Vous êtes sur la page 1sur 13

Answer to Q #2(a)

The FREQ Procedure

Table of severity by TBI

Frequency
severity

Percent

TBI
0

Row Pct
0

Col Pct

Total

53

90

143

26.90

45.69

72.59

37.06

62.94

89.83

65.22

48

54

3.05

24.37

27.41

11.11

88.89

10.17

34.78

59

138

197

29.95

70.05

100.00

Total

Statistics for Table of severity by TBI

Odds Ratio and Relative Risks


Statistic

Value

95% Confidence Limits

Odds Ratio

4.7111

1.8887

11.7511

Relative Risk (Column 1)

3.3357

1.5229

7.3061

Relative Risk (Column 2)

0.7080

0.6050

0.8286

Odds Ratio
Odds Ratio

4.7111

Asymptotic Conf Limits


95% Lower Conf Limit

1.8887

95% Upper Conf Limit

11.7511

Exact Conf Limits

Odds Ratio
95% Lower Conf Limit

1.8292

95% Upper Conf Limit

14.2925
Sample Size = 197

From the table severity and TBI, it is seen that the value of odds ratio is 4.7111 which
indicates indicates the strength of the association between severity and TBI are very
strong..Total sample size 197 and 95% confidence Interval lies the value between
1.8887 and 11.7511.Since the confidence interval does not include the NULL value
1, there is significant association between severity and TBI.

The FREQ Procedure

Table of H_status by TBI

Frequency
H_status

Percent

TBI
0

Row Pct
0

Col Pct

Total

Total

16

62

78

12.12

46.97

59.09

20.51

79.49

72.73

56.36

48

54

4.55

36.36

40.91

11.11

88.89

27.27

43.64

22

110

132

16.67

83.33

100.00

Statistics for Table of H_status by TBI

Odds Ratio and Relative Risks


Statistic

Value

95% Confidence Limits

Odds Ratio

2.0645

0.7511

5.6746

Relative Risk (Column 1)

1.8462

0.7721

4.4143

Odds Ratio and Relative Risks


Statistic

Value

Relative Risk (Column 2)

95% Confidence Limits

0.8942

0.7720

1.0358

Odds Ratio
Odds Ratio

2.0645

Asymptotic Conf Limits


95% Lower Conf Limit

0.7511

95% Upper Conf Limit

5.6746

Exact Conf Limits


95% Lower Conf Limit

0.6968

95% Upper Conf Limit

6.9091
Sample Size = 132

From the table H_status and TBI, it is seen that the value of odds ratio is 2.0645
which indicates an increased odds or risk. Total sample size 132 and 95%
confidence Interval lies the value between 0.7511 and 5.6746.Since the confidence
interval include the NULL value 1, so there is no significant association between
H_status and TBI.

The FREQ Procedure

Table of severity by H_status

Frequency
Percent

severity

H_status
0

Row Pct
Col Pct

Total

114

90

204

36.31

28.66

64.97

55.88

44.12

64.77

65.22

62

48

110

19.75

15.29

35.03

Table of severity by H_status


severity

H_status
0

Total

56.36

43.64

35.23

34.78

176

138

314

56.05

43.95

100.00

Total

Statistics for Table of severity by H_status

Odds Ratio and Relative Risks


Statistic

Value

95% Confidence Limits

Odds Ratio

0.9806

0.6146

1.5648

Relative Risk (Column 1)

0.9915

0.8079

1.2167

Relative Risk (Column 2)

1.0110

0.7775

1.3147

Odds Ratio
Odds Ratio

0.9806

Asymptotic Conf Limits


95% Lower Conf Limit

0.6146

95% Upper Conf Limit

1.5648

Exact Conf Limits


95% Lower Conf Limit

0.5973

95% Upper Conf Limit

1.6058
Sample Size = 314

From the table H_status and severity, it is seen that the value of odds ratio is 0.9806
which is very close to 1 which indicates no strength of association. Total sample size
314 and 95% confidence Interval lies the value between 0.6146 and 1.5648.Since
the confidence interval includes the NULL value 1, so there is no significant
association between H_status and severity.

b) For Severity:

Note: 1 observation with invalid time, censoring, or strata values was deleted.

Test of Equality over Strata


Test

Chi-Square

DF

Pr >
Chi-Square

Log-Rank

63.7654

<.0001

Wilcoxon

54.4177

<.0001

-2Log(LR)

49.2647

<.0001

The logrank test suggests that the two severity groups has significantly
different. The Kaplan Meier survival curves are not so difference at
beginning stage but after the difference is more.

FOR TBI:

Test of Equality over Strata


Test

Chi-Square

DF

Pr >
Chi-Square

Log-Rank

1.4822

0.2234

Wilcoxon

2.1271

0.1447

-2Log(LR)

3.2274

0.0724

The logrank test suggests that the two TBI groups has no significantly
different(P value=0.2234>0.05). The Kaplan Meier survival curves shows
a little difference between two groups.

FOR H_status

Test of Equality over Strata


Chi-Square

DF

Pr >
Chi-Square

1.9735

0.1601

0.1842

0.6678

3.5355

0.0601

The log rank test suggests that the two TBI groups has no significantly
different (P value=0.1601>0.05). The Kaplan Meier survival curves shows
that initial stage the general health status is good but drops it very
quickly.

