2018 Volume 82 Issue 5 Pages 1443-1450
Background: Seasonal variations in the severity and outcomes of stroke remain unclarified.
Methods and Results: A total of 2,965 acute ischemic stroke patients from a single-center prospective registry were studied. Among the total patients, stroke onset did not vary by season, though it varied with a peak in winter when limited to patients >75 years old (P=0.026), when limited to patients with moderate-to-severe initial neurological deficits (National Institutes of Health Stroke Scale Score ≥10, P=0.014), and when limited to those with cardioembolic stroke (n=1,031, P=0.010). In 1,934 patients with noncardioembolic stroke, stroke onset did not vary by season. After multivariable adjustment, moderate-to-severe neurological deficits were more common in winter (odds ratio 1.37, 95% confidence interval 1.10–1.72) and spring (1.27, 1.01–1.60), and death at 1 year was more common in summer than in fall (1.55, 1.03–2.36); death or dependency (modified Rankin Scale score 3–6) and death or bedridden (score of 5–6) were not differently common among the seasons.
Conclusions: Overall ischemic stroke showed a fairly even distribution among the 4 seasons. Cardioembolic stroke was more common in winter. Ischemic stroke patients had more moderate-to-severe initial neurological deficits in winter and spring. Poor clinical outcomes at 1 year were generally similar among the seasons. Ischemic stroke is not necessarily a winter-dominant disease.
It is essential to clarify annual trends in the incidence and severity of stroke, both for a better understanding of stroke pathophysiology and for appropriate medical planning of stroke care. Although most systemic vascular diseases have a peak in the winter months, seasonable variations in stroke remain unclarified.1 At the least, intracerebral hemorrhage has often been reported to occur predominantly in winter.1–5 However, study results on seasonal dominancy of ischemic stroke have been broadly divided: some studies showed a peak in winter,6–10 some in spring,3,11,12 some in summer,13–15 some in fall,16,17 and the others showed no seasonal variation.18–20 A case-crossover study showed a difference in seasonal variation between the sexes, with a higher risk of ischemic stroke in hotter seasons only in men.21 These discordant results may be caused by the heterogeneity of subtypes of ischemic stroke. Because most studies on this theme had community-based designs, the analysis by detailed subtype, quantitative evaluation of neurological severity and functional outcome, and analysis of the hospital care process were often insufficient. To overcome these limitations, large hospital-based registries would be useful.
The National Cerebral and Cardiovascular Center (NCVC) Stroke Registry is a single-center prospective registry of consecutive patients with acute ischemic stroke, intracerebral hemorrhage, and transient ischemic attack who are emergently managed in the stroke care unit of the institute (ClinicalTrials.gov Identifier: NCT02251665).22–24 The NCVC is a core comprehensive stroke center with 24/7 emergency service for acute stroke medicine located in a northern suburban residential area (Toyono medical care zone with a population >1 million) of Osaka prefecture in Japan. A stroke care unit in the NCVC has been active since 1978. All the acute stroke patients were emergently hospitalized and received initial care in the stroke care unit in principle. A small number of patients received initial care in other units and they were also registered in this NCVC Stroke Registry. Between 2013 and 2015, inpatients with acute ischemic stroke in the NCVC amounted to 38% of the overall ischemic stroke inpatients in the major 7 hospitals in the Toyono zone without seasonal variation: 39% in spring, and 38% each in the other seasons. The percentage of cardioembolic stroke patients to overall ischemic stroke patients in the NCVC was relatively higher as compared with the other 6 hospitals or that in nationwide registries.25
Japan has 4 clear-cut seasons with a rainy season between spring and summer and typhoons in late summer and early fall. The climate of Osaka is relatively mild among Japan’s regions. The aim of the present study was to determine seasonal variations in the frequency, severity, management process, and outcomes of ischemic stroke by subtype using data from the NCVC Stroke Registry.
Consecutive patients with ischemic stroke who were emergently admitted to the NCVC from January 2011 through December 2015 were studied. Weather information for Osaka in that period is shown in Figure S1 and Figure S2. Patients were divided into 4 groups according to the season of stroke onset: winter (December–February); spring (March–May); summer (June–August); and fall (September–November). The Regional Ethics Committees approved retrospective analyses using the prospective database of the NCVC Stroke Registry.
