- Email: [email protected]
A High-Resolution Analysis on the Meteorological Influences on Spontaneous Intracerebral Hemorrhage Incidence
Original Article
A High-Resolution Analysis on the Meteorological Influences on Spontaneous Intracerebral Hemorrhage Incidence Marian Christoph Neidert1, Michael Sprenger3, Marius Mader1, Giuseppe Esposito1, Jonas Aurel Hosp1,2, Oliver Bozinov1, Luca Regli1, Jan-Karl Burkhardt1
OBJECTIVE: To investigate the influence of meteorological factors on the incidence of intracerebral hemorrhage (ICH).
-
METHODS: We included 295 ICH admissions between 2005 and 2013. The hourly meteorological parameters considered were surface pressure, temperature, relative humidity, wind gusts, and precipitation. To minimize confounding effects of seasonality, data were analyzed with the following 3 complementary statistical approaches: 1) deviation of daily measure from the 10-year monthly mean at the day of ictus; 2) deviation from monthly average with respect to changes in daily measures between the day of ictus and 2 days before; and 3) evolution of daily measures from 5 days before to 5 days after the ICH occurred. For 1) and 2), the statistical significance of the results was determined with a Monte Carlo simulation combined with a resampling technique (10003).
-
RESULTS: Regarding all patients, no statistically significant and meteorologically meaningful signal could be found. With respect to subgroup-analysis, ICH related to vascular pathologies occurred significantly more frequently at days with especially low relative humidity, whereas an opposite relation was present in patients with cerebral amyloid angiopathy. However, as absolute deviations were small, a strong functional effect is questionable. With respect to seasonal distribution, a greater incidence of ICH could be detected during the cold season, in line with previous reports.
-
CONCLUSIONS: By using high-quality meteorological data analyzed with a sophisticated and robust statistical method that minimizes the confounding effect of seasonality, no clearly identifiable meteorological influence for the ICH events considered can be found.
-
INTRODUCTION
H
emorrhagic stroke due to spontaneous intracerebral hemorrhage (ICH) accounts for 10%e15% of all strokes and has an incidence of around 30 per 100,000/year.1,2 Morbidity and mortality are high, with only 38% of patients surviving the first year.3 Because elderly patients are affected mostly, the socioeconomic importance of ICH is increasing in the context of global aging. In addition to a seasonal increment in ICH incidence in the cold months,4-7 recent studies link the occurrence of ICH to various meteorological factors.8-11 Assuming causal relationships between meteorological phenomena and disease incidences, however, bears the risk of creating pseudo-correlations, because many other factors that may influence ICH may also follow a similar annual cycle, like seasonal changes of weather.12 The aim of this study was to overcome this problem of seasonality with respect of ICH occurrence and meteorology. Hourly meteorological measurements from recording sites all across Switzerland were pursued by the Swiss National Weather Service (MeteoSwiss, Zurich, Switzerland), which allowed a precise correlation to be studied between the ictus and local meteorological data with a high spatial and temporal resolution. Data were analyzed with the use of different statistical approaches to correct for seasonal effects.
Key words Hemorrhagic stroke - ICH - Intracerebral hemorrhage - Meteorology - Weather
From the 1Department of Neurosurgery and 2Vascular Neurology and Neurorehabilitation, Department of Neurology, University of Zurich, Zurich; and 3Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland
Abbreviations and Acronyms CAA: Cerebral amyloid angiopathy ICH: Intracerebral hemorrhage VASC: Vascular pathology
Citation: World Neurosurg. (2016). http://dx.doi.org/10.1016/j.wneu.2016.12.006
-
To whom correspondence should be addressed: Marian Christoph Neidert, M.D. [E-mail: [email protected]] Supplementary digital content available online.
Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
1
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
PATIENTS AND METHODS
measure the parameters considered in this study has not changed significantly within the study period.
Patients We retrospectively analyzed the medical records of 295 consecutive patients admitted to our department for spontaneous ICH between September 2005 and December 2013. Day and time-point of symptom onset (not necessarily the admission day) were considered for this analysis. Patients were included only if they had clinical and imaging findings consistent with ICH. Patients with traumatic ICH, with hemorrhage due to brain malignancies, or with ICH in the setting of aneurysmal subarachnoid hemorrhage were not included in this study. In addition, patients were excluded if the location in which the bleeding occurred was outside Switzerland or unknown. Patients also were excluded in case the day of bleeding was unclear. In patients with multiple bleeding events, only the initial event was considered. Age, sex distribution, ICH location, ICH volume, initial Glasgow Coma Scale, ICH score, and oral anticoagulation were evaluated. Ethics Statement This study was performed in accordance with institutional guidelines and federal law (waiver issued by Cantonal Ethics Committee, KEK Zürich Nr. 75-2015). Meteorological Data All meteorological data are taken from the operational measurement network of the Swiss National Weather Service (MeteoSwiss). The following fields were considered: temperature (in C), relative humidity (in %), and wind gusts (in m/s) at 2 m above ground and surface pressure (in hPa). Further information on the measurement sites and on the quality of the data can be obtained at www.meteoswiss.ch. Meteorological data were taken from the closest weather station to the geographical location where the patient experienced the ICH (Figure 1). The technology used to
4000 3750
BAR SHA
FAH
STCMOE LEI BEZ BAS PSI LAE KLO RUE REH BUS SMA DEM GOE UEB
GRE CHA CDF CRM BRL NEU MUBBER BAN FRE
PAY
GIH
ALT ENG TIT
3250
ARH STG
2750 2500
GLA CMA
CHU
2250 WFJ DAV
NAS SCU
VAB
BUF PMA
80
3000
HOE
2000 1750
SAM
1500
COV ROB
1250 1000 750
70 60 50 40 30 20 10
500 250 0
Figure 1. (A) A map of Switzerland showing the geographical distribution of ICH events (purple dots) and measurement sites (3 letter abbreviations).
www.SCIENCEDIRECT.com
90
GUT
MAH PLF AMS BRZ DIS INT GUE BOL MAE CHE ORO MLS JUN GRH BIE HIR PIO PUY ULR ABO PRE COM SBE ROE EGH DOL MTR CGI BOU AIG DIA MVE VIS CEV LOR GVE SIO CIM EVI OTL MAG EVO ATT ZER LUG GOR GEN GSB SBO
2
100
3500
STK TAE
WYN SAE MOA WAE SCM CHZ EGO AEG QUI VAD NAP LUZ PIL SPF
B
number of cases
A
Statistical Analysis The statistical analysis and the illustration of the data were performed with Matlab (MathWorks, Natick, Massachusetts, USA; www.mathworks.com) in several distinct steps. First, each ICH day was characterized by a single daily value for each meteorological parameter. For temperature, surface pressure, and relative humidity, this was the daily mean; the maximum was taken for winds gusts. In a subgroup of patients, we were able to further stratify the time of ictus into 3 groups: 6 AMe1 PM, 1 PMe8 PM, and 8 PMe6 AM, allowing for analyses with an enhanced temporal resolution. Data were analyzed thereby the following 3 different approaches. 1) To characterize the day of ictus (i.e., day 0), we calculated the deviation of the daily mean of each parameter, e.g., temperature, to the 10-year climatological monthly mean and plotted their distribution (Figure 2, left). Then, the mean of this distribution was determined, shown as the green line in Figure 2 (left). The distribution average of 0.03 C indicates that the temperature at ICH days was essentially what would be expected from a monthly climatology, i.e., they were not “unusual” in terms of meteorological conditions. To assess the statistical significance of the distribution averaged deviation, we applied a Monte Carlo simulation. We repeated the whole analysis illustrated in Figure 2 (left), leading to a distribution mean temperature, but not for the observed ICH days but for a random selection of 295 days during the years 2006e2013 (adjusted time period to ensure whole-year datasets). The 295 randomly selected days were such that we kept the same monthly frequencies as in the observed sample. Again, a distribution of mean temperature values results, although this one was
0 0
2
4
6
8
10
12
14
16
distance [km] (B) A histogram showing the distribution of distances from intracerebral hemorrhage events to the next measurement site.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 33 200
30
150 Frequency
Frequency
mean = 0.31 40
20
100
50
10
0
0 −20
−10 0 Pressure [hPa]
10
−1
0 Pressure [hPa]
160
40
P(>) = 55
35
140
30
120
25
100
Frequency
Frequency
mean = -0.03
20
80
15
60
10
40
5
20
0 −10
−5
0
5
0
10
Temperature [deg C]
−0.6 −0.4 −0.2 0 0.2 0.4 Temperature [deg C]
0.6
P(>) = 33
mean = 0.05 60
150
40
100
Frequency
Frequency
1
50
20
0
0 −4
−2
0 2 Gust [m/s]
4
6
−0.2
8
0 Gust [m/s]
0.2
60 P(>) = 73
mean = -0.48 150
50
Frequency
Frequency
40 30
20
100
50 10 0
0 −20 0 Relative Humidity [%]
20
Figure 2. Left, Histogram of observed meteorological parameters in all patients (n ¼ 295) on intracerebral hemorrhage (ICH) days relative to the 10-year monthly means. The green value gives the mean over all observed ICH days. Right, the green line corresponds to the observed mean given in
WORLD NEUROSURGERY -: ---, - 2016
−1
0
1
2
Relative Humidity [%]
the corresponding histograms on the left and the blue distribution results from a Monte Carlo simulation with 1000 resamplings. A statistical significant signal is found if the green line lies at the outer boundaries of the random distribution (P < 5 or P > 95) number. P ¼ percentile.
