A total of 417 subjects(239 men and 178 women) were included in the final analysis with the mean age(SD) of 67.8(13.2) years. The median baseline NIHSS score was 4(interquartile range, 2–10). Compared with non-pneumonia participants, patients with pneumonia were more likely to be older and higher admission NIHSS score, duration of hospitalization, WBC, CRP and RDW, and higher proportions of great vessels, small-artery occlusion and total anterior circulation infarcts, but lower proportions of atherosclerosis and partial anterior circulation infarcts. Compared with the good functional outcome patients, those who had poor functional outcome were more likely to be men and older, had higher admission NIHSS score, duration of hospitalization, WBC, CRP and RDW, and had higher proportions of great vessels, hypertension history, small-artery occlusion, total anterior circulation infarcts, but lower proportions of atherosclerosis and partial anterior circulation infarcts(all P < 0.05) (Table 1).
Table 1
Baseline characteristics of study participants according to pneumonia and at-discharge functional outcome.
|
Post-stroke pneumonia
|
Functional outcome
|
Characteristics a
|
Non-pneumonia
|
Pneumonia
|
P value
|
Good
|
Poor
|
P value
|
No. of patients
|
328
|
89
|
|
293
|
124
|
|
Age, y
|
66.5 ± 12.9
|
72.9 ± 13.2
|
< 0.001
|
66.2 ± 13.0
|
71.7 ± 13.0
|
< 0.001
|
Male sex
|
193 (58.8)
|
46 (51.7)
|
0.23
|
179 (61.1)
|
60 (48.4)
|
0.02
|
Current cigarette smoking
|
119 (36.3)
|
27 (30.3)
|
0.30
|
111 (37.9)
|
35 (28.2)
|
0.06
|
Admission NIHSS score
|
3 (2–6)
|
12 (7–20)
|
< 0.001
|
3 (2–5)
|
11 (6–16)
|
< 0.001
|
Great vessels
|
45 (13.7)
|
54 (60.7)
|
< 0.001
|
32 (10.9)
|
67 (54.0)
|
< 0.001
|
Medical history
|
|
|
|
|
|
|
Hypertension
|
210 (64.0)
|
62 (69.7)
|
0.32
|
181 (61.8)
|
91 (73.4)
|
0.02
|
Hyperglycemia
|
69 (21.0)
|
18 (20.2)
|
0.87
|
54 (18.4)
|
33 (26.6)
|
0.06
|
Hyperlipidemia
|
68 (20.7)
|
15 (16.9)
|
0.42
|
62 (21.2)
|
21 (16.9)
|
0.32
|
Coronary heart disease
|
21 (6.4)
|
6 (6.7)
|
0.91
|
18 (6.1)
|
9 (7.3)
|
0.67
|
History of stroke
|
44 (13.5)
|
13 (14.6)
|
0.78
|
42 (14.4)
|
15 (12.1)
|
0.57
|
Medications
|
|
|
|
|
|
|
Anticoagulant
|
6 (1.8)
|
1 (1.1)
|
0.65
|
4 (1.4)
|
3 (2.4)
|
0.44
|
Antiplatelet
|
15 (4.6)
|
4 (4.5)
|
0.97
|
13 (4.4)
|
6 (4.8)
|
0.86
|
TOAST
|
|
|
|
|
|
|
A
|
256 (78.0)
|
48 (53.9)
|
< 0.001
|
230 (78.5)
|
74 (59.7)
|
< 0.001
|
C
|
18 (5.5)
|
3 (3.4)
|
0.59
|
19 (6.5)
|
2 (1.6)
|
0.05
|
S
|
48 (14.6)
|
37 (41.6)
|
< 0.001
|
40 (13.7)
|
45 (36.3)
|
< 0.001
|
O
|
6 (1.8)
|
1 (1.1)
|
1.00
|
4 (1.4)
|
3 (2.4)
|
0.43
|
Stroke syndrome
|
|
|
|
|
|
|
TACS
|
37 (11.3)
|
42 (47.2)
|
< 0.001
|
27 (9.2)
|
52 (41.9)
|
< 0.001
|
PACS
|
217 (66.2)
|
36 (40.4)
|
< 0.001
|
199 (67.9)
|
54 (43.5)
|
< 0.001
|
POCS
|
63 (19.2)
|
11 (12.4)
|
0.13
|
56 (19.1)
|
18 (14.5)
|
0.26
|
LACS
|
11 (3.4)
|
0
|
0.13
|
11 (3.8)
|
0
|
0.04
|
Duration of hospitalization, d
|
8 (7–10)
|
13 (8–19)
|
< 0.001
|
8 (7–10)
|
11 (8–16)
|
< 0.001
|
White blood cell
|
7.1 ± 2.3
|
9.4 ± 3.4
|
< 0.001
|
6.9 ± 2.2
|
9.2 ± 3.3
