In this study, 150 patients were divided by gender and age on the symmetrical bar chart of population characteristics shown in figure 1, and the corresponding average hospitalization days was calculated. It can be seen intuitively from the figure that the average hospitalization days of patients increases with the increase of age, and the average hospitalization days of patients aged over 60 is longer than that of patients aged under 40. Between the genders, the average hospitalization days were significantly longer for women (see FIG. 1 for details).
Table 1 demographic characteristic of patients
|
Gender
|
|
Age stratification
|
|
Male (n=84)
|
Female
(n=66)
|
Total
|
<60
(n=49)
|
≥60
(n=101)
|
Total
|
|
HOD
|
20.27±5.77
|
21.11±6.08
|
20.64±5.90
|
20.80±4.578
|
20.56±6.47
|
20.64±5.90
|
|
CRP
|
70.00±47.01
|
33.23±37.38
|
53.82±46.65
|
43.52±42.15
|
58.82±48.09
|
53.82±46.65
|
|
<3
|
3
|
7
|
10
|
6
|
4
|
10
|
|
3-100
|
54
|
53
|
107
|
35
|
72
|
107
|
|
>100
|
27
|
6
|
33
|
8
|
25
|
33
|
|
L
|
0.84±0.61
|
1.03±0.47
|
0.92±0.56
|
0.99±0.42
|
0.89±0.62
|
0.92±0.56
|
|
<1.1
|
66
|
43
|
109
|
30
|
79
|
109
|
|
1.1-3.2
|
16
|
23
|
39
|
19
|
20
|
39
|
|
>3.2
|
2
|
0
|
2
|
0
|
2
|
2
|
|
We divided the patients into two groups: the < 60 year old group and the ≥ 60 year old group. CRP and L were divided into two groups according to the normal reference value range (CRP: < 3, 3-100, > 100; l: 1.1-3.2, < 1.1, > 3.2). The corresponding mean value of L and CRP and 95% confidence interval were collected and calculated respectively, and error bar graph was drawn. In each group of CRP classification, the mean value of CRP in patients over 60 years old was higher than that in patients under 60 years old (see FIG. 2A for details). In the group with L component of 1.1-3.2 and > 3.2, the mean value of patients over 60 years old is higher than that of patients under 60 years old. In the group with L > 3.2, the mean value of patients over 60 years old is significantly higher than that of patients under 60 years old (see FIG.2B for details).
At the time of admission, the patients were divided into three clinical types: normal type, severe type and critical type. Based on the differential test analysis of the age, gender and other classification indicators of 150 patients and their clinical classification (classified as common type, severe type and critical type), the relationship between age, gender and admission clinical classification of patients was explored by using the bilateral chi-square test or Fisher's exact probabilistic test according to the data status. The results showed that there were significant statistical differences between the above indexes and clinical typing. The final outcome of the patients was divided into four outcomes: discharge, improvement, aggravation, and death. The difference analysis results showed that there were significant statistical differences between different age groups and outcome, while there were no or no significant differences between gender and outcome (see table 2 for details).
Table 2 relationship between age, sex and admission clinical classification
Characteristic
|
Clinical typing at admission
|
|
Outcome the ending
|
|
Common
|
Severe, critical type
|
Total
|
Improvement
|
Discharge
|
Aggravation
|
Death
|
Total
|
|
Age a
|
|
|
|
|
|
|
|
|
|
<60
|
30
|
19
|
49
|
35
|
9
|
1
|
4
|
49
|
|
≥60
|
38
|
63
|
101
|
49
|
25
|
6
|
21
|
101
|
|
Total
|
68
|
82
|
150
|
84
|
34
|
7
|
25
|
150
|
|
c2
|
c2=7.415**
|
c2=7.919*
|
|
p
|
P=0.009
|
P=0.048
|
|
Sex a
|
|
|
|
|
|
|
|
|
|
Male
|
31
|
53
|
84
|
44
|
18
|
6
|
16
|
84
|
|
Female
|
37
|
29
|
66
|
40
|
16
|
1
|
9
|
66
|
|
Total
|
68
|
82
|
150
|
84
|
34
|
7
|
25
|
150
|
|
c2
|
c2=5.473*
|
c2=3.733
|
|
p
|
P=0.022
|
P=0.294
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Note: a Chi-square test (double-tail test);b Fisher's exact test(double-tail test)
*At level 0.05 (double-tailed), **At level 0.05 (double-tailed), the correlation was significant.