Ischemic stroke diagnosis was based on neurological and radiological examinations. Magnetic resonance imaging (MRI), including diffusion-weighted imaging and magnetic resonance angiography, was performed for all patients, except those having contraindications to MRI, using a 1.5-T or 3-T system; brain CT was used instead for patients with contraindications to MRI. Subtypes of ischemic stroke, such as cardioembolic stroke, were assessed according to the definition of the Trial of Org 10172 in Acute Stroke Treatment (TOAST)26 with detailed evaluation of emboligenic diseases principally using ECG monitoring in the stroke units, Holter monitoring, transthoracic echocardiography, and transesophageal echocardiography. Neurological deficits were quantified using the National Institutes of Health Stroke Scale (NIHSS) on emergency admission and 7 days later by experienced stroke specialists. Functional outcome was assessed using the modified Rankin Scale (mRS) score at 3 months and 1 year from stroke onset by experienced stroke specialists. The data on the NIHSS and mRS scores, as well as those on risk factors, comorbidities, and the acute management process listed in a table in the result section, were obtained from the database.
OutcomesEssential outcome measures included: seasonal variation of registered patients; moderate-to-severe initial neurological deficits assessed using an NIHSS score ≥10;27,28 neurological deterioration within the initial 7 days assessed by an increase in the NISSS score ≥4 or death;29,30 death or dependency at 1 year assessed by mRS score of 3–6; death or bedridden at 1 year assessed by mRS score of 5–6; and death at 1 year. Outcomes at 3 months were also determined. Outcomes at 3 months and 1 year were ascertained at a hospital clinic (or by phone survey by stroke neurologists or trained research nurses for patients with too severe after-effects to visit the clinic). Patients with a prestroke mRS score of 3–5 were excluded for assessment of mRS scores of 3–6, and those with a prestroke mRS score of 5 were excluded for assessment of mRS scores of 5–6.
Statistical AnalysisData are presented as mean±SD, median values (interquartile range), or number (%). Whether the proportion of patients among the 4 seasonal groups was equal was assessed using the chi-squared test. Underlying characteristics and stroke features were compared among the 4 groups using the chi-squared test, unpaired t-test, and Wilcoxon’s test, as appropriate. Multivariable logistic regression analysis with the forced entry method was performed to identify seasons associated with the aforementioned outcome measures. Age and sex were used as adjusting factors for analysis of initial neurological severity, and age, sex, baseline NIHSS score, and acute endovascular therapy as adjusting factors for the other analyses. All statistical analyses were conducted using JMP 12 statistical software (SAS Institute, Cary, NC, USA). A P-value <0.05 was considered significant.
A total of 2,965 patients (74.1±12.3 years old, 1,170 women) were studied. Overall, 759 patients [25.6%, 95% confidence interval (CI) 24.1–27.2%] developed stroke in winter, 745 (25.1%, 95% CI 23.6–26.7%) in spring, 755 (25.5%, 95% CI 23.9–27.1%) in summer, and 706 (23.8%, 95% CI 22.3–25.4%) in fall; the distribution of patients among the seasons was even, with a gentle trough in fall (P=0.498, Table 1, Figure 1). Seasonal variation showed significance with a peak in winter when limited to the older half of patients (>75 years old, P=0.026) or when limited to patients with moderate-to-severe initial neurological deficits (P=0.014). Underlying characteristics, stroke features, and the acute management process are listed in Table 2. Patients were relatively older (P=0.042), had lower systolic blood pressure (P=0.026), and more commonly had atrial fibrillation (P=0.002) and cardioembolism as a subtype (P=0.002) in colder seasons. Acute endovascular therapy was most commonly performed in fall (P=0.021).