|
< 0.001
|
C-reactive protein
|
6.8 ± 9.9
|
20.7 ± 31.1
|
< 0.001
|
7.3 ± 11.8
|
15.6 ± 26.1
|
< 0.001
|
Red blood cell distribution width
|
42.4 ± 3.2
|
43.9 ± 6.0
|
0.001
|
42.6 ± 3.5
|
43.1 ± 4.9
|
0.20
|
a Continuous variables are expressed as mean ± standard deviation or median (interquartile range). Categorical variables are expressed as number (%). Abbreviations: NIHSS, National Institute of Health Stroke Scale; A, atherosclerosis; C, cardioembolism; S, small-artery occlusion; O, stroke of other causes; TACS, total anterior circulation infarcts; PACS, partial anterior circulation infarcts; POCS, posterior circulation infarcts; LACS, lacunar infarcts.
Independent in-hospital pneumonia predictor: role of WBC or CRP and RDW
89(21.3%) patients experienced in-hospital pneumonia (Table 2). After adjusting for age, baseline NIHSS score, great vessels, and other traditional risk factors, higher WBC, CRP and RDW were associated with increased risk of post-stroke pneumonia. Furthermore, patients with higher levels of RDW and inflammatory biomarkers have the highest risk of post-stroke pneumonia in AIS patients undergoing thrombolysis. Compared to LWLR group, the multivariable-adjusted OR(95%CI) for patients in HWHR was 12.16 (4.21–35.14) for post-stroke pneumonia. And compared to LCLR group, the multivariable-adjusted ORs (95% CIs) for patients in HCHR were 6.93 (2.70-17.78) for post-stroke pneumonia.
Table 2
Joint effects of inflammatory biomarkers and red blood cell distribution width on post-stroke pneumonia of acute ischemic stroke patients with thrombolysis. Data are Odds ratios (95% confidence intervals).
|
|
Unadjusted
|
Multivariable adjusted
|
|
Events, %
|
OR
|
95% CI
|
P value
|
OR
|
95% CI
|
P value
|
High WBC (> 8.85)
|
50 (46.7)
|
6.10
|
3.67–10.12
|
< 0.001
|
6.15
|
3.10-12.55
|
< 0.001
|
High CRP (> 5.75)
|
44 (41.9)
|
4.28
|
2.60–7.06
|
< 0.001
|
3.82
|
1.99–7.33
|
< 0.001
|
High RDW (> 43.85)
|
37 (30.8)
|
2.10
|
1.29–3.43
|
0.003
|
1.97
|
1.02–3.80
|
0.04
|
Joint effect of WBC and RDW
|
|
|
|
|
|
|
WBC (-) and RDW (-)
|
24 (10.6)
|
1.00
|
|
|
1.00
|
|
|
WBC (-) and RDW (+)
|
15 (17.9)
|
1.83
|
0.91–3.69
|
0.09
|
1.57
|
0.64–3.82
|
0.32
|
WBC (+) and RDW (-)
|
28 (39.4)
|
5.48
|
2.90-10.36
|
< 0.001
|
5.18
|
2.17–12.40
|
0.0002
|
WBC (+) and RDW (+)
|
22 (61.1)
|
13.23
|
5.99–29.22
|
< 0.001
|
12.16
|
4.21–35.14
|
< 0.001
|
Joint effect of CRP and RDW
|
|
|
|
|
|
|
CRP (-) and RDW (-)
|
29 (12.6)
|
1.00
|
|
|
1.00
|
|
|
CRP (-) and RDW (+)
|
16 (19.7)
|
1.72
|
0.88–3.36
|
0.12
|
1.56
|
0.67–3.64
|
0.31
|
CRP (+) and RDW (-)
|
23 (34.8)
|
3.73
|
1.97–7.06
|
< 0.001
|
3.86
|
1.72–8.66
|
0.001
|
CRP (+) and RDW (+)
|
21 (53.8)
|
8.13
|
3.88–17.03
|
< 0.001
|
6.93
|
2.70-17.78
|
< 0.001
|
Multivariable adjusted model included age, sex, current smoking, admission NIHSS score, great vessels, medical history (hypertension, hyperglycemia, hyperlipidemia and coronary heart disease), history of stroke, ischemic stroke syndrome, use of anticoagulant and antiplatelet medication, duration of hospitalization.