www.WORLDNEUROSURGERY.org
3
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 41
mean = 0.11 150
40
Frequency
Frequency
30
20
100
50 10
0
−20
−10 0 Pressure [hPa]
10
0
20
−1
0 Pressure [hPa]
1
60 P(>) = 63
mean = -0.06 150
50
Frequency
Frequency
40 30
20
100
50
10
0 −10
0 −5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
P(>) = 28
mean = 0.08 150
80
Frequency
Frequency
60
40
100
50 20
0 −10
−5
0 Gust [m/s]
0 −0.4
5
−0.2
0 0.2 Gust [m/s]
0.4
60 P(>) = 31
mean = 0.4 150
50
Frequency
Frequency
40 30 20
100
50
10 0
0 −20
0 20 Relative Humidity [%]
Figure 3. Left, distribution of the meteorological parameters observed relative to the same parameters 2 days preceding an intracerebral hemorrhage day. The green line gives the mean measure over all bleeding days. Right, the green line corresponds to the observed mean given in the corresponding histograms
4
www.SCIENCEDIRECT.com
−2
−1
0 1 Relative Humidity [%]
2
on the left and the blue distribution results from a Monte Carlo simulation with 1000 resamplings. A statistical significant signal is found if the green line lies at the outer boundaries of the random distribution (P < 5 or P > 95) number. P ¼ percentile.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
due to the randomization not related to ICH. We repeated this random analysis 1000 times (resampling method), resulting in a distribution of 1000 mean temperatures as shown in Figure 2 (right) (referred to as “random mean distribution” in the following). The temperature mean during the observed ICH days can then be compared with the distribution of the 1000 randomly determined temperature means. If the ICH mean of temperature lies outside of the 5% or 95% percentile of the random mean distribution, a statistically significant effect has been found. On the other hand, if the observed value lies near the center of the distribution, no significant meteorological influence can be inferred, as is the case in Figure 2. 2) To evaluate whether rapid weather shifts were related to ICH occurrence, we applied the same method as in 1) (Figure 3). Instead of the daily metric, we used the difference between the parameter at day 0 and 2 days before (i.e., day 2). For 9% of the overall population, another ICH event occurred at the earlier time point, i.e., that day 2 for one patient is day 0 for another patient. Because of this small degree of overlap, we are confident that we received a sufficient temporal discriminatory power to make valid statistical statements. Because we referred the daily metrics in analysis 1) and 2) to the monthly average, no further correction for the seasonal cycle was necessary. In the complementary approach, or 3), we determined the evolution of daily metrics for the preceding and subsequent 5 days and only considered the difference relative to the ICH day (Figure 4). This analysis indicates whether systematic temporal changes occurred around ICH events. Day 0 corresponds to the ICH events and the box plots for the preceding and subsequent days show the distribution of the means relative to the values at the ICH day. RESULTS In the time period between September 2005 and December 2013, 295 patients with ICH were included. The following subgroups were analyzed: patients with vascular pathologies (VASC, n ¼ 80; consisting of patients with arteriovenous malformations and cerebral cavernous malformations), patients without vascular pathologies (n ¼ 215), patients on anticoagulants (n ¼ 37), hemorrhage into the basal ganglia (n ¼ 83), lobar hemorrhage (n ¼ 118), and cerebral amyloid angiopathy (CAA; n ¼ 47). Basic patient characteristics are listed in Table 1 and exhaustive analysis of all subgroups is listed in Supplementary Figures 1e9. The distribution of the geographical locations considered is shown in a map of Switzerland in Figure 1A, and the distances of ICH ictus location and the next weather station is shown in Figure 1B. Between 2006 and 2013 (adjusted time period to ensure whole-year datasets) the median number of cases per months was 2.3. The seasonal distribution shows a peak for the first quarter of the year, i.e. the colder period (Supplementary Figure 10). Three complementary statistical analyses were performed, as outlined in the previous section: 1) the deviation from the 10-year monthly mean with resampling; 2) the change relative to the parameter’s value 2 days before ICH occurrence; and 3) the time evolution from 5 days before to 5 days after the occurrence of the ICH. Regarding the overall patient population (n ¼ 295), the
WORLD NEUROSURGERY -: ---, - 2016
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
parameters pressure, temperature, gusts, and relative humidity matched very well with climatology (Figure 2). No statistically significant signal was seen in the Monte Carlo simulations. Furthermore, no statistically significant signal for a change in the parameters considered could be seen (Figures 3 and 4); however, analysis of subgroups yielded statistically significant differences that are summarized in Figure 5. Deviation from the 10-Year Monthly Mean with Resampling In patients with vascular pathologies (n ¼ 80) a significant signal was present for relative humidity (Supplementary Figure 1). A mean 2.51% below the relative humidity that is expected from climatology was observed on ICH bleeding days (97th percentile). In contrast, no significant signal could be found in the group of patients without vascular pathologies (Supplementary Figure 2). Patients with CAA (n ¼ 47; Supplementary Figure 3) had their ICH events on days with 3.12% greater relative humidity compared with a 10-year monthly mean (first percentile). In the group of patients on oral anticoagulants at the time of ICH (n ¼ 37), a significant signal is seen only for gusts (Supplementary Figure 4). ICH bleeding days in this group were associated with significantly greater gusts of 0.76 m/s more than what can be expected from climatology (first percentile; Supplementary Figure 4). In patients with basal ganglia hemorrhages (n ¼ 83; Supplementary Figure 5) and with lobar hemorrhages (n ¼ 118; Supplementary Figure 6), no significant signals were seen in this climatology analysis. In the analysis with greater temporal resolution, we saw no significant signal in the overall patient population (n ¼ 250; high temporal resolution data were not available for all patients; Supplementary Figure 7). The signal of lower humidity in the cohort of patients with vascular pathologies (n ¼ 69; Supplementary Figure 8) could be confirmed. In patients without vascular pathologies and analyzed with high temporal resolution (n ¼ 181; Supplementary Figure 9) a new significant signal towards greater gusts (0.29 m/s; second percentile) that was not seen with lower temporal resolution could be detected. Change of Meteorological Parameters During the 2 Days Preceding a Bleeding Event In patients with vascular pathologies (n ¼ 80; Supplementary Figure 1) a significant signal towards greater humidity during the 2 days preceding an ICH event was found. The mean for relative humidity was 3.42% greater (first percentile; see Figure 5). This finding is in accordance with the results from the analysis of climatology in 1), where we saw lower relative humidity than expected from climatology. No significant signal can be found in the group of patients without vascular pathologies (see Supplementary Figure 2). Regarding the group of patients on oral anticoagulants (n ¼ 37; Supplementary Figure 4), a significant signal was seen only for pressure with greater pressure (mean 2.41 hPa, fourth percentile; see Figure 5) 2 days preceding ICH. Both in patients with basal ganglia hemorrhages (n ¼ 83; Supplementary Figure 5) and with lobar hemorrhages (n ¼ 118; Supplementary Figure 6) no significant signals were seen. In patients with CAA (n ¼ 47; see Supplementary Figure 3) we saw lower relative humidity 2 days preceding ICH (mean of 4.13% lower relative humidity; 99th
www.WORLDNEUROSURGERY.org
5
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
percentile; see Figure 5). These findings are in line with the higher relative humidity we saw in the previous climatology analysis. In the analyses with greater temporal resolution, we saw no significant signal in the overall patient population (n ¼ 250; Supplementary Figure 7). The signal of greater humidity 2 days before ICH in the cohort of patients with vascular pathologies (n ¼ 69; see Supplementary Figure 8) was even more pronounced in the high temporal resolution cohort (mean 4.22%, first percentile; Figure 5). In patients without vascular pathologies and analyzed with high temporal resolution (n ¼ 181; Supplementary Figure 9) no significant signal could be detected.
Temporal Changes 5 Days Before and 5 Days After the Bleeding Event Box plots depicting the time evolution for the parameters considered from 5 days before to 5 days after the bleeding event are depicted in for the overall patient population in Figure 4 and for all subgroups in Supplementary Figures 1e9. The signal of
greater relative humidity before and after the ICH days and an anomaly towards low relative humidity on the actual bleeding days in the group of patients with vascular pathologies particularly can be seen (Figure 5). In the group of patients with CAA, lower relative humidity is seen before and after ictus, in line with the results from the analyses in 1) and 2). DISCUSSION With respect to the overall population of patients with ICH, our analyses did not show a clearly identifiable influence of the meteorological parameters for the 295 ICH events considered. Analysis of subgroups, however, revealed significant differences in climatology that were reproducible across 3 independent analytic approaches. These differences also remained robust in the population of patients, where an especially high temporal resolution could be obtained. In patients with ICH with a vascular pathology (VASC), relative humidity was significantly lower at day 0 but significantly greater at day 2 compared with the monthly average, indicating a particular
15 Temperature [deg C]
Pressure [hPa]
20 0 −20 −40
10 Relative Humidity [%]
Gust [m/s]
0 −5 −10
50
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding Figure 4. Box plots showing the temporal evolution of the meteorological parameters conserved from 5 days before to 5 days after the bleeding event. The red bars
www.SCIENCEDIRECT.com
5
−15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
6
10
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding represent the median, the boxes depict the interquartile range and whiskers show the 99.3 percentileseoutliers are shown as red crosses.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
strong fluctuation of this parameter. For patients who suffered from an ICH based on CAA, relative humidity follows an opposite profile with a significant peek at day 0 and particular low values at day 2. How changes in relative humidity may influence the incidence of ICH in VASC and CAA patients remains speculative: greater ambient humidity can cause vasodilation in medium-sized conduit arteries13 and increasing air humidity by wearing a moisture exchange mask can lower systolic and mean arterial blood pressure.14 Thus, in turn, low air humidity might promote the occurrence of ICH by elevating blood pressure, the strongest risk factor of ICH in general.15 Air humidity was especially low when bleeding occurred in the VASC patients. Interestingly, elevated blood pressure is a less critical risk factor for the occurrence of bleeding in vascular pathologies such as arteriovenous malformations compared with other etiologies of ICH16; however, low humidity leads to enhanced blood viscosity by promoting insensible water loss,17 which could enhance sheer stress and impair autoregulation in the feeding arterial system of a malformation, thereby enhancing the risk of rupture.18,19 In patients with CAA, ICH became manifest at days with high air humidity. High air humidity reduces the capacity for evaporative heat release, thereby promoting hyperthermia20 that
consecutively compromises microvascular function and integrity of the bloodebrain barrier.21 Both factors facilitate the occurrence of hemorrhage in animal models of CAA.22,23 Furthermore, this inverse sensitivity to deflections in air humidity VASC and CAA points to the interpretation that any extreme—in direction of very low or very high relative humidity— could enhance the risk of ICH occurrence in these subgroups. In line with this hypothesis, increased incidence of hemorrhagic strokes has been reported in a retrospective study at days with very low and very high air humidity.11 It has to be taken into account, however, that the deviation in air humidity from the monthly average in patients with VASC and CAA was quite small in either direction. Small changes in relative air humidity can have large effects when average values lie close to 100%24; however, this was not the case in our study (Supplementary Figure 11), and changes in the magnitude of approximately 3% cannot lead to significant (de-)saturation. Thus, physiological effects of such small changes in air humidity are considered to be only moderate. Other significant differences, e.g., 1) faster gust and 2) greater atmospheric pressure in patients with ICH on anticoagulants, were not consistently reproducible across different analytical approaches and were therefore discarded as false-positive.