| Total | Winter | Spring | Summer | Fall | P value | |
|---|---|---|---|---|---|---|
| Overall ischemic stroke | 2,965 | 759, 25.6% [24.1–27.2%] |
745, 25.1% [23.6–26.7%] |
755, 25.5% [23.9–27.1%] |
706, 23.8% [22.3–25.4%] |
0.498 |
| Women | 1,170 | 311, 26.6% [24.1–29.2%] |
281, 24.0% [21.7–26.5%] |
290, 24.8% [22.4–27.3%] |
288, 24.6% [22.2–27.2%] |
0.634 |
| Men | 1,795 | 448, 25.0% [23.0–27.0%] |
464, 25.9% [23.9–27.9%] |
465, 25.9% [23.9–28.0%] |
418, 23.3% [21.4–25.3%] |
0.360 |
| ≤75 (median) years old | 1,483 | 361, 24.3% [22.2–26.6%] |
352, 23.7% [21.6–26.0%] |
388, 26.2% [24.0–28.5%] |
382, 25.8% [23.6–28.0%] |
0.503 |
| >75 years old | 1,482 | 398, 26.9% [24.7–29.2%] |
393, 26.5% [24.3–28.8%] |
367, 24.8% [22.6–27.0%] |
324, 21.9% [19.8–24.0%] |
0.026 |
| Moderate-to-severe initial neurological deficits |
873 | 249, 28.5% [25.6–31.6%] |
231, 26.5% [23.6–29.5%] |
208, 23.8% [21.1–26.8%] |
185, 21.2% [18.6–24.0%] |
0.014 |
| Mild initial neurological deficits | 2,092 | 510, 24.4% [22.6–26.3%] |
514, 24.6% [22.8–26.5%] |
547, 26.1% [24.3–28.1%] |
521, 24.9% [23.1–26.8%] |
0.662 |
| Cardioembolic stroke | 1,031 | 304, 29.5% [26.8–32.3%] |
248, 24.1% [21.5–26.8%] |
239, 23.2% [20.7–25.9%] |
240, 23.3% [20.8–26.0%] |
0.010 |
| Overall noncardioembolic stroke | 1,934 | 455, 23.5% [21.7–25.5%] |
497, 25.7% [23.8–27.7%] |
516, 26.7% [24.8–28.7%] |
466, 24.1% [22.2–26.1%] |
0.182 |
| Large-artery atherosclerosis | 427 | 120, 28.1% [24.0–32.5%] |
104, 24.3% [20.5–28.6%] |
107, 25.1% [21.2–29.4%] |
96, 22.5% [18.8–26.7%] |
0.424 |
| Small-artery disease | 511 | 117, 22.9% [19.5–26.7%] |
131, 25.6% [22.0–29.6%] |
137, 26.8% [23.2–30.8%] |
126, 24.7% [21.1–28.6%] |
0.639 |
| Other | 996 | 218, 21.9% [19.4–24.6%] |
262, 26.3% [23.7–29.1%] |
272, 27.3% [24.6–30.2%] |
244, 24.5% [21.9–27.3%] |
0.079 |
Data are presented as number (%) [95% confidence interval].
Seasonal frequency of overall patients with ischemic stroke classified by subtype. LAA, large-artery atherosclerosis.
| Total (n=2,965) |
Winter (n=759) |
Spring (n=745) |
Summer (n=755) |
Fall (n=706) |
P value | |
|---|---|---|---|---|---|---|
| Women | 1,170 (39.5) | 311 (41.0) | 281 (37.7) | 290 (38.4) | 288 (40.8) | 0.467 |
| Age, years | 74.1±12.3 | 74.8±12.4 | 74.7±11.9 | 73.4±12.8 | 73.6±12.0 | 0.042 |
| BMI (kg/m2) | 22.4±3.8 | 22.1±3.9 | 22.5±3.7 | 22.5±3.8 | 22.5±3.9 | 0.133 |
| Systolic BP, mmHg | 160.1±29.1 | 157.5±30.2 | 160.2±28.1 | 161.0±29.0 | 161.8±28.8 | 0.026 |
| Diastolic BP, mmHg | 87.6±18.0 | 86.8±18.0 | 87.3±17.7 | 88.2±18.2 | 88.2±18.0 | 0.350 |
| Risk factors | ||||||
| Hypertension | 2,299 (77.5) | 577 (76.0) | 572 (76.8) | 595 (78.8) | 555 (78.6) | 0.491 |
| Diabetes | 736 (24.8) | 174 (22.9) | 197 (26.4) | 181 (24.0) | 184 (26.1) | 0.336 |
| Dyslipidemia | 1,504 (50.7) | 369 (48.6) | 367 (49.3) | 386 (51.1) | 382 (54.1) | 0.153 |
| Atrial fibrillation | 1,041 (35.1) | 306 (40.3) | 256 (34.4) | 233 (30.9) | 246 (34.8) | 0.002 |
| Current smoking | 541 (18.3) | 126 (16.6) | 136 (18.3) | 146 (19.3) | 133 (18.8) | 0.543 |
| Current drinking | 1,122 (37.8) | 269 (35.4) | 272 (36.5) | 309 (40.9) | 272 (38.5) | 0.132 |
