability last long to survive compared with participants in the older age groups like
age group 40 -59 and >60 years old. Survival probability among people who have high
grade glioma in age group <40 years old is 1.00 it means that the people in age group
<40 years old have a long survival life during medication in hospital and keep stable
during hospitalization.
Meanwhile the people in age group >60 years old have a lower survival life (survival
probability) tend to slower decrease during hospitalization start from 0.99 going down
until 0.00 it means that the risk of dying in that group is higher compared than people in
age group <40 years old.
1.00 0.80
Survival Probability
0.20 0.40 0.60
0.00
0 10 20 30 40
analysis time
Observed: event = survive Observed: event = dead
Observed: event = censored Predicted: event = survive
Predicted: event = dead Predicted: event = censored
. sts list
failure _d: event
analysis time _t: id
Beg. Net Survivor Std.
Time Total Fail Lost Function Error [95% Conf. Int.]
1 40 0 1 1.0000 . . .
2 39 1 0 0.9744 0.0253 0.8316 0.9963
3 38 1 0 0.9487 0.0353 0.8102 0.9869
4 37 1 0 0.9231 0.0427 0.7802 0.9745
5 36 1 0 0.8974 0.0486 0.7494 0.9602
6 35 0 1 0.8974 0.0486 0.7494 0.9602
7 34 1 0 0.8710 0.0538 0.7174 0.9442
8 33 1 0 0.8446 0.0583 0.6863 0.9271
9 32 1 0 0.8183 0.0622 0.6560 0.9090
10 31 1 0 0.7919 0.0655 0.6264 0.8901
11 30 1 0 0.7655 0.0685 0.5975 0.8705
12 29 0 1 0.7655 0.0685 0.5975 0.8705
13 28 1 0 0.7381 0.0713 0.5676 0.8498
14 27 1 0 0.7108 0.0737 0.5384 0.8284
15 26 0 1 0.7108 0.0737 0.5384 0.8284
16 25 0 1 0.7108 0.0737 0.5384 0.8284
17 24 0 1 0.7108 0.0737 0.5384 0.8284
18 23 1 0 0.6799 0.0767 0.5044 0.8045
19 22 0 1 0.6799 0.0767 0.5044 0.8045
20 21 1 0 0.6475 0.0796 0.4693 0.7790
21 20 1 0 0.6151 0.0819 0.4354 0.7528
22 19 1 0 0.5828 0.0838 0.4026 0.7258
23 18 0 1 0.5828 0.0838 0.4026 0.7258
24 17 1 0 0.5485 0.0856 0.3681 0.6970
25 16 1 0 0.5142 0.0868 0.3349 0.6673
26 15 1 0 0.4799 0.0875 0.3028 0.6369
27 14 1 0 0.4456 0.0877 0.2718 0.6057
28 13 1 0 0.4114 0.0874 0.2418 0.5736
29 12 1 0 0.3771 0.0866 0.2128 0.5408
30 11 1 0 0.3428 0.0852 0.1849 0.5071
31 10 0 1 0.3428 0.0852 0.1849 0.5071
32 9 1 0 0.3047 0.0839 0.1540 0.4701
33 8 1 0 0.2666 0.0816 0.1249 0.4318
34 7 1 0 0.2285 0.0783 0.0976 0.3920
35 6 0 1 0.2285 0.0783 0.0976 0.3920
36 5 1 0 0.1828 0.0748 0.0656 0.3465
37 4 0 1 0.1828 0.0748 0.0656 0.3465
38 3 1 0 0.1219 0.0704 0.0272 0.2927
39 2 0 1 0.1219 0.0704 0.0272 0.2927
40 1 0 1 0.1219 0.0704 0.0272 0.2927
We can use sts list in our command in stata to get some information about survival function in
our participant, according from the lecture by professor siswanto the survival function is a
function of time, and is defined as the probability of the event in question not occurring by time t
( or the patient surviving until time t or late).
2. a. Make Kaplan-Meier survival curve.
1.00 0.80
Survival Probability
0.40 0.60
0.20
0.00
0 10 20 30 40
analysis time
Observed: sex = Male Observed: sex = Female
Predicted: sex = Male Predicted: sex = Female
According from Kaplan meier curve, female group in this research have a long survival
probability compared with male group. But between two group female and male who
have high grade glioma its have a a same probability to experience died during
hospitalization while the value of survival probability tend decrease slowly start from
1,00 to 0.00.
3. a. Log Rank Test
To do statistical analyses with log rank test, we can use command in stata as follow ” sts
test event , logrank “. By definition, the log rank test is the most commonly used
statistical test for comparing the survival distributions of two or more groups (such as
different treatment groups in a clinical trial).
And the result there is statistically significant about some event related with survival
function toward the event observed and events expected while the p value is
<0.05(0.0002). Participant who died in this research due to glioma is 18 people and there
is 9 people who censored in this research.
Log-rank test for equality of survivor functions
Events Events
event observed expected
survive 0 10.27
dead 18 11.64
censored 9 5.09
Total 27 27.00
chi2(2) = 17.39
Pr>chi2 = 0.0002
4. To analyses cox proportional analysis, based on that we can do statistical analysis use
command as follow “ stcox combined i.agegroup”.
- We find that the combined drug therapy has been shown a result in a lower hazard
and therefore a longer survivor time controlling for age.
- Older patients are more likely to die
- The model as a whole, there is no statistically significant because the p value is more
than 0.05.
- The hazard ratio reported correspond to a one unit change in the corresponding
variable.
. stcox combined i.agegroup
failure _d: event
analysis time _t: id
Iteration 0: log likelihood = -77.89889
Iteration 1: log likelihood = -75.093733
Iteration 2: log likelihood = -74.134359
Iteration 3: log likelihood = -74.131083
Iteration 4: log likelihood = -74.131083
Refining estimates:
Iteration 0: log likelihood = -74.131083
Cox regression -- no ties
No. of subjects = 40 Number of obs = 40
No. of failures = 27
Time at risk = 820
LR chi2(3) = 7.54
Log likelihood = -74.131083 Prob > chi2 = 0.0567
_t Haz. Ratio Std. Err. z P>|z| [95% Conf. Interval]
combined 1.718363 1.158003 0.80 0.422 .4586607 6.437815
agegroup
40-59 .6695187 .335442 -0.80 0.423 .2507806 1.78744
60+ 2.852081 1.650749 1.81 0.070 .9172695 8.86802
5. Here my result in statistical analysis used hazard ratio have shown while the old people
or older participant will have 2.5 times get a higher risk to experience die compared with
non older people but in statistical analysis is not significant. Regarding from journal
related with this assignment about clinical epidemiology and risk factor for survival rate
for high grade glioma patient treated at the National Cancer Centre Hospital Dharmais
Jakarta Indonesia, have been mentioned about the prognosis for patients with HGG is
unfortunately very poor with median survival is less than 12 months, especially among
age 60 years and above, not received radiation and glioblastoma grade IV. This finding
suggests that patients HGG treated at other centre hospitals will have poorer prognosis in
Indonesia. Another reason why in older people or over 60 years old the risk of dying is
almost 4 times but in statistically not significant ? because regrading from this journal,
patients who received sort of treatment like radiation were having relatively better
survival compared those did not received radiation.
