/
R.r
451 lines (372 loc) · 19.7 KB
/
R.r
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
## Author: Bruce Swihart
## Date: 2013/02
## Warranty: None
## This .R file accompanies the paper "Modeling sleep fragmentation in populations of sleep hypnograms".
## 5 STATE / 20 Transition Types: Poisson format / loglinear modeling;
## with one possible bootstrapping procedure coding block;
## analysis of 5598
library("geepack")
library("gee")
library("Matrix")
## matlab library for tic()/toc() timing of function calls.
library("matlab")
library("survival")
## allows us to make nice tables of the linear combinations that ultimately are the effects of interest
linear.comb <- function(fit, cntrst.mtrx1){
hold <- cbind(exp(esticon(fit, cntrst.mtrx1, joint.test=FALSE)[,c("Estimate", "Lower", "Upper")]),
esticon(fit, cntrst.mtrx1, joint.test=FALSE)[,c("Pr(>|X^2|)")])
hold <- round(hold,2)
colnames(hold) <- c("RR est", "95% L", "95% U", "p-value")
hold
}
## read in data
d.5598 <- read.csv("LLfull_allgroups_agesexracesmoking.csv", TRUE)
## the order I want these in the table; models, etc
## I put the line breaks where the 3-state analogues would be.
shifttype <- c("1R", "2R", "SR", ## 3-state analogue: NR
"1W", "2W", "SW", ## 3-state analogue: NW
"R1", "R2", "RS", ## 3-state analogue: RN
"RW", ## 3-state analogue: RW
"W1", "W2", "WS", ## 3-state analogue: WN
"WR", ## 3-state analogue: WR
"S1", "S2", "12", "1S", "21", "2S") ## 3-state analogue: --; these are intra-NREM.
## establish order of types
d.5598$shift <- factor(d.5598$shift,
levels=c("1R", "2R", "SR", "1W", "2W", "SW", "R1", "R2", "RS", "RW", "W1", "W2", "WS", "WR", "S1", "S2", "12", "1S", "21", "2S"))
## waves argument for geeglm(); reinforces data row order.
f.wav <- d.5598$type
tic()
## fit with geeglm()
f.off.y <- geeglm(counts~ offset(I(log(tar))) + I(factor(race)) + sex + I(smokstatus) + age + shift*grouplabel,
id=pptid,
data=d.5598,
family="poisson",
corstr="independence",
scale.fix=TRUE,
wave=f.wav,
control=geese.control(epsilon=1e-4, maxit=as.integer(10), trace=TRUE, scale.fix=TRUE))
toc()
## check if error-free
f.off.y$geese["error"]
summary(f.off.y)
## construct contrast matrix, picking up main effects for group-label
## and corresponding interaction term for shift*group-label For 4 SDB
## severity groups and 20 transitions, we have 3*20 = 60 estimates of
## interest. The 3 non-referent groups each have 20 transition-type
## specific effects. To construct an overall contrast matrix
## ("contog"), we first make matrices for each group ("con0515",
## "con1530", "con30pp", for the mild, moderate and severe SDB groups,
## respectively). For each group, the first row has just one 1 to
## pick up the main effect for the group (the interaction term is zero
## because the first transition type is the referent), stacked on top
## of the remaining 19 rows which need to pick up the same main effect
## AND a corresponding interaction term.
p0 <- 28 ## number of leading zeros in contrast matrix.
con0515<-rbind(c(rep(0,p0),1,0,0,rep(0,19),rep(0,19),rep(0,19)),
cbind(matrix(0,ncol=p0,nrow=19),rep(1,19),rep(0,19), rep(0,19), diag(19)*1, diag(19)*0, diag(19)*0))
con1530<-rbind(c(rep(0,p0),0,1,0,rep(0,19),rep(0,19),rep(0,19)),
cbind(matrix(0,ncol=p0,nrow=19),rep(0,19),rep(1,19), rep(0,19), diag(19)*0, diag(19)*1, diag(19)*0))
con30pp<-rbind(c(rep(0,p0),0,0,1,rep(0,19),rep(0,19),rep(0,19)),
cbind(matrix(0,ncol=p0,nrow=19),rep(0,19),rep(0,19), rep(1,19), diag(19)*0, diag(19)*0, diag(19)*1))
contog<-rbind(con0515, con1530, con30pp) ## stack the group specific contrast matrices
results<-linear.comb(f.off.y, contog) ## use our function to linearly combine; exp()
cbind(shifttype, results) ## long table
cbind(shifttype, results[1:20,],results[21:40,],results[41:60,]) ## wide (utilize common rows)