Table 1. Patient Characteristics
Sex, F/M Mean age, years ICH volume, mL
All (n [ 295)
No Vascular Pathology (n [ 215)
Vascular (n [ 80)
Anticoagulants (n [ 37)
Basal Ganglia (n [ 83)
Lobar (n [ 118)
CAA (n [ 47)
122/173 41%/59%
83/132 39%/61%
39/41 49%/51%
13/24 35%/65%
34/49 41%/59%
42/76 36%/64%
15/32 32%/68%
58.8
58.6
59.4
56.9
59.5
58.2
56.4
25.8
30.6
12.8
52.5
9.4
45.2
66.6
Supratentorial
268 (90.8%)
202 (94.0%)
66 (82.5%)
34 (91.9%)
83 (100%)
118 (100%)
47 (100%)
Infratentorial
27 (9.2%)
13 (6.0%)
14 (17.5%)
3 (8.1%)
0
0
0
91 (30.8%)
83 (38.6%)
8 (10.0%)
15 (40.5%)
83 (100%)
0
1 (2.1%)
54 (67.5%)
Location Basal ganglia Lobar
172 (58.3%)
118 (54.9%)
18 (48.6%)
0
118 (100%)
46 (97.9%)
Cerebellum
21 (7.1%)
9 (4.2%)
12 (15%)
2 (5.4%)
0
0
0
Brain stem
6 (2.0%)
4 (1.9%)
2 (2.5%)
1 (2.7%)
0
0
0
Pure IVH
5 (1.7%)
1 (0.5%)
4 (5.0%)
1 (2.7%)
0
0
0
142 (48.1%)
103 (47.9%)
39 (48.8%)
22 (59.5%)
59 (71.1%)
35 (29.7%)
12 (25.5%)
Side, L/R/B
135/150/10 46%/51%/3%
97/113/5 45%/53%/2%
38/37/5 48%/46%/6%
18/18/1 49%/49%/2%
41/42/0 49%/51%/0%
52/65/1 44%/55%/1%
17/29/1 36.2%/61.7%/2.1%
Craniotomy
109 (36.9%)
73 (34.0%)
36 (45.0%)
8 (21.6%)
14 (16.9%)
54 (45.7%)
28 (59.6%)
IVH
EVD
79 (26.8%)
55 (25.6%)
24 (30.0%)
9 (24.3%)
37 (44.6%)
14 (11.9%)
5 (10.6%)
Anticoagulants
39 (13.2%)
37 (17.2%)
2 (2.5%)
37 (100%)
15 (18.1%)
18 (15.3%)
5 (10.6%)
Antiplatelets
52 (17.6%)
46 (21.4%)
6 (7.5%)
8 (21.6%)
17 (20.5%)
27 (22.9%)
17 (36.2%)
CAA
47 (21.9%)
47 (21.9%)
0
5 (13.5%)
1 (1.2%)
46 (39.0%)
47 (100%)
Vascular pathology
80 (27.1%)
0
80 (100%)
0
0
0
0
CAA, cerebral amyloid angiopathy; F, female; M, male; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; L, left; R, right, B, bilateral; EVD, external ventricular drain.
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
7
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
2)
3)
P(>) = 1
150
Relative Humidity [%]
Frequency
100
50 50
−4
−2
0
−2 0 2 Relative Humidity [%]
150
100
50
2)
0
1
2
3
4
5
4 Day relative to bleeding
3)
P(>) = 99
Relative Humidity [%]
P(>) = 1
Frequency
CAA
Frequency
1)
100
100
50 −20
−5 −4 −3 −2 −1
0 −4
2
Relative Humidity [%]
150
0
−40
0
P(>) = 1
20
40
0
200
−0.5
2)
0 0.5 Gust [m/s]
1
P(>) = 4
20 150 0
Frequency
Frequency
VASC
150
100
1) 150
40
Frequency
P(>) = 97
200
anticoagulants
1)
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
−20
100
50 50 −40
0 −4
−2 0 Relative Humidity [%]
2
4
0
−5
0 Relative Humidity [%]
5
−5 −4 −3 −2 −1
0
1
2
3
Day relative to bleeding
4
5
0
−4
−2 0 Pressure [hPa]
2
4
Figure 5. Statistically significant signals found in the analyses 1)-3). CAA, cerebral amyloid angiopathy; VASC, patient group with vascular pathologies.
Taken together, apart from the aforementioned subgroups, no clearly identifiable and meaningful meteorological influence for the ICH events considered could be found in our study. This finding is surprising, because previous reports claim a clear influence of meteorological factors on ICH incidence. Low ambient temperature is a robust risk factor7,9,10,25,26 for ICH that certainly is associated with the increased incidence of ICH in winter,4-7 in line with the result of the seasonality analysis of our data that indicates a cluster within the “cold months” (i.e., February and March). Furthermore, ICH has been linked to high5,27 or low9,28 air pressure, high8 and low8,9 wind speed, high11 and low9,11 relative air humidity, and the occurrence of west/south-west wind and typhoon.9 These partially contradictory findings could be explained as the result of differences in local meteorological conditions as reported studies were conducted in different climate zones and continents. A “contamination” by other factors, however, like seasonal epidemic exposure to viruses, dietary changes,29 hypovitamnosis,30 or changes in daily habits31 also could lead to “pseudo-correlations” between meteorological phenomena and ICH occurrence.12 To overcome this inaccuracy of “seasonality,” we referred all data to the average of monthly climatology derived from a 10-year period as described previously. When this robust statistical approach was used, a potential relationship to ICH to meteorological conditions could be only detected for defined parameters and selected subgroups. We are aware of some limitations of our study. This study focuses on one center and a relatively small catchment area in global context (Switzerland and mostly Zurich region); however, this approach allowed us to collect high-quality local
8
www.SCIENCEDIRECT.com
meteorological data with the greatest possible spatiotemporal resolution. There are intercultural and interindividual differences in the body’s response to weather parameters, especially differences in comfort levels of humidity and temperature might exist. The data collection was performed retrospectively and the analysis is based on the day of symptom onset (not necessarily the admission day). Impacts of very short within-day and spatially confined meteorological fluctuations are theoretically still possible, but are beyond the scope of this study. Such an analysis would be much more complex to perform as it required detailed information about the exact time and location of ICH occurrences and more importantly, meteorological observational data on even shorter temporal and smaller spatial scales are not readily available. Furthermore, false-positive and false-negative results are possible and even likely, considering the number of parameters and statistical tests. However, the focus on results with similar signals in 3 independent statistical approaches adds robustness to the methods applied in this study. In addition, it is possible that other environmental factors, e.g., solar activity or lunar phase, influence ICH occurrence. Given the quality of data and the thorough methodology of this study, it seems highly unlikely that future studies on the parameters considered will show a clear impact of weather on ICH incidence. ACKNOWLEDGMENTS We thank MeteoSwiss for providing access to the observational data.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
REFERENCES 1. Labovitz DL, Halim A, Boden-Albala B, Hauser WA, Sacco RL. The incidence of deep and lobar intracerebral hemorrhage in whites, blacks, and Hispanics. Neurology. 2005;65:518-522. 2. Qureshi AI, Tuhrim S, Broderick JP, Batjer HH, Hondo H, Hanley DF. Spontaneous intracerebral hemorrhage. N Engl J Med. 2001;344:1450-1460. 3. Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993;24:796-800. 4. Jakovljevic D, Salomaa V, Sivenius J, Tamminen M, Sarti C, Salmi K, et al. Seasonal variation in the occurrence of stroke in a finnish adult population. The FINMONICA Stroke Register. Finnish monitoring trends and determinants in cardiovascular disease. Stroke. 1996; 27:1774-1779. 5. Jimenez-Conde J, Ois A, Gomis M, RodriguezCampello A, Cuadrado-Godia E, Subirana I, et al. Weather as a trigger of stroke. Daily meteorological factors and incidence of stroke subtypes. Cerebrovasc Dis. 2008;26:348-354. 6. Wang Y, Levi CR, Attia JR, D’Este CA, Spratt N, Fisher J. Seasonal variation in stroke in the Hunter Region, Australia: a 5-year hospital-based study, 1995e2000. Stroke. 2003;34:1144-1150. 7. Rothwell PM, Wroe SJ, Slattery J, Warlow CP. Is stroke incidence related to season or temperature? The Oxfordshire Community Stroke Project. Lancet. 1996;347:934-936. 8. Cevik Y, Dogan NO, Das M, Ahmedali A, Kul S, Bayram H. The association between weather conditions and stroke admissions in Turkey. Int J Biometeorol. 2015;59:899-905. 9. Kim HJ, Kim JH, Kim DR, Kang HI, Moon BG, Kim JS. Age and meteorological factors in the occurrence of spontaneous intracerebral hemorrhage in a metropolitan city. J Cerebrovasc Endovasc Neurosurg. 2014;16:209-215. 10. Magalhaes R, Silva MC, Correia M, Bailey T. Are stroke occurrence and outcome related to weather parameters? Results from a population-based study in northern Portugal. Cerebrovasc Dis. 2011; 32:542-551. 11. Slatina E, Music M, Babic N, Dervisevic A, Mujaric E, Salibasic M, et al. Correlation between
