I have weekly data on the last year of life of a sample of decedents (all die in week 52, which differs in calendar time). Each week, patients may experience 4 different types of treatments (T1, T2, T3 AND T4). I am interested in evaluating whether the treatments/ cumulative treatments receiving in the last year of life (cum_T1, cum_T2, cum_T3 and cum_T4) predict deaths. Total_T is the cumulative sum of all prior treatments in any given week. Here is my dataex:
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input long case_number byte week float yw byte(T1 T2 T3) float(T4 cum_T1 cum_T2 cum_T3 cum_T4 total_T dead) byte age float male byte(White Black Hispanic) 100064 1 2550 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 2 2551 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 3 2552 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 4 2553 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 5 2554 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 6 2555 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 7 2556 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 8 2557 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 9 2558 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 10 2559 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 11 2560 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 12 2561 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 13 2562 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 14 2563 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 15 2564 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 16 2565 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 17 2566 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 18 2567 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 19 2568 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 20 2569 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 21 2570 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 22 2571 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 23 2572 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 24 2573 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 25 2574 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 26 2575 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 27 2576 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 28 2577 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 29 2578 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 30 2579 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 31 2580 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 32 2581 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 33 2582 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 34 2583 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 35 2584 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 36 2585 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 37 2586 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 38 2587 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 39 2588 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 40 2589 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 41 2590 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 42 2591 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 43 2592 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 44 2593 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 45 2594 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 46 2595 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 47 2596 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 48 2597 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 49 2598 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 50 2599 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 51 2600 0 0 0 0 0 0 0 0 0 0 20 1 1 0 0 100064 52 2601 0 0 0 0 0 0 0 0 0 1 20 1 1 0 0 100076 1 2551 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 2 2552 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 3 2553 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 4 2554 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 5 2555 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 6 2556 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 7 2557 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 8 2558 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 9 2559 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 10 2560 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 11 2561 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 12 2562 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 13 2563 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 14 2564 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 15 2565 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 16 2566 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 17 2567 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 18 2568 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 19 2569 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 20 2570 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 21 2571 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 22 2572 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 23 2573 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 24 2574 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 25 2575 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 26 2576 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 27 2577 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 28 2578 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 29 2579 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 30 2580 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 31 2581 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 32 2582 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 33 2583 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 34 2584 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 35 2585 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 36 2586 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 37 2587 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 38 2588 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 39 2589 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 40 2590 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 41 2591 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 42 2592 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 43 2593 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 44 2594 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 45 2595 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 46 2596 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 47 2597 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 100076 48 2598 0 0 0 0 0 0 0 0 0 0 42 1 0 1 0 end format %tw yw
Code:
. xtset case_number
panel variable: case_number (balanced)
. stset week, failure(dead==1)
failure event: dead == 1
obs. time interval: (0, week]
exit on or before: failure
------------------------------------------------------------------------------
120,796 total observations
0 exclusions
------------------------------------------------------------------------------
120,796 observations remaining, representing
2,323 failures in single-record/single-failure data
3,201,094 total analysis time at risk and under observation
at risk from t = 0
earliest observed entry t = 0
last observed exit t = 52
.
. eststo r1: xtstreg c.total_T male Black Hispanic Other age c.yw, distribution(exponential) vce(cluster case_number)
> nolog
failure _d: dead == 1
analysis time _t: week
Random-effects exponential PH regression Number of obs = 120,796
Group variable: case_number Number of groups = 2,323
Obs per group:
min = 52
avg = 52.0
max = 52
Integration method: mvaghermite Integration pts. = 12
Wald chi2(7) = 682.67
Log pseudolikelihood = -18974.339 Prob > chi2 = 0.0000
(Std. Err. adjusted for 2,323 clusters in case_number)
------------------------------------------------------------------------------
| Robust
_t | Haz. Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
total_T | 1.103723 .0111509 9.77 0.000 1.082083 1.125796
male | 1.050707 .0226684 2.29 0.022 1.007204 1.09609
Black | .9671728 .0219859 -1.47 0.142 .9250271 1.011239
Hispanic | 1.038662 .0292476 1.35 0.178 .9828912 1.097598
Other | .9710166 .0300546 -0.95 0.342 .9138618 1.031746
age | .9992546 .0005942 -1.25 0.210 .9980906 1.00042
yw | 1.001171 .0000688 17.04 0.000 1.001036 1.001305
_cons | .0000221 4.32e-06 -54.88 0.000 .0000151 .0000324
-------------+----------------------------------------------------------------
/sigma2_u | 1.27e-32 4.69e-34 1.18e-32 1.36e-32
------------------------------------------------------------------------------
Note: Estimates are transformed only in the first equation.
Note: _cons estimates baseline hazard (conditional on zero random effects).
. stcurve, survival at1(male=0) at2(male=1) lpattern(solid dash) lcolor(cranberry ebblue) title("Estimated Survival by
> Gender")
(option marginal assumed)1-- is this the appropriate model given all individuals died in week 52? I apologize if this is not the right place to ask this statistical (rather than Stata) question, but will be very grateful for any advice.
2-- Is the interpretation the above estimate of 1.103723 for total_t that one additional treatment increases the hazard of death by 10%?
Thank you for your time. Ever gratefully,
Sumedha.
Hello everyone, I want to say a special thanks to Dr OGU. for helping me get cured from herpes virus 2019 , I contacted him base on the testimonies I saw about him on the internet I was diagnosed of HERPES Virus i have tried all I can to get cured but all to know avail, until i saw a post in a health forum about a herbal man who prepare herbal medication to cure all kind of diseases including HERPES virus, at first i doubted if it was real but decided to give it a try I was cured by his herbal medicine and natural herbs, kindly contact him today through his email: drogugusolutionhome@gmail.com or text/call: +1 (719) 629 0982
ReplyDeleteHe’s waiting to help you.
HE ALSO SPECIALIZE IN THE FOLLOWING Illness;
If you have any sickness like : H I V/AIDS , CANCER , HERPES HSV 1 or 2 , GENITAL WARTS, Yeast-infection's (HPV), and Hepatitis A, B. Stroke, Chlamydia, Genital herpes, Alzheimer’s, Trichomoniasis , Tuberculosis, CAD, Gonorrhea, Epilepsy, and Syphilis.