## bootstrapping can be done to "correct" confidence intervals. Doing so can be computationally intensive.
## This is my BEE function: "Bootstrap Estimating Equations"
bee <- function(gee.fit, data, id, B=1000) {
mat <- matrix(nrow=B, ncol=length(coef(gee.fit)))
unique.ids <- unique(id)
for (b in 1:B) {
# Boostrap ids
ids.b <- sample(unique.ids, replace=TRUE)
dat.b <- data[as.vector(sapply(ids.b, function(x) which(id==x))),]
dat.b$pptid <- id
# Fit GEE and get coefficients
mat[b,] <- coef(update(gee.fit, data=dat.b, control=geese.control(maxit=as.integer(1))))
}
mat
}
## took 168 sec for 2 reps
## estimating at 9.33 hours for 1000 reps (actually took: 79143 seconds = 21.98 hours). Ouch.
tic()
coeff<-bee(f.off.y, d.5598, d.5598$pptid, B=1000)
toc()
## linearly combine and exp() as after the original geeglm call
ests <- exp(contog%*%coeff[1,])
cbind(shifttype, ests[1:20], ests[21:40], ests[41:60])
## Gives in the 60 rows, 1000 samples of my estimate(s) in the columns
hold<-exp(contog%*%t(coeff))
apply(hold, 1, function(W){quantile(W, c(.025,.975))})
cbind(shifttype,t(round(apply(hold, 1, function(W){quantile(W, c(.025,.975))}),2)))
write.csv(coeff, "geeIND_boot1000_coeff.csv", row.names=FALSE)
write.csv(hold, "geeIND_boot1000_rr.csv", row.names=FALSE)
## 3 STATE / 6 Transition Types: Poisson format / loglinear modeling;
d.5598 <- read.csv("LLfull_allgroups_agesexracesmoking_0306.csv", TRUE)
##quick calculation for example in paper: 671 has 0 RN. Whereas 6269 have 0 RS.
table(d.5598$count[d.5598$shift=="RN"])
table(d.5598$count[d.5598$shift=="RN"]<1)
## the order I want these in the table; models, etc
shifttype <- c("NR", "NW", "RN", "RW", "WN", "WR")
## establish order of types
d.5598$shift <- factor(d.5598$shift,
levels=c("NR", "NW", "RN", "RW", "WN", "WR"))
## waves argument
f.wav <- d.5598$type
tic()
f.off.y <- geeglm(counts~ offset(I(log(tar))) + I(factor(race)) + sex + I(smokstatus) + age + shift*grouplabel,
id=pptid,
data=d.5598,
family="poisson",
corstr="independence",
scale.fix=TRUE,
wave=f.wav,
control=geese.control(epsilon=1e-4, maxit=as.integer(10), trace=TRUE, scale.fix=TRUE))
toc()
f.off.y$geese["error"]
summary(f.off.y)
p0 <- 14 ##28 ##32+4
con0515<-rbind(c(rep(0,p0),1,0,0,rep(0,5),rep(0,5),rep(0,5)), cbind(matrix(0,ncol=p0,nrow=5),rep(1,5),rep(0,5), rep(0,5), diag(5)*1, diag(5)*0, diag(5)*0))
con1530<-rbind(c(rep(0,p0),0,1,0,rep(0,5),rep(0,5),rep(0,5)), cbind(matrix(0,ncol=p0,nrow=5),rep(0,5),rep(1,5), rep(0,5), diag(5)*0, diag(5)*1, diag(5)*0))
con30pp<-rbind(c(rep(0,p0),0,0,1,rep(0,5),rep(0,5),rep(0,5)), cbind(matrix(0,ncol=p0,nrow=5),rep(0,5),rep(0,5), rep(1,5), diag(5)*0, diag(5)*0, diag(5)*1))
contog<-rbind(con0515, con1530, con30pp)
results<-linear.comb(f.off.y, contog)
cbind(shifttype, results) ## long table
cbind(shifttype, results[1:6,],results[7:12,],results[13:18,]) ## wide (utlze common rows)
