Regresion de cox

TRABAJO ENCARGADO DE SUPERVIVENCIA Y CONFIABILIDAD
REGRESIÓN DE COX

1er paso
Ajustamos todos los modelos con una sola covariable. Luego, incluimos todas las covariables que fueron significativas a un nivel [pic].

Realizamos el modelo de regresión Cox para cada una de las variables:

Variable Race
[pic]
g(X´β)=exp(X´ β)=exp(β1X1)

> modelo1 summary(modelo1)
Call:
coxph(formula =Surv(duration, delta) ~ race, data = bfeed)

n= 927
coef exp(coef) se(coef) z Pr(>|z|)
race 0.12515 1.13331 0.04447 2.814 0.00489 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

exp(coef) exp(-coef) lower .95 upper .95
race 1.133 0.8824 1.039 1.237

Rsquare= 0.008 (max possible= 1 )
Likelihood ratio test= 7.6 on 1 df,p=0.005828
Wald test = 7.92 on 1 df, p=0.004894
Score (logrank) test = 7.95 on 1 df, p=0.004816

Notamos que la variable race resulta ser signifcativa

[pic]

Con un α=0.1 de significancia existe suficiente evidencia estadística para afirmar que la variable race contribuye significativamente al modelo.

Variable poverty
[pic]
g(X´β)=exp(X´ β)=exp(β2X2)

> modelo2summary(modelo2)
Call:
coxph(formula = Surv(duration, delta) ~ poverty, data = bfeed)

n= 927

coef exp(coef) se(coef) z Pr(>|z|)
poverty -0.07584 0.92696 0.08618 -0.88 0.379

exp(coef) exp(-coef) lower .95 upper .95
poverty 0.927 1.079 0.7829 1.098

Rsquare= 0.001 (max possible= 1 )
Likelihood ratio test= 0.79 on 1 df, p=0.3752
Wald test= 0.77 on 1 df, p=0.3788
Score (logrank) test = 0.77 on 1 df, p=0.3787

[pic]

Con un α=0.1 de significancia no existe suficiente evidencia estadística para afirmar que la variable poverty contribuye significativamente al modelo.

Variable smoke
[pic]
g(X´β)=exp(X´ β)=exp(β3X3)

> modelo3 summary(modelo3)
Call:
coxph(formula = Surv(duration, delta) ~ smoke, data =bfeed)

n= 927

coef exp(coef) se(coef) z Pr(>|z|)
smoke 0.22702 1.25486 0.07374 3.079 0.00208 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

exp(coef) exp(-coef) lower .95 upper .95
smoke 1.255 0.7969 1.086 1.45

Rsquare= 0.01 (max possible= 1 )
Likelihood ratio test= 9.18 on 1 df, p=0.002447
Wald test =9.48 on 1 df, p=0.00208
Score (logrank) test = 9.52 on 1 df, p=0.002035

[pic]

A un nivel de significación del 10% existe suficiente evidencia estadística para afirmar que la variable smoke contribuye significativamente al modelo.

Variable alcohol
[pic]
g(X´β)=exp(X´ β)=exp(β4X4)

> modelo4 summary(modelo4)
Call:
coxph(formula = Surv(duration, delta) ~ alcohol, data = bfeed)n= 927

coef exp(coef) se(coef) z Pr(>|z|)
alcohol 0.1657 1.1803 0.1203 1.378 0.168

exp(coef) exp(-coef) lower .95 upper .95
alcohol 1.180 0.8473 0.9324 1.494

Rsquare= 0.002 (max possible= 1 )
Likelihood ratio test= 1.82 on 1 df, p=0.1779
Wald test = 1.9 on 1 df, p=0.1682
Score (logrank) test = 1.9 on 1 df,p=0.1677

[pic]

A un nivel de significación del 10% no existe suficiente evidencia estadística para afirmar que la variable alcohol contribuye significativamente al modelo.

Variable agemth
[pic]
g(X´β)=exp(X´ β)=exp(β5X5)

> modelo5 summary(modelo5)
Call:
coxph(formula = Surv(duration, delta) ~ agemth, data = bfeed)

n= 927

coef exp(coef) se(coef) z Pr(>|z|)agemth -0.006291 0.993729 0.013145 -0.479 0.632

exp(coef) exp(-coef) lower .95 upper .95
agemth 0.9937 1.006 0.9685 1.020

Rsquare= 0 (max possible= 1 )
Likelihood ratio test= 0.23 on 1 df, p=0.6321
Wald test = 0.23 on 1 df, p=0.6322
Score (logrank) test = 0.23 on 1 df, p=0.6322

[pic]

A un nivel de significación del 10% no existe...