WORLD NEUROSURGERY -: ---, - 2016
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
change in air humidity and the incidence of stroke. Mater Sociomed. 2013;25:242-245. 12. Barnett AJ, Dobson AJ. Analysing Seasonal Health Data. New York, NY: Springer; 2010. 13. Zanobetti A, Luttmann-Gibson H, Horton ES, Cohen A, Coull BA, Hoffmann B, et al. Brachial artery responses to ambient pollution, temperature, and humidity in people with type 2 diabetes: a repeated-measures study. Environ Health Perspect. 2014;122:242-248. 14. Seifert J, McNair M, Declercq P, St Cyr J. A heat and moisture mask attenuates cardiovascular stress during cold air exposure. Ther Adv Cardiovasc Dis. 2013;7:123-129. 15. Rymer MM. Hemorrhagic stroke: intracerebral hemorrhage. Mo Med. 2011;108:50-54. 16. Russin JJ, Spetzler RF, Giannotta S, Meyer FB, Lawton MT, Cohen-Gadol AA. Introduction: strategies for management and outcomes of arteriovenous malformations. Neurosurg Focus. 2014;37. Introduction; http://dx.doi.org/10.3171/ 2014.7.FOCUS14421. 17. Hashiguchi N, Takeda A, Yasuyama Y, Chishaki A, Tochihara Y. Effects of 6-h exposure to low relative humidity and low air pressure on body fluid loss and blood viscosity. Indoor Air. 2013;23:430-436. 18. Morgan M, Winder M. Haemodynamics of arteriovenous malformations of the brain and consequences of resection: a review. J Clin Neurosci. 2001; 8:216-224. 19. Rossitti S, Svendsen P. Shear stress in cerebral arteries supplying arteriovenous malformations. Acta Neurochir (Wien). 1995;137:138-145 [discussion: 145]. 20. Maughan RJ, Shirreffs SM, Ozgunen KT, Kurdak SS, Ersoz G, Binnet MS, et al. Living, training and playing in the heat: challenges to the football player and strategies for coping with environmental extremes. Scand J Med Sci Sports. 2010;20(suppl 3):117-124. 21. Bain AR, Nybo L, Ainslie PN. Cerebral vascular control and metabolism in heat stress. Compr Physiol. 2015;5:1345-1380. 22. Han BH, Zhou ML, Johnson AW, Singh I, Liao F, Vellimana AK, et al. Contribution of reactive oxygen species to cerebral amyloid angiopathy, vasomotor dysfunction, and microhemorrhage in aged tg2576 mice. Proc Natl Acad Sci U S A. 2015;112: E881-E890.
23. Tanifum EA, Starosolski ZA, Fowler SW, Jankowsky JL, Annapragada AV. Cerebral vascular leak in a mouse model of amyloid neuropathology. J Cereb Blood Flow Metab. 2014;34: 1646-1654. 24. Yau MK, Rogers RR. A Short Course in Cloud Physics. Amsterdam, Netherlands: Elsevier; 1996. 25. Passero S, Reale F, Ciacci G, Zei E. Differing temporal patterns of onset in subgroups of patients with intracerebral hemorrhage. Stroke. 2000; 31:1538-1544. 26. Shinkawa A, Ueda K, Hasuo Y, Kiyohara Y, Fujishima M. Seasonal variation in stroke incidence in Hisayama, Japan. Stroke. 1990;21: 1262-1267. 27. Nakaguchi H, Teraoka A. Relationship between the occurrence of spontaneous intracerebral hemorrhage and holidays and traditionally unlucky days in Fukuyama City, Hiroshima Prefecture, Japan. J Stroke Cerebrovasc Dis. 2007;16:194-198. 28. Dawson J, Weir C, Wright F, Bryden C, Aslanyan S, Lees K, et al. Associations between meteorological variables and acute stroke hospital admissions in the west of Scotland. Acta Neurol Scand. 2008;117:85-89. 29. Fares A. Global patterns of seasonal variation in gastrointestinal diseases. J Postgrad Med. 2013;59: 203-207. 30. Karras S, Paschou SA, Kandaraki E, Anagnostis P, Annweiler C, Tarlatzis BC, et al. Hypovitaminosis D in pregnancy in the mediterranean region: a systematic review. Eur J Clin Nutr. 2016;70:979-986. 31. Menz V, Grimm W, Hoffmann J, Maisch B. Alcohol and rhythm disturbance: the holiday heart syndrome. Herz. 1996;21:227-231.
Conflict of interest statement: The authors declare that the article content was composed in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Received 18 September 2016; accepted 2 December 2016 Citation: World Neurosurg. (2016). http://dx.doi.org/10.1016/j.wneu.2016.12.006 Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
www.WORLDNEUROSURGERY.org
9
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
SUPPLEMENTARY DATA
P(>) = 48
200
10
Frequency
Frequency
mean = 0.16
5
150 100 50
0 −20
0
−10 0 10 Pressure [hPa]
−2
0 Pressure [hPa]
2
12 mean = 0.23
P(>) = 30
150 Frequency
Frequency
10 8 6 4
100
50
2 0
mean = −0.12
−1 0 1 Temperature [deg C] P(>) = 71
150 Frequency
20 Frequency
0
−5 0 5 10 Temperature [deg C]
15 10
100
50 5 0
−2
0
2 4 Gust [m/s]
6
0
8
−0.5
12 mean = −2.51
200 Frequency
Frequency
10 8 6 4
0.5
P(>) = 97
150 100 50
2 0
0 Gust [m/s]
−20 0 Relative Humidity [%]
20
0 −4
−2 0 2 Relative Humidity [%]
Supplementary Figure 1. Patients with vascular pathologies, analysis (i), (ii) and (iii).
9.e1
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
mean = 0.85
15
P(>) = 14
10
20 Pressure [hPa]
150 Frequency
Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
100
5
50
0
0
10 0 −10 −20
0 10 Pressure [hPa] mean = −0.07
10
5
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
100
50
−5
0
0 5 10 Temperature [deg C] mean = 0.09
10
10 5 0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−0.5 0 0.5 1 Temperature [deg C] P(>) = 39
200 Frequency
15
−1
5 Gust [m/s]
0
Frequency
2
P(>) = 59
150 Frequency
Frequency
15
−1 0 1 Pressure [hPa]
Temperature [deg C]
−10
−2
150 100
5
0 −5
50 −10 −5
5
mean = 3.42
15
150 Frequency
Frequency
0 Gust [m/s]
10
−0.5
0 0.5 Gust [m/s]
P(>) = 1
100
50
5
0
0
−20
0 20 Relative Humidity [%]
0 −4
−2 0 2 4 Relative Humidity [%]
1
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding Relative Humidity [%]
0
40 20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 1. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e2
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
P(>) = 28
mean = 0.39
Frequency
20 Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
15 10
100
50
5 0
−20
−10 0 Pressure [hPa]
mean = −0.13
20
10
P(>) = 66
100
50
0 −10
0 −1
−5 0 5 10 Temperature [deg C] mean = 0.11
−0.5 0 0.5 Temperature [deg C] P(>) = 20
150 Frequency
40 Frequency
−1 0 1 Pressure [hPa]
150 Frequency
30 Frequency
0
10
30 20
100
50 10 0
−4
−2
0 2 4 Gust [m/s]
mean = 0.33
30 20
−0.2
0 0.2 Gust [m/s]
0.4
P(>) = 29
200 Frequency
Frequency
40
150 100 50
10 0
0
6
−20 0 20 Relative Humidity [%]
0
−3
−2 −1 0 1 Relative Humidity [%]
Supplementary Figure 2. Patients without vascular pathologies, analysis (i), (ii) and (iii).