## bootstrap ...
# This is my BEE function
bee <- function(gee.fit, data, id, B=1000) {
mat <- matrix(nrow=B, ncol=length(coef(gee.fit)))
unique.ids <- unique(id)
for (b in 1:B) {
# Boostrap ids
ids.b <- sample(unique.ids, replace=TRUE)
dat.b <- data[as.vector(sapply(ids.b, function(x) which(id==x))),]
dat.b$pptid <- id
# Fit GEE and get coefficients
mat[b,] <- coef(update(gee.fit, data=dat.b, control=geese.control(maxit=as.integer(1))))
}
mat
}
tic()
coeff<-bee(f.off.y, d.5598, d.5598$pptid, B=1000)
toc()
%% tinker
ests <- exp(contog%*%coeff[1,])
cbind(shifttype, ests[1:20], ests[21:40], ests[41:60])
exp(contog%*%t(coeff[1:2,]))
hold<-exp(contog%*%t(coeff))
apply(hold, 1, function(W){quantile(W, c(.025,.975))})
cbind(shifttype,t(round(apply(hold, 1, function(W){quantile(W, c(.025,.975))}),2)))
write.csv(coeff, "geeIND_boot1000_coeff_0306.csv", row.names=FALSE)
write.csv(hold, "geeIND_boot1000_rr_0306.csv", row.names=FALSE)
## Survival analysis only has interaction terms, so linearly combining
## as in the Poisson examples above is not required. The naming of the
## predictors gX.tYY.ZZ is for the X group (1-mild, 2-moderate, 3-severe SDB),
## YY transition-type (1-20) (ZZ was the label version of YY).
## The line breaks occur where the analogue sets would be and a space between groups.
## The interaction terms were preprocessed; that is they are found in the accompanying datasets ready-to-go.
## 5 STATE / 20 Transition Types: multistate survival;
d.5598 <- read.csv("SAfull_allgroups_agesexracesmoking.csv", TRUE)
## "SA 5/20";
tic()
SA520 <- coxph(Surv(time, status) ~
g1t01.1R + g1t02.2R + g1t03.SR
+ g1t04.1W + g1t05.2W + g1t06.SW
+ g1t07.R1 + g1t08.R2 + g1t09.RS
+ g1t10.RW
+ g1t11.W1 + g1t12.W2 + g1t13.WS
+ g1t14.WR
+ g1t15.S1 + g1t16.S2 + g1t17.12 + g1t18.1S + g1t19.21 + g1t20.2S
+ g2t01.1R + g2t02.2R + g2t03.SR
+ g2t04.1W + g2t05.2W + g2t06.SW
+ g2t07.R1 + g2t08.R2 + g2t09.RS
+ g2t10.RW
+ g2t11.W1 + g2t12.W2 + g2t13.WS
+ g2t14.WR
+ g2t15.S1 + g2t16.S2 + g2t17.12 + g2t18.1S + g2t19.21 + g2t20.2S
+ g3t01.1R + g3t02.2R + g3t03.SR
+ g3t04.1W + g3t05.2W + g3t06.SW
+ g3t07.R1 + g3t08.R2 + g3t09.RS
+ g3t10.RW
+ g3t11.W1 + g3t12.W2 + g3t13.WS
+ g3t14.WR
+ g3t15.S1 + g3t16.S2 + g3t17.12 + g3t18.1S + g3t19.21 + g3t20.2S
+ age + sex + I(factor(race)) + smokstatus
+ strata(type) + cluster(pptid),
data=d.5598)
toc()