At the time of admission, the first clinical classification was: normal 68 patients (45.3%), severe and critical 82 patients (54.7%). There are more men than women and more patients over 60 than those under 60. The age of severe, critical and dead patients is too old. From the above description and difference analysis, we found that the elderly population (aged over 60 years old) had significant differences in hospitalization time, clinical classification and outcome after admission, and there were significant statistical differences with the clinical classification and outcome after admission. Therefore, this study extracted 101 samples of the elderly population (over 60 years old), and studied the relationship between the demographic and clinical characteristics of the elderly population and the admission classification, outcome, and death outcome. In the analysis of the difference between CRP, L indexes of the elderly and admission clinical typing, we converted CRP,L into classification variables according to the range of normal reference values, and explored the relationship between CRP, L and admission clinical typing of the elderly respectively by using the bilateral chi-square test or Fisher's exact probability test according to the data state. The results showed that there was a significant difference between CRP index and clinical typing in the elderly population, while there was no significant statistical difference between index L and clinical typing at admission.
Table 3 Relationship between CRP, L and admission clinical typing in the elderly aged over 60 years
Characteristic
|
Clinical typing at admission
|
|
Outcome the ending
|
|
Common
|
Severe, critical type
|
Total
|
Improvement
|
Discharge
|
Aggravation
|
Death
|
Total
|
|
CRP a
|
33.93
±32.63
|
73.83
±49.86
|
58.82
±48.09
|
42.41
±37.20
|
57.81
±45.86
|
82.65
±69.71
|
91.51
±50.49
|
58.82
±48.09
|
|
<3
|
2
|
2
|
4
|
4
|
0
|
0
|
0
|
4
|
|
3-100
|
33
|
39
|
72
|
40
|
20
|
3
|
9
|
72
|
|
>100
|
3
|
22
|
25
|
5
|
5
|
3
|
12
|
25
|
|
Total
|
38
|
63
|
101
|
49
|
25
|
6
|
21
|
101
|
|
c2
|
c2=9.323**
|
c2=20.095**
|
|
p
|
P=0.009
|
P=0.001
|
|
L b
|
1.05
±0.67
|
0.79
±0.56
|
0.89
±0.62
|
0.91
±0.48
|
1.11
±0.95
|
0.65
±0.34
|
0.65
±0.33
|
0.89
±0.0.62
|
|
<1.1
|
27
|
52
|
79
|
36
|
19
|
5
|
19
|
79
|
|
1.1-3.2
|
10
|
10
|
20
|
13
|
4
|
1
|
2
|
20
|
|
>3.2
|
1
|
1
|
2
|
0
|
2
|
0
|
0
|
2
|
|
Total
|
38
|
63
|
101
|
49
|
25
|
6
|
21
|
101
|
|
c2
|
c2=2.149
|
c2=7.289
|
|
p
|
P=0.355
|
P=0.230
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Note: Normal reference values: CRP (<3mg / L); L (1.1× 109 / L -3.2 × 109 / L)
a Chi-square test (double-tail test);b Fisher's exact test(double-tail test)
*At level 0.05 (double-tailed), **At level 0.05 (double-tailed), the correlation was significant.
In the binary Logistic regression analysis of elderly patients with or without death outcome, we used univariate Logistic regression analysis to preliminarily determine the factors affecting death outcome and risk degree. The results showed that gender, hospitalization time, CRP, L and other factors had a significant impact on the death outcome of patients. In the gender classification, taking women as the baseline, the risk degree of death of men was 2.750 times that of women, and the 95% interval of OR value was all greater than 1, satisfying the conditions of risk factors. The decrease in the number of days in hospital, relative to the increase in the number of days in hospital, the risk of death was 0.930 times, and the 95% interval of OR value was all less than 1, satisfying the condition of constituting a protective factor. The risk of death increased by 1.022 times for every unit of increase in CRP index, and the 95% interval of OR value was all greater than 1, satisfying the conditions of risk factors. In order to correct the influence of confounder, we included the above variables into the multivariate Logistic regression analysis, and the results showed that the risk of death increased by 1.018 times for every unit of increase in CRP index, and the 95% interval of OR value was all greater than 1, satisfying the conditions constituting risk factors. (see table 4 for details)