9.e3
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
30 mean = −0.31
P(>) = 68
20 15 10
20 Pressure [hPa]
150 Frequency
Frequency
25
100
50
5 0
0 −20 −40
−20
−10 0 10 Pressure [hPa]
0
20
−1
0 1 Pressure [hPa]
2
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 15
20
Temperature [deg C]
30
100
50
10 0 −10
P(>) = 57
150 Frequency
Frequency
40
mean = −0.05
0
−5 0 5 Temperature [deg C]
mean = 0.08
60
−0.5
5 0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
0 0.5 Temperature [deg C]
10
P(>) = 34
200
10
40
Gust [m/s]
Frequency
Frequency
5 150 100
0 −5
20 50
−10 0 Gust [m/s]
5
mean = −0.79
20
10
0
150
−0.5
0 Gust [m/s]
100
0
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 50
P(>) = 81
50
−20 0 20 Relative Humidity [%]
0.5
Relative Humidity [%]
Frequency
30
−5
0
Frequency
0 −10
−2 0 2 Relative Humidity [%]
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 2. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e4
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
mean = −0.99
2
1
100
50
0
−20
−10 0 Pressure [hPa]
0
10
mean = −0.11
−4
−2 0 2 Pressure [hPa]
4
2
4
P(>) = 57
150 Frequency
6 Frequency
P(>) = 69
150 Frequency
3 Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
100
50
0 −10
0
−5 0 5 Temperature [deg C]
−1 0 1 Temperature [deg C]
2
12 mean = 0.22
8 6 4
P(>) = 22
150 Frequency
Frequency
10
100
50
2 0
−2
0 2 Gust [m/s]
0 −1
4
0 0.5 Gust [m/s]
1
mean = 3.12
8
150
6
Frequency
Frequency
−0.5
4
P(>) = 1
100
50
2 0
0 −20
−10 0 10 20 Relative Humidity [%]
−4 −2 0 2 Relative Humidity [%]
4
Supplementary Figure 3. Patients with cerebral, amyloid angiopathy; analysis (i), (ii) and (iii).
9.e5
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
30
mean = 1.86 150
4
100
50
2
0 10 Pressure [hPa]
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Temperature [deg C]
Frequency
4
150 100 50
2 0
−5 0 Temperature [deg C]
0
5
mean = −0.39
15
−10
P(>) = 66
200
6
0
−2 0 2 Pressure [hPa]
20
mean = −0.2
8
10
−20
0
0
Frequency
20 Pressure [hPa]
Frequency
Frequency
6
P(>) = 9
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−1 0 1 Temperature [deg C] P(>) = 87
150
10
5
Gust [m/s]
Frequency
Frequency
5 100
0
50 −5
Frequency
8
0 Gust [m/s]
0
5
mean = −4.13 150
6 4
−1
−0.5 0 0.5 Gust [m/s]
1
100
2
−10 0 10 20 Relative Humidity [%]
0
40
P(>) = 99
50
0 −20
−5 −4 −3 −2 −1 0 1 2 3 4 5 Day relative to bleeding
Relative Humidity [%]
−5
Frequency
0
−5 0 5 Relative Humidity [%]
20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 3. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e6
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = 1.51
P(>) = 20 Frequency
Frequency
3
2
100
50
1
0 −10
−5
−2
0 2 Pressure [hPa]
mean = −0.55
8 6 4
4
P(>) = 82
150 Frequency
Frequency
0
0 5 10 Pressure [hPa]
100
50
2 0
0 −2
−5 0 5 10 Temperature [deg C]
−1 0 1 Temperature [deg C]
2
6 mean = 0.76
P(>) = 1
150
4
Frequency
Frequency
5
3 2
100
50
1 0 −2
0
2 Gust [m/s]
Frequency
150
6 4
0 0.5 Gust [m/s]
1
P(>) = 12
100
50
2 0
−0.5
4
mean = 1.7
8 Frequency
0
−30
−20 −10 0 10 Relative Humidity [%]
0
−4
−2 0 2 4 Relative Humidity [%]
Supplementary Figure 4. Patients with anticoagulants, analysis (i), (ii) and (iii).
9.e7
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
20
mean = 2.41
3
P(>) = 4
2
1
10 Pressure [hPa]
200 Frequency
Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
150 100 50
0 −10 −20 −30 −40
mean = −0.15
−4
6 4
−2 0 2 Pressure [hPa]
4
100
50
2
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 10
P(>) = 62
150 Frequency
8 Frequency
0 10 Pressure [hPa]
Temperature [deg C]
0 −10
0
5 0 −5 −10
0
−5 0 Temperature [deg C]
Gust [m/s]
Frequency
5
100
50
2
0 Gust [m/s]
0
5
3 2
−1 −0.5 0 0.5 Gust [m/s]
1
100
50
1 0 −20
0
−10 0 10 Relative Humidity [%]
−5 −4 −3 −2 −1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 78
150 Frequency
4
0
−5
Relative Humidity [%]
−5
6 mean = −1.31 5 Frequency
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 91
150
4
0
−1 0 1 Temperature [deg C]
mean = −0.52
6 Frequency
0
5
−5
0 5 Relative Humidity [%]
40 20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 4. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e8
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = −0.44
P(>) = 70
150 Frequency
Frequency
8 6 4
100
50
2 0
−10
−5 0 5 Pressure [hPa]
0
10
10
5
2
P(>) = 66
200 Frequency
Frequency
mean = −0.12
−2 0 Pressure [hPa]
150 100 50
0
mean = 0.18
10
5
Frequency
15
−4
−2
0 2 4 Gust [m/s]
100
0
6
mean = −0.49
−0.5
150
10
5
0
P(>) = 20
50
Frequency
0
−1 −0.5 0 0.5 1 Temperature [deg C]
150 Frequency
15 Frequency
0
−5 0 5 10 Temperature [deg C]
0 Gust [m/s]
0.5
P(>) = 68
100
50
−20 0 20 Relative Humidity [%]
0 −4
−2 0 2 Relative Humidity [%]
Supplementary Figure 5. Patients with basal ganglia hemorrhage, analysis (i), (ii) and (iii).
9.e9
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
12
6 4
Pressure [hPa]
8
20
P(>) = 49
150 Frequency
Frequency
10
mean = 0.03
100
50
−20
0
−10 0 10 Pressure [hPa]
0 −10 −20
2 0
10
−2
−30 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
0 2 Pressure [hPa]
15 P(>) = 60
10
5
Temperature [deg C]
150 Frequency
Frequency
mean = −0.05
100
50
0
−0.5 0 0.5 1 Temperature [deg C]
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 10
P(>) = 44
150
10
−5
0 Gust [m/s]
0
5
mean = 0.57
15 10
−0.5
0 0.5 Gust [m/s]
50
P(>) = 37
150
100
50
−20 0 20 Relative Humidity [%]
0
−10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
1
5 0
0 −5
Relative Humidity [%]
20
100
50
Frequency
0
Gust [m/s]
Frequency
5
5
Frequency
−1
mean = 0.05
15 Frequency
0
−5 0 5 Temperature [deg C]
10
−4
−2 0 2 Relative Humidity [%]
4
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 5. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e10
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
12 mean = 0.68
P(>) = 23
150 Frequency
Frequency
10 8 6 4
100
50
2 0
−20
−10 0 Pressure [hPa]
mean = 0.1
−2
−1 0 1 Pressure [hPa]
15 10
2
P(>) = 41
150 Frequency
20 Frequency
0
10
100
50 5 0 −10
−5 0 5 Temperature [deg C] mean = 0.11
−0.5 0 0.5 1 Temperature [deg C] P(>) = 27
150 Frequency
20 Frequency
0 −1
10
15 10
100
50 5
Frequency
20
−4
−2
0 2 4 Gust [m/s]
mean = 0.65
150
15 10
−0.5
0 Gust [m/s]
0.5
P(>) = 25
100
50
5 0
0
6
Frequency
0
−20 0 20 Relative Humidity [%]
0
−2 0 2 Relative Humidity [%]
Supplementary Figure 6. Patients with lobar hemorrhage, analysis (i), (ii) and (iii).
9.e11
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 63
10
5
20 Pressure [hPa]
150 Frequency
Frequency
mean = −0.27
100
50
0 −20 −40
−10
20
mean = −0.23
15 10 5
0 2 Pressure [hPa]
10
150 100
0 −1
−5 0 5 Temperature [deg C]
mean = −0.04
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 75
50
0 −10
40
−2
200 Frequency
20 Frequency
0 10 Pressure [hPa]
0
Temperature [deg C]
0
0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−0.5 0 0.5 1 Temperature [deg C]
10
P(>) = 62
200
5
20
Gust [m/s]
Frequency
Frequency
5 30
150 100
0 −5
10
50
0 −10
−5
0 Gust [m/s]
0
5
−10 −0.5
0 Gust [m/s]
0.5
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 50
15 10
P(>) = 89 Relative Humidity [%]
150 Frequency
Frequency
20
mean = −1.39
100
50 5 0
−20 0 20 Relative Humidity [%]
0
−2 0 2 Relative Humidity [%]
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 6. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e12
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
35 mean = 0.32
P(>) = 33
30
200
Frequency
Frequency
25 20 15
150
100
10 50
5 0
−20
−10 0 Pressure [hPa]
−1
0 Pressure [hPa]
1
2
160
45
mean = 0.01
P(>) = 47
40
140
35
120
30 Frequency
Frequency
0 −2
10
25 20
100 80 60
15 40
10
20
5 0
0
−10
−5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
180
70 mean = 0.19
P(>) = 6
160
60
140
50 Frequency
Frequency
120
40 30
100 80 60
20 40
10 0
20
−2
0
2 4 Gust [m/s]
6
0 −0.3
8
45
mean = −0.75
0 0.1 Gust [m/s]
0.2
0.3
140
35
120
30 Frequency
Frequency
−0.1
P(>) = 82
160
40
25 20
100 80
15
60
10
40
5
20
0
−0.2
180
−30
−20
−10 0 10 20 Relative Humidity [%]
30
0
−2
−1 0 1 Relative Humidity [%]
2
Supplementary Figure 7. All patients with high temporal resolution, analysis (i), (ii) and (iii).
