The final study cohort consisted of 8136 SAH patients available for analysis (Fig 1).
Among the study population, 4802 (59.0%) were female, and the mean age was 60.2±13.0 years. More than one-third of patients presented with symptoms during the nighttime (Table 1). 1791 (22.0%) patients had in-hospital mortality or withdrawal of life-sustaining treatment. Patients who had an unfavorable outcome at discharge were significantly older, more often presented in the nighttime, presented with higher SBP, and had a history of hypertension. Patients with poor outcome were more likely to have a higher score on the prestroke modified Rankin Scale and a higher WFNS score at presentation. There was no difference in nimodipine use between the two groups. The incidence of rebleeding, seizure, hydrocephalus, respiratory arrest, atrial fibrillation, myocardial infarction, and gastrointestinal bleeding were significantly higher in patients with poor outcome.
Table 1. Characteristics of the Study Population
Characteristics
|
Total
(n=8136)
|
Good outcome
(n=6345)
|
Poor outcome
(n=1791)
|
P value
|
Age, mean±SD, y
|
60.2±13.0
|
59.4±12.6
|
62.9±13.9
|
<0.001
|
Female, n (%)
|
4802 (59.0)
|
3746 (59.0)
|
1056 (59.0)
|
0.95
|
Hospital level, n (%)
|
|
|
|
<0.001
|
Secondary
|
2656 (32.6)
|
1912 (30.1)
|
744 (41.5)
|
|
Tertiary
|
5480 (67.4)
|
4433 (69.9)
|
1047 (58.5)
|
|
Onset Time, n (%)
|
|
|
|
0.004
|
Daytime
|
5283 (64.9)
|
4171 (65.7)
|
1112 (62.1)
|
|
Nighttime
|
2853 (35.1)
|
2174 (34.3)
|
679 (37.9)
|
|
Prestroke modified Rankin Scale, n (%)*
|
|
|
|
0.007
|
0-2
|
6500 (79.9)
|
5108 (80.5)
|
1392 (77.7)
|
|
3-5
|
1635 (20.1)
|
1237 (19.5)
|
398 (22.2)
|
|
WFNS score, n (%)†
|
|
|
|
<0.001
|
I
|
2358 (29.0)
|
1972 (31.1)
|
386 (21.6)
|
|
II
|
538 (6.6)
|
459 (7.2)
|
79 (4.4)
|
|
III
|
223 (2.7)
|
176 (2.8)
|
47 (2.6)
|
|
IV
|
880 (10.8)
|
648 (10.2)
|
232 (13.0)
|
|
V
|
776 (9.5)
|
423 (6.7)
|
353 (19.7)
|
|
Admission SBP, mean±SD, mmHg
|
152.9±27.8
|
151.2±26.2
|
159.2±32.3
|
<0.001
|
Medical history, n (%)
|
|
|
|
|
Hypertension
|
4050 (49.8)
|
3049 (48.1)
|
1001 (55.9)
|
<0.001
|
Diabetes mellitus
|
554 (6.8)
|
423 (6.7)
|
131 (7.3)
|
0.34
|
Dyslipidemia
|
255 (3.1)
|
208 (3.3)
|
47 (2.6)
|
0.16
|
Current or previous smoking, n (%)
|
1710 (21.0)
|
1354 (21.3)
|
356 (19.9)
|
0.18
|
Current or previous drinking, n (%)
|
1309 (16.1)
|
1017 (16.0)
|
292 (16.3)
|
0.78
|
Treatment methods, n (%)
|
|
|
|
<0.001
|
Internal medicine
|
4988 (61.3)
|
3495 (55.1)
|
1493 (83.4)
|
|
Neurosurgery
|
762 (9.4)
|
673 (10.6)
|
89 (5.0)
|
|
Endovascular treatment
|
2323 (28.6)
|
2120 (33.4)
|
203 (11.3)
|
|
Combined therapy
|
63 (0.8)
|
57 (0.9)
|
6 (0.3)
|
|
Nimodipine, n (%)
|
6593 (81.0)
|
5134 (80.9)
|
1459 (81.5)
|
0.60
|
CSF diversion, n (%)
|
255 (3.1)
|
184 (2.9)
|
71 (4.0)
|
0.02
|
Complications, n (%)
|
|
|
|
|
Rebleeding
|
541 (6.6)
|
326 (5.1)
|
215 (12.0)
|
<0.001
|
Ischemic stroke
|
233 (2.9)
|
180 (2.8)
|
53 (3.0)
|
0.78
|
Seizure
|
151 (1.9)
|
106 (1.7)
|
45 (2.5)
|
0.02
|
Hydrocephalus
|
393 (4.8)
|
256 (4.0)
|
137 (7.6)
|
<0.001
|
DVT
|
126 (1.5)
|
105 (1.7)
|
21 (1.2)
|
0.14
|
Respiratory arrest
|
314 (3.9)
|
50 (0.8)
|
264 (14.7)
|
<0.001
|
Atrial fibrillation
|
100 (1.2)
|
62 (1.0)
|
38 (2.1)
|
0.001
|
Myocardial infarction
|
33 (0.4)
|
18 (0.3)
|
15 (0.8)
|
0.001
|
Gastrointestinal bleeding
|
233 (2.9)
|
139 (2.2)
|
94 (5.2)
|
<0.001
|
Abbreviations: SD, standard deviation; WFNS, World Federation of Neurosurgical Societies; SBP, systolic blood pressure; CSF, cerebrospinal fluid; DVT, deep vein thrombosis. P value for comparison between patients with good outcome and poor outcome.