| Comorbidity | ||||||
| Prior ischemic stroke | 831 (28.0) | 224 (29.5) | 202 (27.1) | 222 (29.4) | 183 (25.9) | 0.334 |
| Prior intracerebral hemorrhage | 158 (5.3) | 39 (5.1) | 44 (5.9) | 41 (5.4) | 34 (4.8) | 0.818 |
| Ischemic heart disease | 336 (11.3) | 85 (11.2) | 83 (11.1) | 87 (11.5) | 81 (11.5) | 0.994 |
| Stroke subtype | 0.020 | |||||
| Cardioembolism | 1,031 (34.8) | 304 (40.1) | 248 (33.3) | 239 (31.7) | 240 (34.0) | |
| Large-artery atherosclerosis | 427 (14.4) | 120 (15.8) | 104 (14.0) | 107 (14.2) | 96 (13.6) | |
| Small-artery disease | 511 (17.2) | 117 (15.4) | 131 (17.6) | 137 (18.2) | 126 (17.9) | |
| Other determined or undetermined etiology |
996 (33.6) | 218 (28.7) | 262 (35.2) | 272 (36.0) | 244 (34.6) | |
| Infarct location | ||||||
| Infratentorial* | 662 (22.3) | 168 (22.1) | 162 (21.7) | 179 (23.7) | 153 (21.7) | 0.769 |
| Acute reperfusion therapy | ||||||
| Any of the following 2 therapies | 438 (14.8) | 111 (14.6) | 99 (13.3) | 111 (14.7) | 117 (16.6) | 0.371 |
| Intravenous thrombolysis | 379 (12.8) | 97 (12.8) | 92 (12.3) | 97 (12.9) | 93 (12.8) | 0.974 |
| Acute endovascular therapy | 148 (5.0) | 33 (4.4) | 31 (4.2) | 33 (4.4) | 51 (7.2) | 0.021 |
Data are presented as mean±SD or number (%). *Including patients with both infratentorial and supratentorial lesions. BMI, body mass index; BP, blood pressure.
NIHSS and mRS scores in each season are shown in Table 3 and Figure 2. The overall median NIHSS score was 4 [interquartile range (IQR) 2–12] at baseline and 2 [1–7] at 7 days. Moderate-to-severe initial neurological deficits were identified in 873 patients (29.4%) and showed significant seasonal variation, with a peak in winter (P=0.019). Early neurological deterioration was identified in 178 patients (6.0%) and did not show seasonal variation. A 3-month follow-up was done in 2,728 patients (95.2%): the mean mRS score was 2 [IQR 1–4], and it was marginally significantly higher in winter (P=0.054). A 1-year follow-up was done in 2,412 patients (81.3%): the mean mRS score was also 2 [IQR 1–4], and it was insignificantly higher in winter (P=0.096).
| Total (n=2,965) |
Winter (n=759) |
Spring (n=745) |
Summer (n=755) |
Fall (n=706) |
P value | |
|---|---|---|---|---|---|---|
| National Institutes of Health Stroke Scale | ||||||
| Score at baseline, n=2,965 | 4 [2–12] (8.2±8.9) |
4 [2–14] (8.8±9.4) |
4 [2–13.5] (8.5±9.1) |
4 [2–11] (7.7±8.4) |
4 [2–10] (7.8±8.6) |
0.107 (0.027) |
| Moderate-to-severe neurological deficits* | 873 (29.4) | 249 (32.8) | 231 (31.0) | 208 (27.6) | 185 (26.2) | 0.019 |
| Score at 7 days, n=2,964 | 2 [1–7] (6.2±9.0) |
3 [1–9] (6.7±9.5) |
2 [1–8] (6.5±9.3) |
2 [1–7] (5.9±8.7) |
2 [1–7] (5.5±8.1) |
0.085 (0.043) |
| Neurological deterioration† | 178 (6.0) | 51 (6.7) | 46 (6.2) | 43 (5.7) | 38 (5.4) | 0.723 |
| Modified Rankin Scale | ||||||
| Score at 3 months, n=2,728 | 2 [1–4] (2.2±1.9) |
2 [1–4] (2.4±1.9) |
2 [1–4] (2.2±1.9) |
2 [1–4] (2.1±1.9) |
2 [1–4] (2.1±1.8) |
0.054 (0.057) |
| 3–6 at 3 months, n=2,280‡ | 640 (28.1) | 181 (31.3) | 161 (27.8) | 148 (25.8) | 150 (27.3) | 0.201 |