9.e13
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = 0.12
20
140
35
10
30 25 20
100 80
15
60
10
40
5
20
−20
−10
0 Pressure [hPa]
10
0
20
60
Pressure [hPa]
120 Frequency
Frequency
P(>) = 41
160
40
0
30
180
45
0 −10 −20 −30 −40
−1.5
−1
−0.5 0 0.5 Pressure [hPa]
1
1.5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
180
mean = −0.04
15
P(>) = 59
160
50 10
120 Frequency
Frequency
40
Temperature [deg C]
140
30
100 80 60
20
5 0 −5
40
−10
20
−15
10
0
0 −10
−5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
180
70
10
mean = −0.02
P(>) = 56
160
60
8
140
6
120
4
40 30
Gust [m/s]
Frequency
Frequency
50
100 80
0
−4
40
−6 −8
20
−5
0 Gust [m/s]
5
0 −0.4
10
60
−10
−0.2
0 0.2 Gust [m/s]
0.4
180
mean = 0.55
60 P(>) = 31
160
50
40
120 Frequency
Frequency
Relative Humidity [%]
140
40
30
100 80 60
20
0 −2
60
20 10
2
40
20
0
−20
−40
10 20
−60
0
−40
−20 0 20 Relative Humidity [%]
40
60
0
−2
−1 0 1 Relative Humidity [%]
2
3
Supplementary Figure 7. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e14
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.48
14
140
12
120
10 8
80 60
4
40
2
20 0
−15
−10
−5 0 5 Pressure [hPa]
10
15
18
−3
−2
−1 0 Pressure [hPa]
1
2
180
mean = 0.46
16 14
140
12
120
10 8
100 80
6
60
4
40
2
20
0
−5
0 5 Temperature [deg C]
P(>) = 16
160
Frequency
Frequency
100
6
0
P(>) = 33
160
Frequency
Frequency
16
0 −1.5
10
−1
−0.5 0 0.5 Temperature [deg C]
1
180
mean = −0.09
P(>) = 65
160
20
120
15 Frequency
Frequency
140
10
100 80 60 40
5
20
0
−2
0
2 Gust [m/s]
4
0 −0.6
6
12
−0.4
−0.2
0 0.2 Gust [m/s]
0.4
0.6
180
mean = −2.55
160
P(>) = 96
10 140 120 Frequency
Frequency
8
6
100 80 60
4
40
2 20
0 −30
−20
−10 0 Relative Humidity [%]
10
0 −5
0 Relative Humidity [%]
5
Supplementary Figure 8. Patients with vascular pathologies studied with high temporal resolution, analysis (i), (ii) and (iii).
9.e15
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.54
20
P(>) = 28
160
15
140
10
12
120
5
10 8
Pressure [hPa]
14
Frequency
Frequency
16
100 80
0 −5
6
60
−10
4
40
−15
2
20
−20
0
0
−15
−10
−5 0 5 Pressure [hPa]
10
15
12
−25
−3
−2
−1 0 Pressure [hPa]
1
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
4
5
2
180
mean = −0.43
P(>) = 83
160
10
10
120 Frequency
Frequency
8
Temperature [deg C]
140
6
100 80 60
4
5
0
−5
−10
40
2 20 0
0
−5
0 5 Temperature [deg C]
18
−15
−1
−0.5 0 0.5 Temperature [deg C]
1
1.5
180
mean = 0.03
16
8
P(>) = 46
160
6
14
140
12
120
10 8
100 80
6
60
4
40
2
20
0
−6
−4
−2 0 Gust [m/s]
2
Gust [m/s]
Frequency
Frequency
4
0 −2 −4 −6 −8
0 −1
4
2
−0.5
0 Gust [m/s]
0.5
18 180 P(>) = 1
160
14
40 Relative Humidity [%]
140
12
120 Frequency
Frequency
60
mean = 4.22
16
10 8
100 80
6
60
4
40
20
0
−20
20
2 0 −20
0
20 40 Relative Humidity [%]
60
0 −6
−40 −4
−2 0 2 Relative Humidity [%]
4
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
Supplementary Figure 8. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e16
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.22
140
12
120
10 8
100 80
6
60
4
40
2
20
0
P(>) = 40
160
14
Frequency
Frequency
16
0
−20
−10 0 Pressure [hPa]
−2
10
30
−1
0 Pressure [hPa]
1
2
180
mean = −0.16
P(>) = 70
160
25 140 120 Frequency
Frequency
20
15
100 80 60
10
40
5 20
0
0
−10
−5 0 5 Temperature [deg C]
−0.5 0 Temperature [deg C]
mean = 0.29 160
35
140
30
120 Frequency
40
25 20
0.5
80 60
10
40
5
20
−2
0
2 4 Gust [m/s]
6
P(>) = 2
100
15
0
−1
180
45
Frequency
10
0 −0.4
8
−0.2
0 Gust [m/s]
0.2
0.4
35 mean = −0.03
P(>) = 49
30
200
Frequency
Frequency
25 20 15
150
100
10 50
5 0
0
−30
−20
−10 0 10 20 Relative Humidity [%]
30
−3
−2
−1 0 1 Relative Humidity [%]
2
3
Supplementary Figure 9. Patients without vascular pathologies studied with high temporal resolution, analysis (i), (ii) and (iii).
9.e17
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
180
mean = −0.14
30
P(>) = 58
160
20
20 10
15
10
Pressure [hPa]
120 Frequency
Frequency
140
100 80
0 −10 −20
60 −30
40
5
−40
20
0
−20
−10
45
0 Pressure [hPa]
10
0
20
−2
−1
0 Pressure [hPa]
mean = 0.11
1
2
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
200 10
Frequency
30 25 20
Temperature [deg C]
35 Frequency
4
15
P(>) = 34
40
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
150
100
15 50
10
5 0 −5 −10
5 −15
0
−5
0 5 Temperature [deg C]
10
0 0.5 Temperature [deg C]
1
180
45
10
mean = −0.03
P(>) = 60
160
40 35 30 Frequency
Frequency
−0.5
25 20
8
140
6
120
4 Gust [m/s]
0 −10
100 80
2 0 −2
15
60
−4
10
40
−6
5
20
0
−5
0 Gust [m/s]
5
0
10
−8 −10 −0.4
−0.2
0 0.2 Gust [m/s]
0.4
0.6
60
45
mean = −0.9
P(>) = 81
40
200
40
25 20
Relative Humidity [%]
30
Frequency
Frequency
35 150
100
15 50
10
20
0
−20
−40
5 0
−40
−20 0 Relative Humidity [%]
20
40
0 −4
−60
−2
0 Relative Humidity [%]
2
Supplementary Figure 9. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e18
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
600
500
# Cases
400
300
200
100
0 20
40
60 80 Relative Humidity [%]
100
Supplementary Figure 10. Distribution of relative humidity at bleeding day.
Median Number of Cases/Month
6
5
4
3
2
1
0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Supplementary Figure 11. Seasonal distribution between 2006 and 2013 (adjusted time period to ensure whole-year datasets).
9.e19
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
A High-Resolution Analysis on the Meteorological Influences on Spontaneous Intracerebral Hemorrhage Incidence Marian Christoph Neidert1, Michael Sprenger3, Marius Mader1, Giuseppe Esposito1, Jonas Aurel Hosp1,2, Oliver Bozinov1, Luca Regli1, Jan-Karl Burkhardt1
OBJECTIVE: To investigate the influence of meteorological factors on the incidence of intracerebral hemorrhage (ICH).
-
METHODS: We included 295 ICH admissions between 2005 and 2013. The hourly meteorological parameters considered were surface pressure, temperature, relative humidity, wind gusts, and precipitation. To minimize confounding effects of seasonality, data were analyzed with the following 3 complementary statistical approaches: 1) deviation of daily measure from the 10-year monthly mean at the day of ictus; 2) deviation from monthly average with respect to changes in daily measures between the day of ictus and 2 days before; and 3) evolution of daily measures from 5 days before to 5 days after the ICH occurred. For 1) and 2), the statistical significance of the results was determined with a Monte Carlo simulation combined with a resampling technique (10003).
-
RESULTS: Regarding all patients, no statistically significant and meteorologically meaningful signal could be found. With respect to subgroup-analysis, ICH related to vascular pathologies occurred significantly more frequently at days with especially low relative humidity, whereas an opposite relation was present in patients with cerebral amyloid angiopathy. However, as absolute deviations were small, a strong functional effect is questionable. With respect to seasonal distribution, a greater incidence of ICH could be detected during the cold season, in line with previous reports.
-
CONCLUSIONS: By using high-quality meteorological data analyzed with a sophisticated and robust statistical method that minimizes the confounding effect of seasonality, no clearly identifiable meteorological influence for the ICH events considered can be found.
-
INTRODUCTION
H
emorrhagic stroke due to spontaneous intracerebral hemorrhage (ICH) accounts for 10%e15% of all strokes and has an incidence of around 30 per 100,000/year.1,2 Morbidity and mortality are high, with only 38% of patients surviving the first year.3 Because elderly patients are affected mostly, the socioeconomic importance of ICH is increasing in the context of global aging. In addition to a seasonal increment in ICH incidence in the cold months,4-7 recent studies link the occurrence of ICH to various meteorological factors.8-11 Assuming causal relationships between meteorological phenomena and disease incidences, however, bears the risk of creating pseudo-correlations, because many other factors that may influence ICH may also follow a similar annual cycle, like seasonal changes of weather.12 The aim of this study was to overcome this problem of seasonality with respect of ICH occurrence and meteorology. Hourly meteorological measurements from recording sites all across Switzerland were pursued by the Swiss National Weather Service (MeteoSwiss, Zurich, Switzerland), which allowed a precise correlation to be studied between the ictus and local meteorological data with a high spatial and temporal resolution. Data were analyzed with the use of different statistical approaches to correct for seasonal effects.