*Missing value for 1 (0.01%) patient.
†Missing value for 3361 (41.3%) patients.
Significant factors from the univariate analysis were entered into a multiple regression model to determine the independent predictors of poor outcome, as shown in Table 2. Previously well-accepted independent predictors of a negative outcome, such as age, high SBP, rebleeding, and hydrocephalus were confirmed in this cohort. Respiratory arrest and gastrointestinal bleeding are also factors that significantly increase the odds of poor outcome. Nighttime onset was identified as an independent predictor of poor outcome (OR 1.260 [95% CI 1.118-1.419], p<0.001). WFNS was removed from the main analysis due to high percentage of missing data. A baseline comparison in patients with WFNS score is shown in the supplementary (Table S1).
Table 2. Predictive Factors for In-Hospital Mortality or Withdrawal of Life-Sustaining Treatment
Factors
|
Unadjusted OR
(95% CI)
|
P value
|
Adjusted OR
(95% CI)
|
P value
|
Age
|
1.021 (1.017-1.026)
|
<0.001
|
1.007 (1.002-1.012)
|
0.004
|
Female
|
0.998 (0.897-1.111)
|
0.97
|
1.049 (0.932-1.181)
|
0.43
|
Hospital level
|
|
|
|
|
Secondary
|
-
|
-
|
-
|
-
|
Tertiary
|
0.608 (0.545-0.677)
|
<0.001
|
0.822 (0.728-0.927)
|
0.001
|
Onset Time
|
|
|
|
|
Daytime
|
-
|
-
|
-
|
-
|
Nighttime
|
1.170 (1.049-1.304)
|
0.005
|
1.260 (1.118-1.419)
|
<0.001
|
Admission SBP
|
1.010 (1.008-1.012)
|
<0.001
|
1.007 (1.005-1.009)
|
<0.001
|
Treatment methods
|
|
|
|
|
Internal medicine
|
-
|
-
|
-
|
-
|
Neurosurgery
|
0.310 (0.246-0.390)
|
<0.001
|
0.294 (0.229-0.376)
|
<0.001
|
Endovascular treatment
|
0.224 (0.192-0.262)
|
<0.001
|
0.250 (0.212-0.296)
|
<0.001
|
Combined therapy
|
0.247 (0.106-0.573)
|
0.001
|
0.210 (0.082-0.536)
|
0.001
|
Rebleeding
|
2.520 (2.103-3.020)
|
<0.001
|
1.534 (1.240-1.898)
|
<0.001
|
Hydrocephalus
|
1.971 (1.591-2.443)
|
<0.001
|
1.569 (1.207-2.040)
|
<0.001
|
Respiratory arrest
|
21.779 (16.015-29.618)
|
<0.001
|
16.522 (11.941-22.860)
|
<0.001
|
Gastrointestinal bleeding
|
2.475 (1.894-3.232)
|
<0.001
|
1.415 (1.021-1.961)
|
0.04
|
Abbreviations: OR, odds ratio; SBP, systolic blood pressure. Adjusted for age, female, hospital level, time of onset, prestroke modified Rankin Scale, systolic blood pressure on admission, treatment methods, cerebrospinal fluid diversion, rebleeding, seizure, hydrocephalus, respiratory arrest, atrial fibrillation, myocardial infarction, gastrointestinal bleeding.
To further understand the relationship between onset time and outcome, the distribution of SAH patients’ onset and discharge outcome according to 1-h intervals in a nychthemeron is shown in Figure 2. The frequency distribution of the onset of SAH presented a bimodal distribution in two peak times: a slightly higher peak between 8 a.m. to 10 a.m., and a relatively lower peak between 3 p.m. to 5 p.m. Nighttime onset nadir was observed from midnight to 4 a.m. (Fig 2A).
Conversely, the distribution of poor outcome proportion showed an evident spike from midnight to 4 a.m., peaking between 2 a.m. to 3 a.m. The downturn started from 8 a.m., lasting to 6 p.m. (Fig 2B). Upon reversing the scale for poor discharge outcome, the inverse trend of onset and poor outcome was more apparent (Fig 2C).