| 5–6 at 3 months, n=2,687§ | 404 (15.0) | 118 (17.0) | 111 (16.4) | 91 (13.5) | 84 (13.1) | 0.097 |
| 6 (death) at 3 months, n=2,728 | 94 (4.5) | 26 (4.9) | 19 (3.5) | 26 (4.8) | 23 (4.6) | 0.674 |
| Score at 1 year, n=2,412 | 2 [1–4] (2.3±2.1) |
2 [1–4] (2.4±2.1) |
1 [0–4] (2.2±2.1) |
2 [0–4] (2.3±2.1) |
1 [0–4] (2.1±2.0) |
0.096 (0.134) |
| 3–6 at 1 year, n=2,027‡ | 554 (27.3) | 154 (30.0) | 148 (27.1) | 134 (26.5) | 118 (26.0) | 0.591 |
| 5–6 at 1 year, n=2,379§ | 471 (19.5) | 125 (20.0) | 124 (19.7) | 122 (18.7) | 85 (16.0) | 0.212 |
| 6 at 1 year, n=2,412 | 318 (13.2) | 86 (13.6) | 88 (13.8) | 87 (14.3) | 57 (10.7) | 0.296 |
Data are presented as mean±SD, median [interquartile range], or number (%). *Baseline score ≥10, †increase in the total score ≥4 or death at day 7. ‡Patients with prestroke modified Rankin Scale score 3–5 were excluded. §Patients with prestroke modified Rankin Scale score 5 were excluded.
Baseline National Institutes of Health Stroke Scale (NIHSS, Left) and 1-year modified Rankin Scale (mRS, Right) scores in the overall patients.
After multivariable adjustment, moderate-to-severe initial neurological deficits were more common in winter [odds ratio (OR) 1.37, 95% CI 1.08–1.72] and spring (1.28, 1.01–1.62) than in fall (Table 4), and death at 1 year (1.55, 1.03–2.36) was more common in summer than in fall (Table 5). Death or dependency (mRS 3–6) and death or bedridden (mRS 5–6) were not differently common among seasons. When limited to the patients with moderate-to-severe initial neurological deficits, 1-year outcomes did not show differences among the seasons. No significant seasonal variation was shown in 3-month outcomes after multivariable adjustment (data not shown).
| Winter | Spring | Summer | |
|---|---|---|---|
| Moderate-to-severe initial neurological deficits [baseline NIHSS score ≥10] | |||
| Overall ischemic stroke | |||
| Crude | 1.37 (1.10–1.72)* | 1.27 (1.01–1.60)* | 1.07 (0.85–1.35) |
| Adjusted by age and sex | 1.37 (1.08–1.72)* | 1.28 (1.01–1.62)* | 1.10 (0.87–1.39) |
| Cardioembolic stroke | |||
| Crude | 1.05 (0.75–1.47) | 1.30 (0.91–1.85) | 1.03 (0.72–1.47) |
| Adjusted by age and sex | 1.07 (0.75–1.51) | 1.31 (0.91–1.90) | 1.09 (0.75–1.57) |
| Noncardioembolic stroke | |||
| Crude | 1.61 (1.12–2.31)* | 1.42 (1.00–2.05) | 1.28 (0.90–1.85) |
| Adjusted by age and sex | 1.62 (1.13–2.33)* | 1.45 (1.01–2.08)* | 1.30 (0.91–1.87) |
| Neurological deterioration (increase in NIHSS score ≥4 or death at day 7) | |||
| Overall ischemic stroke | |||
| Crude | 1.26 (0.82–1.96) | 1.16 (0.74–1.81) | 1.06 (0.68–1.67) |
| Adjusted† | 1.16 (0.74–1.81) | 1.09 (0.69–1.71) | 1.08 (0.69–1.72) |
| Cardioembolic stroke | |||
| Crude | 1.43 (0.72–2.96) | 1.96 (0.99–4.04) | 1.00 (0.45–2.23) |
| Adjusted† | 1.36 (0.67–2.89) | 1.86 (0.92–3.93) | 1.05 (0.46–2.39) |
| Noncardioembolic stroke | |||
| Crude | 1.15 (0.66–2.02) | 0.78 (0.42–1.41) | 1.09 (0.63–1.89) |
| Adjusted† | 1.08 (0.61–1.92) | 0.75 (0.40–1.37) | 1.09 (0.62–1.90) |
*P<0.05, †adjusted by age, sex, baseline NIHSS, and acute endovascular therapy. NIHSS, National Institutes of Health Stroke Scale.