## output from SA520:
## Warning messages:
## 1: In matrix(0, n, nvar) :
## Reached total allocation of 16296Mb: see help(memory.size)
## 2: In matrix(0, n, nvar) :
## Reached total allocation of 16296Mb: see help(memory.size)
## > elapsed time is 25881.810000 seconds
## > 25881/3600
## [1] 7.189167 ## took 7 hours to run on 64 bit machine
## > SA520
## Call:
## coxph(formula = Surv(time, status) ~ g1t01.1R + g1t02.2R + g1t03.SR +
## g1t04.1W + g1t05.2W + g1t06.SW + g1t07.R1 + g1t08.R2 + g1t09.RS +
## g1t10.RW + g1t11.W1 + g1t12.W2 + g1t13.WS + g1t14.WR + g1t15.S1 +
## g1t16.S2 + g1t17.12 + g1t18.1S + g1t19.21 + g1t20.2S + g2t01.1R +
## g2t02.2R + g2t03.SR + g2t04.1W + g2t05.2W + g2t06.SW + g2t07.R1 +
## g2t08.R2 + g2t09.RS + g2t10.RW + g2t11.W1 + g2t12.W2 + g2t13.WS +
## g2t14.WR + g2t15.S1 + g2t16.S2 + g2t17.12 + g2t18.1S + g2t19.21 +
## g2t20.2S + g3t01.1R + g3t02.2R + g3t03.SR + g3t04.1W + g3t05.2W +
## g3t06.SW + g3t07.R1 + g3t08.R2 + g3t09.RS + g3t10.RW + g3t11.W1 +
## g3t12.W2 + g3t13.WS + g3t14.WR + g3t15.S1 + g3t16.S2 + g3t17.12 +
## g3t18.1S + g3t19.21 + g3t20.2S + age + sex + I(factor(race)) +
## smokstatus + strata(type) + cluster(pptid), data = d.5598)
## coef exp(coef) se(coef) robust se z p
## g1t01.1R 0.051333 1.053 0.025571 0.047676 1.076704 2.8e-01
## g1t02.2R -0.008662 0.991 0.013936 0.015198 -0.569921 5.7e-01
## g1t03.SR 0.069485 1.072 0.067863 0.073931 0.939854 3.5e-01
## g1t04.1W 0.185714 1.204 0.014154 0.024106 7.703975 1.3e-14
## g1t05.2W 0.095205 1.100 0.007919 0.014662 6.493418 8.4e-11
## g1t06.SW 0.034416 1.035 0.022924 0.029099 1.182730 2.4e-01
## g1t07.R1 0.227274 1.255 0.028898 0.056279 4.038363 5.4e-05
## g1t08.R2 0.002209 1.002 0.021905 0.030290 0.072930 9.4e-01
## g1t09.RS -0.089109 0.915 0.258597 0.324726 -0.274414 7.8e-01
## g1t10.RW 0.197191 1.218 0.012814 0.022395 8.805263 0.0e+00
## g1t11.W1 0.000618 1.001 0.007798 0.015284 0.040425 9.7e-01
## g1t12.W2 -0.016763 0.983 0.010507 0.024466 -0.685129 4.9e-01
## g1t13.WS 0.000122 1.000 0.077133 0.134980 0.000905 1.0e+00
## g1t14.WR 0.167531 1.182 0.019698 0.037370 4.483068 7.4e-06
## g1t15.S1 -0.096194 0.908 0.260157 0.261271 -0.368177 7.1e-01
## g1t16.S2 0.028888 1.029 0.006712 0.011333 2.549086 1.1e-02
## g1t17.12 -0.083974 0.919 0.009074 0.016778 -5.004943 5.6e-07
## g1t18.1S -0.251288 0.778 0.232362 0.239032 -1.051274 2.9e-01
## g1t19.21 0.211747 1.236 0.064400 0.080785 2.621104 8.8e-03
## g1t20.2S -0.080072 0.923 0.006446 0.017631 -4.541512 5.6e-06
## g2t01.1R 0.084791 1.088 0.034605 0.067002 1.265496 2.1e-01
## g2t02.2R -0.040070 0.961 0.020062 0.023330 -1.717539 8.6e-02
## g2t03.SR -0.064428 0.938 0.105998 0.129175 -0.498767 6.2e-01
## g2t04.1W 0.310558 1.364 0.018358 0.033774 9.195062 0.0e+00
## g2t05.2W 0.206118 1.229 0.010684 0.022556 9.138093 0.0e+00
## g2t06.SW -0.051814 0.950 0.035067 0.041355 -1.252903 2.1e-01