Patients were divided into 4 groups according to onset time periods (daytime or nighttime) and whether they were hypertensive (SBP>160mmHg) or normotensive (SBP≤160mmHg) at admission: daytime onset + normotensive (3390 patients), daytime onset + hypertensive (1666 patients), nighttime onset + normotensive (1909 patients), and nighttime onset + hypertensive (1171 patients). Characteristics of the included patients grouped in the above approach are shown in Table 3.
Table 3. Characteristics of the Status of Onset Time and Admission SBP in the Study Population
Characteristics
|
Daytime onset
|
Nighttime onset
|
P value
|
Normotensive
(n=3390)
|
Hypertensive
(n=1666)
|
Normotensive
(n=1909)
|
Hypertensive
(n=1171)
|
Age, mean±SD, y
|
59.3±13.1
|
63.5±12.3
|
57.4±12.9
|
62.5±12.2
|
<0.001
|
Female, n (%)
|
2043 (58.3)
|
1062 (59.6)
|
1047 (58.3)
|
650 (61.6)
|
0.24
|
Hospital level, n (%)
|
|
|
|
|
<0.001
|
Secondary
|
1038 (29.6)
|
692 (38.9)
|
484 (26.9)
|
442 (41.9)
|
|
Tertiary
|
2464 (70.4)
|
1089 (61.1)
|
1313 (73.1)
|
614 (58.1)
|
|
Prestroke modified Rankin Scale, n (%)*
|
|
|
|
|
0.10
|
0-2
|
2800 (80.0)
|
1396 (78.4)
|
1449 (80.6)
|
855 (81.0)
|
|
3-5
|
702 (20.0)
|
385 (21.6)
|
348 (19.4)
|
200 (18.9)
|
|
WFNS score, n (%)†
|
|
|
|
|
<0.001
|
I
|
1034 (29.5)
|
480 (27.0)
|
550 (30.6)
|
294 (27.8)
|
|
II
|
247 (7.1)
|
113 (6.3)
|
125 (7.0)
|
53 (5.0)
|
|
III
|
91 (2.6)
|
51 (2.9)
|
42 (2.3)
|
39 (3.7)
|
|
IV
|
349 (10.0)
|
205 (11.5)
|
193 (10.7)
|
133 (12.6)
|
|
V
|
274 (7.8)
|
225 (12.6)
|
138 (7.7)
|
139 (13.2)
|
|
Medical history, n (%)
|
|
|
|
|
|
Hypertension
|
1342 (38.3)
|
1258 (70.6)
|
720 (40.1)
|
730 (69.1)
|
<0.001
|
Diabetes mellitus
|
212 (6.1)
|
136 (7.6)
|
116 (6.5)
|
90 (8.5)
|
0.02
|
Dyslipidemia
|
112 (3.2)
|
61 (3.4)
|
48 (2.7)
|
34 (3.2)
|
0.61
|
Current or previous smoking, n (%)
|
737 (21.0)
|
371 (20.8)
|
397 (22.1)
|
205 (19.4)
|
0.40
|
Current or previous drinking, n (%)
|
544 (15.5)
|
304 (17.1)
|
288 (16.0)
|
173 (16.4)
|
0.54
|
Treatment methods, n (%)
|
|
|
|
|
<0.001
|
Internal medicine
|
2229 (63.6)
|
1112 (62.4)
|
999 (55.6)
|
648 (61.4)
|
|
Neurosurgery
|
305 (8.7)
|
186 (10.4)
|
156 (8.7)
|
115 (10.9)
|
|
Endovascular treatment
|
937 (26.8)
|
466 (26.2)
|
630 (35.1)
|
290 (27.5)
|
|
Combined therapy
|
31 (0.9)
|
17 (1.0)
|
12 (0.7)
|
3 (0.3)
|
|
Nimodipine, n (%)
|
2730 (78.0)
|
1460 (82.0)
|
1488 (82.8)
|
915 (86.6)
|
<0.001
|
CSF diversion, n (%)
|
77 (2.2)
|
76 (4.3)
|
53 (2.9)
|
49 (4.6)
|
<0.001
|
Complications, n (%)
|
|
|
|
|
|
Rebleeding
|
210 (6.0)
|
137 (7.7)
|
107 (6.0)
|
87 (8.2)
|
0.01
|
Ischemic stroke
|
89 (2.5)
|
63 (3.5)
|
53 (2.9)
|
28 (2.7)
|
0.22
|
Seizure
|
70 (2.0)
|
30 (1.7)
|
28 (1.6)
|
23 (2.2)
|
0.54
|
Hydrocephalus
|
122 (3.5)
|
108 (6.1)
|
86 (4.8)
|
77 (7.3)
|
<0.001
|
DVT
|
36 (1.0)
|
36 (2.0)
|
31 (1.7)
|
23 (2.2)
|
0.008
|
Respiratory arrest
|
103 (2.9)
|
101 (5.7)
|
43 (2.4)
|
67 (6.3)
|
<0.001
|
Atrial fibrillation
|
49 (1.4)
|
27 (1.5)
|
14 (0.8)
|
10 (0.9)
|
0.13
|
Myocardial infarction
|
15 (0.4)
|
7 (0.4)
|
7 (0.4)
|
4 (0.4)
|
0.99
|
Gastrointestinal bleeding
|
64 (1.8)
|
74 (4.2)
|
54 (3.0)
|
41 (3.9)
|
<0.001
|
In-hospital mortality or withdrawal of life-sustaining treatment, n (%)
|
639 (18.2)
|
473 (26.6)
|
359 (20.0)
|
320 (30.3)
|
<0.001
|
Abbreviations: SD, standard deviation; WFNS, World Federation of Neurosurgical Societies; CSF, cerebrospinal fluid; DVT, deep vein thrombosis. P value for comparison between 4 groups of hypertension/ normotensive patients with daytime/ nighttime onset.