| Winter | Spring | Summer | |
|---|---|---|---|
| Death or dependency: mRS score 3–6 | |||
| Overall ischemic stroke | |||
| Crude | 1.19 (0.90–1.59) | 1.06 (0.80–1.41) | 1.03 (0.77–1.37) |
| Adjusted† | 0.92 (0.64–1.31) | 0.88 (0.62–1.25) | 1.07 (0.76–1.52) |
| Cardioembolic stroke | |||
| Crude | 0.93 (0.61–1.42) | 0.99 (0.64–1.54) | 1.04 (0.66–1.65) |
| Adjusted† | 1.03 (0.60–1.77) | 0.88 (0.50–1.55) | 1.19 (0.67–2.13) |
| Noncardioembolic stroke | |||
| Crude | 1.07 (0.72–1.61) | 1.09 (0.74–1.62) | 1.13 (0.77–1.67) |
| Adjusted† | 0.86 (0.53–1.36) | 0.89 (0.57–1.38) | 0.99 (0.64–1.55) |
| Moderate-to-severe initial neurological deficits | |||
| Crude | 1.54 (0.94–2.52) | 1.32 (0.81–2.14) | 1.42 (0.86–2.36) |
| Adjusted† | 1.20 (0.66–2.20) | 1.19 (0.65–2.18) | 1.44 (0.77–2.71) |
| Death or bedridden: mRS score 5–6 | |||
| Overall ischemic stroke | |||
| Crude | 1.36 (1.00–1.85)* | 1.33 (0.98–1.81) | 1.24 (0.91–1.70) |
| Adjusted† | 1.09 (0.73–1.62) | 1.25 (0.84–1.86) | 1.42 (0.95–2.12) |
| Cardioembolic stroke | |||
| Crude | 0.98 (0.65–1.50) | 1.14 (0.74–1.76) | 1.27 (0.82–1.97) |
| Adjusted† | 0.94 (0.55–1.60) | 1.00 (0.58–1.75) | 1.61 (0.92–2.84) |
| Noncardioembolic stroke | |||
| Crude | 1.68 (1.04–2.75)* | 1.69 (1.06–2.75)* | 1.42 (0.88–2.33) |
| Adjusted† | 1.26 (0.70–2.29) | 1.54 (0.88–2.75) | 1.27 (0.72–2.30) |
| Moderate-to-severe initial neurological deficits | |||
| Crude | 1.11 (0.72–1.72) | 1.36 (0.87–2.11) | 1.16 (0.74–1.83) |
| Adjusted† | 0.93 (0.54–1.58) | 1.22 (0.71–2.11) | 1.22 (0.70–2.15) |
| Death: mRS score 6 | |||
| Overall ischemic stroke | |||
| Crude | 1.31 (0.92–1.88) | 1.33 (0.93–1.90) | 1.38 (0.97–1.98) |
| Adjusted† | 1.07 (0.71–1.63) | 1.19 (0.78–1.81) | 1.55 (1.03–2.36)* |
| Cardioembolic stroke | |||
| Crude | 0.96 (0.60–1.54) | 1.41 (0.88–2.27) | 1.33 (0.82–2.18) |
| Adjusted† | 0.93 (0.53–1.62) | 1.34 (0.77–2.36) | 1.60 (0.90–2.86) |
| Noncardioembolic stroke | |||
| Crude | 1.67 (0.95–3.02) | 1.33 (0.75–2.41) | 1.65 (0.95–2.94) |
| Adjusted† | 1.28 (0.68–2.43) | 1.01 (0.53–1.93) | 1.55 (0.85–2.90) |
| Moderate-to-severe initial neurological deficits | |||
| Crude | 1.09 (0.68–1.75) | 1.27 (0.79–2.05) | 1.18 (0.73–1.94) |
| Adjusted† | 0.99 (0.58–1.70) | 1.17 (0.69–2.01) | 1.33 (0.76–2.33) |
*P<0.05, †adjusted by age, sex, baseline National Institutes of Health Stroke Scale Score, and acute endovascular therapy. mRS, modified Rankin Scale.