## g2t07.R1 0.295024 1.343 0.040109 0.082452 3.578133 3.5e-04
## g2t08.R2 0.028918 1.029 0.031612 0.042723 0.676868 5.0e-01
## g2t09.RS 0.504464 1.656 0.296996 0.368477 1.369050 1.7e-01
## g2t10.RW 0.316460 1.372 0.017612 0.036406 8.692633 0.0e+00
## g2t11.W1 -0.011131 0.989 0.010680 0.021737 -0.512074 6.1e-01
## g2t12.W2 0.053191 1.055 0.013978 0.032770 1.623170 1.0e-01
## g2t13.WS -0.185127 0.831 0.112896 0.179751 -1.029905 3.0e-01
## g2t14.WR 0.198719 1.220 0.026087 0.052292 3.800175 1.4e-04
## g2t15.S1 0.059581 1.061 0.362156 0.363195 0.164046 8.7e-01
## g2t16.S2 0.031723 1.032 0.009898 0.016537 1.918269 5.5e-02
## g2t17.12 -0.142443 0.867 0.012824 0.024448 -5.826452 5.7e-09
## g2t18.1S -0.465896 0.628 0.357861 0.392251 -1.187750 2.3e-01
## g2t19.21 0.209726 1.233 0.089929 0.114378 1.833622 6.7e-02
## g2t20.2S -0.163465 0.849 0.009536 0.027096 -6.032872 1.6e-09
## g3t01.1R 0.250123 1.284 0.042165 0.095344 2.623382 8.7e-03
## g3t02.2R -0.219493 0.803 0.028474 0.031382 -6.994323 2.7e-12
## g3t03.SR 0.402372 1.495 0.125769 0.142208 2.829457 4.7e-03
## g3t04.1W 0.442225 1.556 0.022617 0.055627 7.949854 1.9e-15
## g3t05.2W 0.394735 1.484 0.012809 0.042642 9.256964 0.0e+00
## g3t06.SW -0.106631 0.899 0.053805 0.076468 -1.394450 1.6e-01
## g3t07.R1 0.763347 2.145 0.045443 0.108070 7.063433 1.6e-12
## g3t08.R2 0.064954 1.067 0.043853 0.055104 1.178757 2.4e-01
## g3t09.RS 1.053801 2.869 0.324010 0.558388 1.887218 5.9e-02
## g3t10.RW 0.230134 1.259 0.025348 0.050193 4.585019 4.5e-06
## g3t11.W1 -0.029422 0.971 0.013993 0.037796 -0.778451 4.4e-01
## g3t12.W2 0.346996 1.415 0.016023 0.048640 7.133894 9.8e-13
## g3t13.WS -0.086156 0.917 0.137445 0.231173 -0.372691 7.1e-01
## g3t14.WR -0.100069 0.905 0.037097 0.078893 -1.268409 2.0e-01
## g3t15.S1 0.121554 1.129 0.519840 0.513861 0.236550 8.1e-01
## g3t16.S2 0.077289 1.080 0.013962 0.021743 3.554692 3.8e-04
## g3t17.12 -0.183433 0.832 0.017211 0.027394 -6.696129 2.1e-11
## g3t18.1S -0.643352 0.526 0.516701 0.518080 -1.241800 2.1e-01
## g3t19.21 0.666314 1.947 0.096308 0.138849 4.798850 1.6e-06
## g3t20.2S -0.332604 0.717 0.013536 0.046203 -7.198731 6.1e-13
## age -0.000681 0.999 0.000119 0.000246 -2.771197 5.6e-03
## sex 0.001470 1.001 0.002586 0.005069 0.289959 7.7e-01
## I(factor(race))2 -0.016677 0.983 0.004977 0.009598 -1.737470 8.2e-02
## I(factor(race))3 -0.077522 0.925 0.004464 0.009624 -8.055465 7.8e-16
## I(factor(race))4 0.030283 1.031 0.009950 0.019409 1.560305 1.2e-01
## I(factor(race))5 -0.032723 0.968 0.006143 0.011640 -2.811127 4.9e-03
## smokstatusFormer 0.039114 1.040 0.004337 0.008545 4.577186 4.7e-06
## smokstatusNever 0.037947 1.039 0.004313 0.008455 4.488375 7.2e-06
## Likelihood ratio test=4883 on 68 df, p=0 n= 2716188, number of events= 677295
## 3 STATE / 6 Transition Types: multistate survival;
d.5598.36 <- read.csv("SAfull_allgroups_agesexracesmoking_0306.csv", TRUE)