*Missing value for 1 (0.1%) patient.
†Missing value for 3361 (41.3%) patients.
Hypertensive patients who had nighttime onset were more likely to be elderly, have a higher severity of the stroke, and have medical histories of hypertension and diabetes mellitus. The percentage of neurosurgical and interventional treatment was relatively high in patients with nighttime onset, as were the use of nimodipine and cerebrospinal fluid (CSF) diversion. Hypertensive patients who had nighttime onset were at high risk for complications including rebleeding (8.2%), hydrocephalus (7.3%), and deep vein thrombosis (2.2%). This subgroup of patients had the highest rate of poor outcome, whereas normotensive patients with daytime onset had the lowest rate of poor outcome (30.3% vs 18.2%, p<0.001).
After adjustment using the above model, all three other groups of patients were associated with worse outcome in comparison to the daytime onset + normotensive group (Table 4). The risk of poor discharge outcome in the nighttime onset + hypertensive group was significantly highest in the four groups (OR 1.796 [95% CI 1.511-2.136], p<0.001). Nighttime onset + normotensive and daytime onset + hypertensive groups were also significantly more at risk for poor outcome (OR 1.295 [95% CI 1.109-1.511] and OR 1.463 [95% CI 1.259-1.700], respectively) in comparison to daytime onset + normotensive group. Other independent risk factors remain the same. Subgroup analysis in patients with WFNS score is shown in Table S2.
Table 4. Effect of Onset Time and Admission SBP on In-Hospital Mortality or Withdrawal of Life-Sustaining Treatment
Factors
|
|
Unadjusted OR
(95% CI)
|
P value
|
Adjusted OR
(95% CI)
|
P value
|
Normotensive
|
|
|
|
|
|
|
Daytime
|
-
|
-
|
-
|
-
|
|
Nighttime
|
1.119 (0.969-1.292)
|
0.13
|
1.295 (1.109-1.511)
|
0.001
|
Hypertensive
|
|
|
|
|
|
|
Daytime
|
1.620 (1.415-1.856)
|
<0.001
|
1.463 (1.259-1.700)
|
<0.001
|
|
Nighttime
|
1.948 (1.665-2.279)
|
<0.001
|
1.796 (1.511-2.136)
|
<0.001
|
Abbreviations: SBP, systolic blood pressure; OR, odds ratio. Adjusted for age, female, hospital level, prestroke modified Rankin Scale, onset time and admission SBP, treatment methods, cerebrospinal fluid diversion, rebleeding, seizure, hydrocephalus, respiratory arrest, atrial fibrillation, myocardial infarction, gastrointestinal bleeding.
We further analyzed the association of onset time periods with outcomes by using 6-hour intervals, sorting onset time into forenoon (6:00 a.m.- 11:59 a.m.), afternoon (12:00 a.m.- 5:59 p.m.), before-midnight (6:00 p.m.- 11:59 p.m.), and after-midnight (12:00 p.m.- 5:59 a.m.). Logistic regression showed between four periods of onset time, patients with an after-midnight onset had the highest risk of poor outcome (OR 1.353 [95% CI 1.103-1.658], p=0.004), as shown in Table S3. Admission SBP remained an independent risk factor. When combining the 6-h interval onset time with admission SBP, compared to normotensive patients with afternoon onset, patients with after-midnight onset combined with admission SBP > 160 mmHg had the highest risk of poor outcome (OR 1.951 [95% CI 1.440-2.643], p<0.001) (Table S4).