C)
Model Fit Statistics
Criterion

Without
Covariates

With
Covariates

-2 LOG L

1203.270

1203.129

AIC

1203.270

1205.129

SBC

1203.270

1207.857

Testing Global Null Hypothesis: BETA=0


Test

Chi-Square

DF

Pr > ChiSq

Likelihood Ratio

0.1410

0.7073

Score

0.1426

0.7057

Wald

0.1426

0.7057

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

age

0.00204

0.00539

0.1426

0.7057

1.002

95% Hazard
Ratio
Confidence
Limits
0.992

1.013

From maximum likelihood estimates it is seen that p value is larger than


0.05 and 95% CI includes the null value 1. So there is no association
between age and survival time. They are statistically insignificant.

d)

Analysis of Maximum Likelihood Estimates


Paramete
r

age

DF Parameter Standard
Estimate
Error

0.00204

0.00539

ChiSquare

0.1426

Pr > ChiS Hazard


q
Ratio

0.7057

95% Hazard
Ratio
Confidence
Limits

1.002

0.992

1.013

Analysis of Maximum Likelihood Estimates


Paramete
r

severity

DF Parameter
Estimate

1.51092

Standar
d
Error

Chi- Pr > ChiSq


Square

0.20446 54.6103

<.0001

Hazar
d
Ratio

95% Hazard
Ratio
Confidence
Limits

4.531

3.035

6.764

Analysis of Maximum Likelihood Estimates


Paramete
r

DF Parameter
Estimate

severity

1.57397

age

-0.00644

Standar
d
Error

Chi- Pr > ChiSq


Square

Hazar
d
Ratio

95% Hazard
Ratio
Confidence
Limits

0.21190 55.1742

<.0001

4.826

3.186

7.310

0.00556

0.2470

0.994

0.983

1.004

1.3404

Analysis of Maximum Likelihood Estimates


Parameter

D Paramete Standar
F
r
d
Estimate
Error

age*severit
y

0.02123

Chi- Pr > ChiS Hazar


Squar
q
d
e
Ratio

0.00347 37.317
6

EFFECT MODIFICATION:

<.0001

95%
Label
Hazard
Ratio
Confidenc
e
Limits
.

. age *
severit
y

From maximum likelihood estimate it is seen that the P value(.0001) is


very small which is less than 0.05. So there is a significant effect
modification between age and severity.

Confounding:
When considered age alone and when considering with severity, the P
value of age is changed. So there is confounding on the effects of age
due to severity.

e)
Analysis of Maximum Likelihood Estimates
Paramete
r

D
F

Paramete
r
Estimate

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

TBI

0.26540

0.21884

1.4708

0.2252

1.304

95% Hazard
Ratio
Confidence
Limits
0.849

2.002

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

severity

1.51092

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

0.20446 54.6103

<.0001

4.531

95% Hazard
Ratio
Confidence
Limits
3.035

6.764

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

95% Hazard
Ratio
Confidence
Limits

TBI

0.11562

0.22122

0.2731

0.6012

1.123

0.728

1.732

severity

1.49895

0.20581 53.0431

<.0001

4.477

2.991

6.701

Analysis of Maximum Likelihood Estimates


Parameter

D Paramete Standar
Chi- Pr > ChiS Hazar
F
r
d Squar
q
d
Estimate
Error
e
Ratio

TBI*severit
y

1.61851

0.20766 60.744
2

<.0001

95%
Label
Hazard
Ratio
Confidenc
e
Limits
.

. TBI *
severit
y

EFFECT MODIFICATION:
From maximum likelihood estimate it is seen that the P value (.0001) is
very small which is less than 0.05. So there is a significant effect
modification between TBI and severity.

Confounding:
When considered TBI alone and when considering with severity, the P
value of TBI is changed. So there is confounding on the effects of TBI due
to severity.

f)

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

H_status

0.26758

0.19120

1.9586

0.1617

1.307

95% Hazard
Ratio
Confidence
Limits
0.898

1.901

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

severity

1.51092

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

0.20446 54.6103

<.0001

4.531

95% Hazard
Ratio
Confidence
Limits
3.035

6.764

Analysis of Maximum Likelihood Estimates


Paramete
r

D
F

Paramete
r
Estimate

Standar
d
Error

ChiSquare

Pr > ChiS
q

Hazar
d
Ratio

95% Hazard
Ratio
Confidence
Limits

H_status

0.31282

0.19239

2.6439

0.1040

1.367

0.938

1.994

severity

1.52641

0.20540 55.2253

<.0001

4.602

3.077

6.883

Analysis of Maximum Likelihood Estimates


Parameter

D Paramet
F
er
Estimate

H_status*sever
ity

Standa Chird Squar


Error
e

0.85967 0.27376

EFFECT MODIFICATION:

9.861
1

Pr > Chi Hazar


Sq
d
Ratio

0.0017

95%
Label
Hazard
Ratio
Confiden
ce
Limits
.

. H_stat
us *
severit
y

From maximum likelihood estimate it is seen that the P value(.0017) is


very small which is less than 0.05. So there is a significant effect
modification between H_status and severity.

Confounding:
When considered H_status alone and when considering with severity , the
P value of H_status is slightly changed. So there is little confounding on
the effects of H_status due to severity.

Vous aimerez peut-être aussi