## "SA 3/06";
tic()
SA306 <- coxph(Surv(time, status) ~
g1t01.NR
+ g1t02.NW
+ g1t03.RN
+ g1t04.RW
+ g1t05.WN
+ g1t06.WR
+ g2t01.NR
+ g2t02.NW
+ g2t03.RN
+ g2t04.RW
+ g2t05.WN
+ g2t06.WR
+ g3t01.NR
+ g3t02.NW
+ g3t03.RN
+ g3t04.RW
+ g3t05.WN
+ g3t06.WR
+ age + sex + I(factor(race)) + smokstatus
+ strata(type) + cluster(pptid),
data=d.5598.36)
toc()
## Bonus: time varying SA 5/20;
## takes a while to run: possibly double the previous SA520 fit
## 5 STATE / 20 Transition Types: multistate survival;
d.5598 <- read.csv("SAfull_allgroups_agesexracesmoking.csv", TRUE)
## "time-varying SA 5/20";
tic()
SA520.tt <- coxph(Surv(time, status) ~
g1t01.1R + g1t02.2R + g1t03.SR
+ g1t04.1W + g1t05.2W + g1t06.SW
+ g1t07.R1 + g1t08.R2 + g1t09.RS
+ g1t10.RW
+ g1t11.W1 + g1t12.W2 + g1t13.WS
+ g1t14.WR
+ g1t15.S1 + g1t16.S2 + g1t17.12 + g1t18.1S + g1t19.21 + g1t20.2S
+ g2t01.1R + g2t02.2R + g2t03.SR
+ g2t04.1W + g2t05.2W + g2t06.SW
+ g2t07.R1 + g2t08.R2 + g2t09.RS
+ g2t10.RW
+ g2t11.W1 + g2t12.W2 + g2t13.WS
+ g2t14.WR
+ g2t15.S1 + g2t16.S2 + g2t17.12 + g2t18.1S + g2t19.21 + g2t20.2S
+ g3t01.1R + g3t02.2R + g3t03.SR
+ g3t04.1W + g3t05.2W + g3t06.SW
+ g3t07.R1 + g3t08.R2 + g3t09.RS
+ g3t10.RW
+ g3t11.W1 + g3t12.W2 + g3t13.WS
+ g3t14.WR
+ g3t15.S1 + g3t16.S2 + g3t17.12 + g3t18.1S + g3t19.21 + g3t20.2S
+ tt(g1t01.1R) + tt(g1t02.2R) + tt(g1t03.SR)
+ tt(g1t04.1W) + tt(g1t05.2W) + tt(g1t06.SW)
+ tt(g1t07.R1) + tt(g1t08.R2) + tt(g1t09.RS)
+ tt(g1t10.RW)
+ tt(g1t11.W1) + tt(g1t12.W2) + tt(g1t13.WS)
+ tt(g1t14.WR)
+ tt(g1t15.S1) + tt(g1t16.S2) + tt(g1t17.12) + tt(g1t18.1S) + tt(g1t19.21) + tt(g1t20.2S)
+ tt(g2t01.1R) + tt(g2t02.2R) + tt(g2t03.SR)
+ tt(g2t04.1W) + tt(g2t05.2W) + tt(g2t06.SW)
+ tt(g2t07.R1) + tt(g2t08.R2) + tt(g2t09.RS)
+ tt(g2t10.RW)
+ tt(g2t11.W1) + tt(g2t12.W2) + tt(g2t13.WS)
+ tt(g2t14.WR
+ tt(g2t15.S1) + tt(g2t16.S2) + tt(g2t17.12) + tt(g2t18.1S) + tt(g2t19.21) + tt(g2t20.2S)
+ tt(g3t01.1R) + tt(g3t02.2R) + tt(g3t03.SR)
+ tt(g3t04.1W) + tt(g3t05.2W) + tt(g3t06.SW)
+ tt(g3t07.R1) + tt(g3t08.R2) + tt(g3t09.RS)
+ tt(g3t10.RW)
+ tt(g3t11.W1) + tt(g3t12.W2) + tt(g3t13.WS)
+ tt(g3t14.WR)
+ tt(g3t15.S1) + tt(g3t16.S2) + tt(g3t17.12) + tt(g3t18.1S) + tt(g3t19.21) + tt(g3t20.2S)
+ age + sex + I(factor(race)) + smokstatus
+ strata(type) + cluster(pptid),
data=d.5598,
tt=function(x,t, ...){x*log(t)})
toc()