Key words Hemorrhagic stroke - ICH - Intracerebral hemorrhage - Meteorology - Weather
From the 1Department of Neurosurgery and 2Vascular Neurology and Neurorehabilitation, Department of Neurology, University of Zurich, Zurich; and 3Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland
Abbreviations and Acronyms CAA: Cerebral amyloid angiopathy ICH: Intracerebral hemorrhage VASC: Vascular pathology
Citation: World Neurosurg. (2016). http://dx.doi.org/10.1016/j.wneu.2016.12.006
-
To whom correspondence should be addressed: Marian Christoph Neidert, M.D. [E-mail: [email protected]] Supplementary digital content available online.
Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
1
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
PATIENTS AND METHODS
measure the parameters considered in this study has not changed significantly within the study period.
Patients We retrospectively analyzed the medical records of 295 consecutive patients admitted to our department for spontaneous ICH between September 2005 and December 2013. Day and time-point of symptom onset (not necessarily the admission day) were considered for this analysis. Patients were included only if they had clinical and imaging findings consistent with ICH. Patients with traumatic ICH, with hemorrhage due to brain malignancies, or with ICH in the setting of aneurysmal subarachnoid hemorrhage were not included in this study. In addition, patients were excluded if the location in which the bleeding occurred was outside Switzerland or unknown. Patients also were excluded in case the day of bleeding was unclear. In patients with multiple bleeding events, only the initial event was considered. Age, sex distribution, ICH location, ICH volume, initial Glasgow Coma Scale, ICH score, and oral anticoagulation were evaluated. Ethics Statement This study was performed in accordance with institutional guidelines and federal law (waiver issued by Cantonal Ethics Committee, KEK Zürich Nr. 75-2015). Meteorological Data All meteorological data are taken from the operational measurement network of the Swiss National Weather Service (MeteoSwiss). The following fields were considered: temperature (in C), relative humidity (in %), and wind gusts (in m/s) at 2 m above ground and surface pressure (in hPa). Further information on the measurement sites and on the quality of the data can be obtained at www.meteoswiss.ch. Meteorological data were taken from the closest weather station to the geographical location where the patient experienced the ICH (Figure 1). The technology used to
4000 3750
BAR SHA
FAH
STCMOE LEI BEZ BAS PSI LAE KLO RUE REH BUS SMA DEM GOE UEB
GRE CHA CDF CRM BRL NEU MUBBER BAN FRE
PAY
GIH
ALT ENG TIT
3250
ARH STG
2750 2500
GLA CMA
CHU
2250 WFJ DAV
NAS SCU
VAB
BUF PMA
80
3000
HOE
2000 1750
SAM
1500
COV ROB
1250 1000 750
70 60 50 40 30 20 10
500 250 0
Figure 1. (A) A map of Switzerland showing the geographical distribution of ICH events (purple dots) and measurement sites (3 letter abbreviations).
www.SCIENCEDIRECT.com
90
GUT
MAH PLF AMS BRZ DIS INT GUE BOL MAE CHE ORO MLS JUN GRH BIE HIR PIO PUY ULR ABO PRE COM SBE ROE EGH DOL MTR CGI BOU AIG DIA MVE VIS CEV LOR GVE SIO CIM EVI OTL MAG EVO ATT ZER LUG GOR GEN GSB SBO
2
100
3500
STK TAE
WYN SAE MOA WAE SCM CHZ EGO AEG QUI VAD NAP LUZ PIL SPF
B
number of cases
A
Statistical Analysis The statistical analysis and the illustration of the data were performed with Matlab (MathWorks, Natick, Massachusetts, USA; www.mathworks.com) in several distinct steps. First, each ICH day was characterized by a single daily value for each meteorological parameter. For temperature, surface pressure, and relative humidity, this was the daily mean; the maximum was taken for winds gusts. In a subgroup of patients, we were able to further stratify the time of ictus into 3 groups: 6 AMe1 PM, 1 PMe8 PM, and 8 PMe6 AM, allowing for analyses with an enhanced temporal resolution. Data were analyzed thereby the following 3 different approaches. 1) To characterize the day of ictus (i.e., day 0), we calculated the deviation of the daily mean of each parameter, e.g., temperature, to the 10-year climatological monthly mean and plotted their distribution (Figure 2, left). Then, the mean of this distribution was determined, shown as the green line in Figure 2 (left). The distribution average of 0.03 C indicates that the temperature at ICH days was essentially what would be expected from a monthly climatology, i.e., they were not “unusual” in terms of meteorological conditions. To assess the statistical significance of the distribution averaged deviation, we applied a Monte Carlo simulation. We repeated the whole analysis illustrated in Figure 2 (left), leading to a distribution mean temperature, but not for the observed ICH days but for a random selection of 295 days during the years 2006e2013 (adjusted time period to ensure whole-year datasets). The 295 randomly selected days were such that we kept the same monthly frequencies as in the observed sample. Again, a distribution of mean temperature values results, although this one was
0 0
2
4
6
8
10
12
14
16
distance [km] (B) A histogram showing the distribution of distances from intracerebral hemorrhage events to the next measurement site.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 33 200
30
150 Frequency
Frequency
mean = 0.31 40
20
100
50
10
0
0 −20
−10 0 Pressure [hPa]
10
−1
0 Pressure [hPa]
160
40
P(>) = 55
35
140
30
120
25
100
Frequency
Frequency
mean = -0.03
20
80
15
60
10
40
5
20
0 −10
−5
0
5
0
10
Temperature [deg C]
−0.6 −0.4 −0.2 0 0.2 0.4 Temperature [deg C]
0.6
P(>) = 33
mean = 0.05 60
150
40
100
Frequency
Frequency
1
50
20
0
0 −4
−2
0 2 Gust [m/s]
4
6
−0.2
8
0 Gust [m/s]
0.2
60 P(>) = 73
mean = -0.48 150
50
Frequency
Frequency
40 30
20
100
50 10 0
0 −20 0 Relative Humidity [%]
20
Figure 2. Left, Histogram of observed meteorological parameters in all patients (n ¼ 295) on intracerebral hemorrhage (ICH) days relative to the 10-year monthly means. The green value gives the mean over all observed ICH days. Right, the green line corresponds to the observed mean given in
WORLD NEUROSURGERY -: ---, - 2016
−1
0
1
2
Relative Humidity [%]
the corresponding histograms on the left and the blue distribution results from a Monte Carlo simulation with 1000 resamplings. A statistical significant signal is found if the green line lies at the outer boundaries of the random distribution (P < 5 or P > 95) number. P ¼ percentile.
www.WORLDNEUROSURGERY.org
3
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 41
mean = 0.11 150
40
Frequency
Frequency
30
20
100
50 10
0
−20
−10 0 Pressure [hPa]
10
0
20
−1
0 Pressure [hPa]
1
60 P(>) = 63
mean = -0.06 150
50
Frequency
Frequency
40 30
20
100
50
10
0 −10
0 −5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
P(>) = 28
mean = 0.08 150
80
Frequency
Frequency
60
40
100
50 20
0 −10
−5
0 Gust [m/s]
0 −0.4
5
−0.2
0 0.2 Gust [m/s]
0.4
60 P(>) = 31
mean = 0.4 150
50
Frequency
Frequency
40 30 20
100
50
10 0
0 −20
0 20 Relative Humidity [%]
Figure 3. Left, distribution of the meteorological parameters observed relative to the same parameters 2 days preceding an intracerebral hemorrhage day. The green line gives the mean measure over all bleeding days. Right, the green line corresponds to the observed mean given in the corresponding histograms
4
www.SCIENCEDIRECT.com
−2
−1
0 1 Relative Humidity [%]
2
on the left and the blue distribution results from a Monte Carlo simulation with 1000 resamplings. A statistical significant signal is found if the green line lies at the outer boundaries of the random distribution (P < 5 or P > 95) number. P ¼ percentile.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
due to the randomization not related to ICH. We repeated this random analysis 1000 times (resampling method), resulting in a distribution of 1000 mean temperatures as shown in Figure 2 (right) (referred to as “random mean distribution” in the following). The temperature mean during the observed ICH days can then be compared with the distribution of the 1000 randomly determined temperature means. If the ICH mean of temperature lies outside of the 5% or 95% percentile of the random mean distribution, a statistically significant effect has been found. On the other hand, if the observed value lies near the center of the distribution, no significant meteorological influence can be inferred, as is the case in Figure 2. 2) To evaluate whether rapid weather shifts were related to ICH occurrence, we applied the same method as in 1) (Figure 3). Instead of the daily metric, we used the difference between the parameter at day 0 and 2 days before (i.e., day 2). For 9% of the overall population, another ICH event occurred at the earlier time point, i.e., that day 2 for one patient is day 0 for another patient. Because of this small degree of overlap, we are confident that we received a sufficient temporal discriminatory power to make valid statistical statements. Because we referred the daily metrics in analysis 1) and 2) to the monthly average, no further correction for the seasonal cycle was necessary. In the complementary approach, or 3), we determined the evolution of daily metrics for the preceding and subsequent 5 days and only considered the difference relative to the ICH day (Figure 4). This analysis indicates whether systematic temporal changes occurred around ICH events. Day 0 corresponds to the ICH events and the box plots for the preceding and subsequent days show the distribution of the means relative to the values at the ICH day. RESULTS In the time period between September 2005 and December 2013, 295 patients with ICH were included. The following subgroups were analyzed: patients with vascular pathologies (VASC, n ¼ 80; consisting of patients with arteriovenous malformations and cerebral cavernous malformations), patients without vascular pathologies (n ¼ 215), patients on anticoagulants (n ¼ 37), hemorrhage into the basal ganglia (n ¼ 83), lobar hemorrhage (n ¼ 118), and cerebral amyloid angiopathy (CAA; n ¼ 47). Basic patient characteristics are listed in Table 1 and exhaustive analysis of all subgroups is listed in Supplementary Figures 1e9. The distribution of the geographical locations considered is shown in a map of Switzerland in Figure 1A, and the distances of ICH ictus location and the next weather station is shown in Figure 1B. Between 2006 and 2013 (adjusted time period to ensure whole-year datasets) the median number of cases per months was 2.3. The seasonal distribution shows a peak for the first quarter of the year, i.e. the colder period (Supplementary Figure 10). Three complementary statistical analyses were performed, as outlined in the previous section: 1) the deviation from the 10-year monthly mean with resampling; 2) the change relative to the parameter’s value 2 days before ICH occurrence; and 3) the time evolution from 5 days before to 5 days after the occurrence of the ICH. Regarding the overall patient population (n ¼ 295), the