Of the total 2,965 patients studied, 1,031 developed cardioembolic stroke (34.8%, 77.8±11.5 years old, 500 women). These patients developed cardioembolic stroke with a clear seasonal variation, showing an excess in winter (29.5%, 95% CI 26.8–32.3%, P=0.010, Table 1).
NIHSS and mRS scores in each season are shown in Table S1. No variables listed in the table showed significant differences among the seasons.
After multivariable adjustment, no significant seasonal variation was shown in NIHSS score (Table 4) or outcomes at 1 year (Table 5).
Analysis of Patients With Noncardioembolic StrokeOf the total patients, 1,934 developed noncardioembolic stroke (65.2%, 72.2±12.3 years old, 670 women), and they were further classified into 427 patients with large-artery atherosclerosis, 511 with small-artery disease, and 996 with stroke of other determined or undetermined etiology. Patients with noncardioembolic stroke were younger and included more men than those with cardioembolic stroke (P<0.001 for both). There was no significant seasonal variation, with a gentle peak in summer (Table 1, Figure 1).
NIHSS and mRS scores in each season are shown in Table S2. Death at 3 months showed a significant seasonal variation, with dual peaks in winter and summer (P=0.006). When assessing patients with large-artery atherosclerosis, those with small-artery disease, or those with other stroke, there were no variables showing significant seasonal variation (data not shown).
After multivariable adjustment, moderate-to-severe neurological deficits were more common in winter (OR 1.62, 95% CI 1.13–2.33) and spring (1.45, 1.01–2.08) than in fall (Table 4), and no significant seasonal variation was seen in mRS scores (Table 5). When assessing patients in 3 separate subgroups, no outcome measures showed significant seasonal variation or the analysis was incomplete because of the small patient numbers (data not shown).
Whether ischemic stroke shows seasonal variation with a peak in winter months, like most systemic vascular diseases, was evaluated in the present study using a single-center registry. The major findings included, first, that stroke onset did not vary by season in patients with ischemic stroke overall, though it varied with a peak in winter when limited to patients >75 years old, when limited to those with moderate-to-severe initial neurological deficits, and when limited to those with cardioembolic stroke. Second, in patients overall, as well as in patients with noncardioembolic stroke, more moderate-to-severe initial neurological deficits were seen in winter and spring than in fall. Third, death at 1 year was more common in summer than in fall in the patients overall, although the relationship no longer existed in the analysis after exclusion of mild stroke patients. This is the first observational study that has assessed seasonal variations of long-term functional stroke outcomes.
Peak seasons for ischemic stroke have varied among reports in the past 30 years,3,6–21 though around half showed a winter-excess.1 The discordance of findings might be partly related to differences in races, countries, population size, hospital-based or community-based designs, and stroke subtypes among reports. Many reports assessed stroke onset, and some evaluated death after stroke, but few evaluated initial neurological severity or chronic functional outcomes as indicators of seasonal variation. The strength of the present study was the large number of registered patients with full MRI-based examinations for diagnosis of stroke subtype, quantitative evaluation of neurological deficits (NIHSS) and chronic functional outcomes (mRS), and a relatively high follow-up rate at 1 year, exceeding four-fifths.
When the present patients were divided into cardioembolic and noncardioembolic stroke groups, the former showed a winter-excess frequency, and the latter showed an insignificant peak in summer. Patients with large-artery atherosclerosis had an insignificant peak in winter, and both those with small-artery disease and those with stroke of other etiology had insignificant peaks in summer, as shown in Table 1. Partly because cardioembolic patients were 5.6 years older on average than the other patients in the present study, the older half of the overall patients showed a winter-excess in onset. Similarly, partly because cardioembolic patients showed more severe neurological deficits than the other patients in the present study, patients with moderate-to-severe deficits showed a winter-excess in onset. The winter-excess in cardioembolic stroke can be partly explained by peaks occurring in new atrial fibrillation in winter, with troughs in summer.31–34 These findings can explain why previous findings on seasonal proportions were discordant. The age- and subtype-composition of each study would have a large effect on the seasonal variation of stroke onset. Most community-based studies did not refer to subtypes of ischemic stroke. In hospital-based studies with analysis based on subtypes, cardioembolic stroke was generally most common in winter, and other subtypes of stroke were common in warmer seasons.15,35,36 However, one should be careful about discussing seasonal variation of stroke onset based on single-center studies, because there might be some differences in the prehospital transportation of stroke patients in medical care zones between peak and trough times. Fortunately, the percentage of patients with acute ischemic stroke at NCVC to those in the medical care zone consistently amounted to 38–39% in each season.