Abbreviations: WBC, white blood cell; CRP, C-reactive protein; RDW, red blood cell distribution width.
Independent functional outcome predictor: role of WBC or CRP and RDW
124(29.7%) patients experienced poor functional outcome (Table 3). After adjusting for age, baseline NIHSS score, great vessels, and other traditional risk factors, higher WBC and CRP levels were associated with increased risk of poor functional outome. In addition, patients with higher levels of RDW and inflammatory biomarkers have the highest risk of poor functional outcome in AIS patients undergoing thrombolysis. Compared to LWLR group, the multivariable-adjusted OR(95%CI) for patients in HWHR group was 9.31 (3.19-27.17) for functional outcome. And compared to LCLR group, the multivariable-adjusted OR(95%CI) for patients in HCHR was 3.38 (1.10-10.39) for functional outcome.
Table 3
Joint effects of inflammatory biomarkers and red blood cell distribution width on at-discharge functional outcome of acute ischemic stroke patients with thrombolysis.
|
|
Unadjusted
|
Multivariable adjusted
|
|
Events, %
|
OR
|
95% CI
|
P value
|
OR
|
95% CI
|
P value
|
High WBC (> 8.75)
|
66 (60.6)
|
6.62
|
4.10-10.68
|
< 0.001
|
5.93
|
3.08–11.41
|
< 0.001
|
High CRP (> 8.40)
|
35 (48.9)
|
5.09
|
2.82–9.20
|
< 0.001
|
2.51
|
1.25–5.06
|
0.01
|
High RDW (> 43.95)
|
43 (37.1)
|
1.60
|
1.02–2.52
|
0.04
|
1.29
|
0.71–2.35
|
0.41
|
Joint effect of WBC and RDW
|
|
|
|
|
|
|
WBC (-) and RDW (-)
|
40 (17.6)
|
1.00
|
|
|
1.00
|
|
|
WBC (-) and RDW (+)
|
18 (22.2)
|
1.34
|
0.72–2.50
|
0.35
|
0.99
|
0.46–2.16
|
0.99
|
WBC (+) and RDW (-)
|
41 (55.4)
|
5.81
|
3.28–10.29
|
< 0.001
|
4.93
|
2.32–10.49
|
< 0.001
|
WBC (+) and RDW (+)
|
25 (71.4)
|
11.69
|
5.21–26.24
|
< 0.001
|
9.31
|
3.19–27.17
|
< 0.001
|
Joint effect of CRP and RDW
|
|
|
|
|
|
|
CRP (-) and RDW (-)
|
61 (22.8)
|
1.00
|
|
|
1.00
|
|
|
CRP (-) and RDW (+)
|
28 (29.8)
|
1.43
|
0.85–2.43
|
0.18
|
1.21
|
0.63–2.32
|
0.56
|
CRP (+) and RDW (-)
|
20 (58.8)
|
4.82
|
2.30-10.12
|
< 0.001
|
3.16
|
1.33–7.54
|
0.01
|
CRP (+) and RDW (+)
|
15 (68.2)
|
7.24
|
2.82–18.55
|
< 0.001
|
3.38
|
1.10-10.39
|
0.03
|
Multivariable adjusted model included age, sex, current smoking, admission NIHSS score, great vessels, medical history (hypertension, hyperglycemia, hyperlipidemia and coronary heart disease), history of stroke, ischemic stroke syndrome, use of anticoagulant and antiplatelet medication, duration of hospitalization. Abbreviations: WBC, white blood cell, CRP, C-reactive protein, RDW, red blood cell distribution width.