Table 4 Logistic regression analysis of influencing factors of admission classification of patients over 60 years old
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Factor
|
Univariate Logistic regression analysis
|
|
B
|
SE
|
Wald c2
|
p
|
Exp(B)
|
95%CI
|
|
Sex(male)
|
1.012
|
0.424
|
5.405
|
0.017
|
2.750
|
(1.199,6.307)
|
|
HOD
|
-0.073
|
0.035
|
4.354
|
0.037
|
0.930
|
(0.869,0.996)
|
|
CRP
|
0.022
|
0.006
|
13.827
|
0.000
|
1.022
|
(1.011,1.035)
|
|
L
|
-0.708
|
0.373
|
3.602
|
0.058
|
0.493
|
(0.237,1.023)
|
|
Factor
|
Multivariate Logistic regression analysis
|
0.017多因素Logistic回归分析
|
B
|
SE
|
Wald c2
|
p
|
Exp(B)
|
95%CI
|
|
Sex(male)
|
0.480
|
0.488
|
0.969
|
0.325
|
1.616
|
(0.621,4.203)
|
|
HOD
|
-0.054
|
0.040
|
1.792
|
0.181
|
0.948
|
(0.876,1.025)
|
|
CRP
|
0.018
|
0.006
|
8.022
|
0.005
|
1.018
|
(1.006,1.031)
|
|
L
|
-0.568
|
0.367
|
2.395
|
0.122
|
0.567
|
(0.276,1.163)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Note: HOD: Hospitalization days
In the multivariate Logistic regression analysis of hospitalization days, gender, CRP, L and outcome of the elderly patients, compared with the elderly patients who died, the risk of death in discharged patients was 0.985 times for every unit of decrease in CRP, and the 95% interval of OR value was all less than 1. The decline of CRP constituted a condition of protective factors. The outcomes of discharge, improvement and aggravation were compared with those of the elderly patients who died,the increase of hospitalization days all constituted the risk factors for death, with the OR values of 1.306, 1.354 and 1.339, respectively (see table 5 for details).
Table 5 multivariate Logistic regression analysis of the influencing factors of mortality outcome in patients
Factors
|
B
|
SE
|
Wald c2
|
p
|
Exp(B)
|
95%CI
|
Discharged
|
|
|
|
|
|
|
|
HOD
|
0.267
|
0.069
|
14.831
|
0.000
|
1.306
|
1.140
|
1.496
|
CRP
|
-0.015
|
0.008
|
4.132
|
0.042
|
0.985
|
0.970
|
0.999
|
L
|
1.163
|
0.880
|
1.746
|
0.186
|
3.198
|
0.570
|
17.944
|
[sex = male]
|
0.097
|
0.726
|
0.018
|
0.894
|
1.102
|
0.265
|
4.574
|
[sex = female]
|
0
|
|
|
|
|
|
|
Improvement
|
|
|
|
|
|
|
|
HOD
|
0.303
|
0.075
|
16.271
|
0.000
|
1.354
|
1.169
|
1.570
|
CRP
|
-0.006
|
0.008
|
0.482
|
0.487
|
0.994
|
0.978
|
1.010
|
L
|
1.719
|
0.904
|
3.620
|
0.057
|
5.579
|
0.949
|
32.785
|
[sex = male]
|
0.256
|
0.801
|
0.102
|
0.749
|
1.292
|
0.269
|
6.205
|
[sex = female]
|
0
|
|
|
|
|
|
|
Exacerbation
|
|
|
|
|
|
|
|
HOD
|
0.292
|
0.103
|
7.962
|
0.005
|
1.339
|
1.093
|
1.639
|
CRP
|
0.001
|
0.011
|
0.006
|
0.940
|
1.001
|
0.979
|
1.023
|
L
|
0.611
|
1.349
|
0.205
|
0.651
|
1.842
|
0.131
|
25.931
|
[sex = male]
|
1.125
|
1.292
|
0.758
|
0.384
|
3.081
|
0.245
|
38.774
|
[sex = female]
|
0
|
|
|
|
|
|
|
Note: HOD: Hospitalization days
In the diagnostic relationship between CRP, L and death outcome in elderly patients, non-death outcome and death outcome were taken as the positive classification basis. We took AUC=0.5 as the null hypothesis, and the significance of CRP and L in ROC curve analysis was less than 0.05. The area under the CRP curve is greater than 0.7(AUC=0.732, 95%CI: 0.61-0.86), and the area under the L curve is close to 0.7(AUC=0.648, 95%CI: 0.51-0.79). It can be seen in the ROC curve that both CRP and L have a good diagnostic value for the outcome of death. The cutoff values of CRP and L diagnosis of death outcome were determined by the maximum Youden index, which were 91.5 and 0.615 respectively (see FIG.3 for details).