Our nationwide research involving 12,660 patients with ischemic stroke over Japan was the first large-population study to use the NIHSS to assess seasonal variation in neurological severity, and it did not show any seasonal differences.15 In contrast, the present categorical classification of the NIHSS score into ≥10 and <10 showed that neurologically moderate-to-severe patients at the time of their emergency visits were more common in winter even after adjustment by age and sex. This tendency was not observed in patients with cardioembolic stroke. Because the percentage of moderate-to-severe patients was very different between cardioembolic patients (54.5%) and the others (16.1%), there might be a kind of ceiling effect in cardioembolic patients.
The aforementioned nationwide research was also the first to use the mRS for a large number of patients to assess seasonal variation in functional outcomes, and it did not show any seasonal differences in the mRS scores at acute hospital discharge.15 Most previous studies of seasonal variation of stroke outcomes focused on mortality, which was generally highest in winter.8,37–39 In our multivariable analysis, summer was the season with the highest number of deaths, which might be caused by technical problems with the statistical analysis; the predominance of summer was absent in the analysis after exclusion of mild stroke patients.
To determine the association of season with stroke outcomes, it would be important to check seasonal differences in the therapeutic process that may have a great influence on outcomes. In Japan, the use of intravenous alteplase within 4.5 h of symptom onset became covered by insurance in August, 2012.40 The enlargement of the therapeutic time window did not influence seasonal differences in thrombolysis in this cohort. Merci Retriever, the first intracranial thrombectomy device, was officially approved in Japan in 2010, and stent retrievers were approved in 2014.41 These trends seemed to result in higher percentages of patients receiving endovascular therapy in the latter season of the year. At the least, the total patients receiving either thrombolysis or acute endovascular therapy were similar among the 4 seasons in the present cohort.
Study LimitationsThese included, first, a single-center study design and ethnicity limited to Japanese, which would make generalization of the present findings to other institutes or countries difficult. A single-center study might cause selection bias. Second, the season might be a rougher explanatory variable than other meteorological conditions such as temperature, atmospheric pressure, relative humidity, and newer social environments such as air pollution.16,18–20,36,42 Third, it was difficult to clarify the definite results of noncardioembolic stroke patients divided by 3 further subtypes because of the small patient numbers. Fourth, recent therapeutic strategies such as endovascular thrombectomy and direct oral anticoagulants were adopted rapidly during the 5-year study period. These changes might have affected the present results for long-term outcomes.
Ischemic stroke showed different seasonal variations according to the subtype of ischemic stroke and outcome measures. Overall, ischemic stroke showed a fairly even distribution among the 4 seasons, though the proportion varied, with a peak in winter, when limited to cardioembolic stroke patients. Overall, the ischemic stroke patients, as well as noncardioembolic stroke patients, had more moderate-to-severe neurological deficits in winter and spring. Poor clinical outcomes at 1 year were generally similar among the seasons. Thus, ischemic stroke cannot simply be recognized as a winter-dominant disease.
The authors thank the Toyono Rehabilitation Hospital Liaison Council for their data on the medical environment.
K.T. has received honorarium from Bayer, BMS, Boehringer Ingelheim and Daiichi Sankyo.
Study Funded by an Intramural Research Fund (H28-4-1) for Cardiovascular Diseases of the National Cerebral and Cardiovascular Center, Grants from the Japan Agency for Medical Research and Development (AMED: 17ek0210091 h0001, 17ek0210055 h0001), and JSPS KAKENHI JP17H04308.
Supplementary File 1
Figure S1. Average air temperature in Osaka between January 2011 and December 2015: comparison with 3 famous cities having similar trends of temperature.
Figure S2. Average humidity and rainfall in Osaka between January 2011 and December 2015.
Table S1. Initial neurological severity and long-term outcomes in patients with cardioembolic stroke
Table S2. Initial neurological severity and long-term outcomes in patients with noncardioembolic Stroke
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-17-1310