The predictive ability of inflammatory biomarkers combined with RDW for in-hospital pneumonia and functional outcomes
We further assessed the predictive value of inflammatory biomarkers and RDW beyond established risk factors for in-hospital pneumonia and functional outcome (Table 4). First, the risk discrimination for study outcomes by inflammatory biomarkers and RDW was analyzed (Figure 2). The basic model included age, sex, current smoking, admission NIHSS score, great vessels, medical history (hypertension, hyperglycemia, hyperlipidemia and coronary heart disease), history of stroke, ischemic stroke syndrome, duration of hospitalization, use of anticoagulant and antiplatelet medication. The addition of inflammatory biomarkers and RDW to the basic model significantly improved discriminatory power, as C statistics increased from 0.874 (95% CI: 0.838-0.904) to 0.898 (95% CI: 0.865-0.925) or 0.891 (95% CI: 0.857-0.920) for post-stroke pneumonia; and from 0.859 (95% CI: 0.821-0.891) to 0.885 (95% CI: 0.851-0.914) or 0.864 (95% CI: 0.827-0.895) for functional outcome(Figure 2). Second, the addition of inflammation biomarkers and RDW to the basic model significantly improved risk classification for both post-stroke pneumonia (category-free NRIs were 80.4% (95% CI 58.4%–102.4%, P <0.001) for WBC and RDW and 60.8% (95% CI 38.1%–83.4%, P <0.001) for CPR and RDW, and the corresponding IDIs were 7.5% (95% CI 4.0%–11.0%, P <0.001) and 4.9% (95% CI 2.1%–7.8%, P=0.0006), respectively). Similarly, simultaneously adding inflammation biomarkers and RDW to the basic model improved risk classification for functional outcome (category-free NRIs were 70.8% (95% CI 51.2%–90.5%, P <0.001) for WBC and RDW and 26.2% (95% CI 7.0%–45.4%, P = 0.01) for CPR and RDW, and the corresponding IDIs were 6.1% (95% CI 3.4%–8.9%, P <0.001) and 1.5% (95% CI 0.03%–3.0%, P = 0.04), respectively). Furthermore, the Hosmer Lemeshow tests showed adequate calibration of models after the addition of inflammation biomarkers and RDW (all P>0.05)(Table 4).
Table 4
Performance of models with inflammation biomarkers and red blood cell distribution width.
|
NRI (category free)
|
IDI
|
Calibration
|
|
Estimate (95% CI), %
|
P value
|
Estimate (95% CI), %
|
P value
|
χ2 value
|
P value
|
Post-stroke Pneumonia
|
|
|
|
|
|
Basic model
|
Reference
|
|
Reference
|
|
9.14
|
0.33
|
Basic model + WBC + RDW
|
80.4 (58.4-102.4)
|
< 0.001
|
7.5 (4.0–11.0)
|
< 0.001
|
3.42
|
0.91
|
Basic model + CRP + RDW
|
60.8 (38.1–83.4)
|
< 0.001
|
4.9 (2.1–7.8)
|
0.0006
|
10.13
|
0.26
|
Functional outcome
|
|
|
|
|
|
Basic model
|
Reference
|
|
Reference
|
|
9.79
|
0.28
|
Basic model + WBC + RDW
|
70.8 (51.2–90.5)
|
< 0.001
|
6.1 (3.4–8.9)
|
< 0.001
|
3.98
|
0.86
|
Basic model + CRP + RDW
|
26.2 (7.0-45.4)
|
0.01
|
1.5 (0.03-3.0)
|
0.04
|
9.68
|
0.29
|
Basic model included age, sex, current smoking, admission NIHSS score, great vessels, medical history (hypertension, hyperglycemia, hyperlipidemia and coronary heart disease), history of stroke, ischemic stroke syndrome, use of anticoagulant and antiplatelet medication, duration of hospitalization. Abbreviations: WBC, white blood cell; CRP, C-reactive protein; RDW, red blood cell distribution width; CI, confidence interval; IDI, integrated discrimination improvement; NRI, net reclassification index.
HWHR or HCHR had worse functional outcomes at discharge.
LWLR patients had a median mRS score of 1 (IQR 0-2), in comparison to HWHR patients with a median mRS score of 4 (IQR 2-5) at discharge (Figure 3). LCLR patients had a median mRS score of 1 (IQR 0-2), in comparison to HCHR patients with a median mRS score of 4 (IQR 2-5) at discharge (Figure 4). The risk of poor outcome at discharge was significantly higher among HWHR and HCHR participants than that among those in LWLR and LCLR groups (P-trend <0.001) .