WORLD NEUROSURGERY -: ---, - 2016
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
parameters pressure, temperature, gusts, and relative humidity matched very well with climatology (Figure 2). No statistically significant signal was seen in the Monte Carlo simulations. Furthermore, no statistically significant signal for a change in the parameters considered could be seen (Figures 3 and 4); however, analysis of subgroups yielded statistically significant differences that are summarized in Figure 5. Deviation from the 10-Year Monthly Mean with Resampling In patients with vascular pathologies (n ¼ 80) a significant signal was present for relative humidity (Supplementary Figure 1). A mean 2.51% below the relative humidity that is expected from climatology was observed on ICH bleeding days (97th percentile). In contrast, no significant signal could be found in the group of patients without vascular pathologies (Supplementary Figure 2). Patients with CAA (n ¼ 47; Supplementary Figure 3) had their ICH events on days with 3.12% greater relative humidity compared with a 10-year monthly mean (first percentile). In the group of patients on oral anticoagulants at the time of ICH (n ¼ 37), a significant signal is seen only for gusts (Supplementary Figure 4). ICH bleeding days in this group were associated with significantly greater gusts of 0.76 m/s more than what can be expected from climatology (first percentile; Supplementary Figure 4). In patients with basal ganglia hemorrhages (n ¼ 83; Supplementary Figure 5) and with lobar hemorrhages (n ¼ 118; Supplementary Figure 6), no significant signals were seen in this climatology analysis. In the analysis with greater temporal resolution, we saw no significant signal in the overall patient population (n ¼ 250; high temporal resolution data were not available for all patients; Supplementary Figure 7). The signal of lower humidity in the cohort of patients with vascular pathologies (n ¼ 69; Supplementary Figure 8) could be confirmed. In patients without vascular pathologies and analyzed with high temporal resolution (n ¼ 181; Supplementary Figure 9) a new significant signal towards greater gusts (0.29 m/s; second percentile) that was not seen with lower temporal resolution could be detected. Change of Meteorological Parameters During the 2 Days Preceding a Bleeding Event In patients with vascular pathologies (n ¼ 80; Supplementary Figure 1) a significant signal towards greater humidity during the 2 days preceding an ICH event was found. The mean for relative humidity was 3.42% greater (first percentile; see Figure 5). This finding is in accordance with the results from the analysis of climatology in 1), where we saw lower relative humidity than expected from climatology. No significant signal can be found in the group of patients without vascular pathologies (see Supplementary Figure 2). Regarding the group of patients on oral anticoagulants (n ¼ 37; Supplementary Figure 4), a significant signal was seen only for pressure with greater pressure (mean 2.41 hPa, fourth percentile; see Figure 5) 2 days preceding ICH. Both in patients with basal ganglia hemorrhages (n ¼ 83; Supplementary Figure 5) and with lobar hemorrhages (n ¼ 118; Supplementary Figure 6) no significant signals were seen. In patients with CAA (n ¼ 47; see Supplementary Figure 3) we saw lower relative humidity 2 days preceding ICH (mean of 4.13% lower relative humidity; 99th
www.WORLDNEUROSURGERY.org
5
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
percentile; see Figure 5). These findings are in line with the higher relative humidity we saw in the previous climatology analysis. In the analyses with greater temporal resolution, we saw no significant signal in the overall patient population (n ¼ 250; Supplementary Figure 7). The signal of greater humidity 2 days before ICH in the cohort of patients with vascular pathologies (n ¼ 69; see Supplementary Figure 8) was even more pronounced in the high temporal resolution cohort (mean 4.22%, first percentile; Figure 5). In patients without vascular pathologies and analyzed with high temporal resolution (n ¼ 181; Supplementary Figure 9) no significant signal could be detected.
Temporal Changes 5 Days Before and 5 Days After the Bleeding Event Box plots depicting the time evolution for the parameters considered from 5 days before to 5 days after the bleeding event are depicted in for the overall patient population in Figure 4 and for all subgroups in Supplementary Figures 1e9. The signal of
greater relative humidity before and after the ICH days and an anomaly towards low relative humidity on the actual bleeding days in the group of patients with vascular pathologies particularly can be seen (Figure 5). In the group of patients with CAA, lower relative humidity is seen before and after ictus, in line with the results from the analyses in 1) and 2). DISCUSSION With respect to the overall population of patients with ICH, our analyses did not show a clearly identifiable influence of the meteorological parameters for the 295 ICH events considered. Analysis of subgroups, however, revealed significant differences in climatology that were reproducible across 3 independent analytic approaches. These differences also remained robust in the population of patients, where an especially high temporal resolution could be obtained. In patients with ICH with a vascular pathology (VASC), relative humidity was significantly lower at day 0 but significantly greater at day 2 compared with the monthly average, indicating a particular
15 Temperature [deg C]
Pressure [hPa]
20 0 −20 −40
10 Relative Humidity [%]
Gust [m/s]
0 −5 −10
50
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding Figure 4. Box plots showing the temporal evolution of the meteorological parameters conserved from 5 days before to 5 days after the bleeding event. The red bars
www.SCIENCEDIRECT.com
5
−15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
6
10
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding represent the median, the boxes depict the interquartile range and whiskers show the 99.3 percentileseoutliers are shown as red crosses.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
strong fluctuation of this parameter. For patients who suffered from an ICH based on CAA, relative humidity follows an opposite profile with a significant peek at day 0 and particular low values at day 2. How changes in relative humidity may influence the incidence of ICH in VASC and CAA patients remains speculative: greater ambient humidity can cause vasodilation in medium-sized conduit arteries13 and increasing air humidity by wearing a moisture exchange mask can lower systolic and mean arterial blood pressure.14 Thus, in turn, low air humidity might promote the occurrence of ICH by elevating blood pressure, the strongest risk factor of ICH in general.15 Air humidity was especially low when bleeding occurred in the VASC patients. Interestingly, elevated blood pressure is a less critical risk factor for the occurrence of bleeding in vascular pathologies such as arteriovenous malformations compared with other etiologies of ICH16; however, low humidity leads to enhanced blood viscosity by promoting insensible water loss,17 which could enhance sheer stress and impair autoregulation in the feeding arterial system of a malformation, thereby enhancing the risk of rupture.18,19 In patients with CAA, ICH became manifest at days with high air humidity. High air humidity reduces the capacity for evaporative heat release, thereby promoting hyperthermia20 that
consecutively compromises microvascular function and integrity of the bloodebrain barrier.21 Both factors facilitate the occurrence of hemorrhage in animal models of CAA.22,23 Furthermore, this inverse sensitivity to deflections in air humidity VASC and CAA points to the interpretation that any extreme—in direction of very low or very high relative humidity— could enhance the risk of ICH occurrence in these subgroups. In line with this hypothesis, increased incidence of hemorrhagic strokes has been reported in a retrospective study at days with very low and very high air humidity.11 It has to be taken into account, however, that the deviation in air humidity from the monthly average in patients with VASC and CAA was quite small in either direction. Small changes in relative air humidity can have large effects when average values lie close to 100%24; however, this was not the case in our study (Supplementary Figure 11), and changes in the magnitude of approximately 3% cannot lead to significant (de-)saturation. Thus, physiological effects of such small changes in air humidity are considered to be only moderate. Other significant differences, e.g., 1) faster gust and 2) greater atmospheric pressure in patients with ICH on anticoagulants, were not consistently reproducible across different analytical approaches and were therefore discarded as false-positive.
Table 1. Patient Characteristics
Sex, F/M Mean age, years ICH volume, mL
All (n [ 295)
No Vascular Pathology (n [ 215)
Vascular (n [ 80)
Anticoagulants (n [ 37)
Basal Ganglia (n [ 83)
Lobar (n [ 118)
CAA (n [ 47)
122/173 41%/59%
83/132 39%/61%
39/41 49%/51%
13/24 35%/65%
34/49 41%/59%
42/76 36%/64%
15/32 32%/68%
58.8
58.6
59.4
56.9
59.5
58.2
56.4
25.8
30.6
12.8
52.5
9.4
45.2
66.6
Supratentorial
268 (90.8%)
202 (94.0%)
66 (82.5%)
34 (91.9%)
83 (100%)
118 (100%)
47 (100%)
Infratentorial
27 (9.2%)
13 (6.0%)
14 (17.5%)
3 (8.1%)
0
0
0
91 (30.8%)
83 (38.6%)
8 (10.0%)
15 (40.5%)
83 (100%)
0
1 (2.1%)
54 (67.5%)
Location Basal ganglia Lobar
172 (58.3%)
118 (54.9%)
18 (48.6%)
0
118 (100%)
46 (97.9%)
Cerebellum
21 (7.1%)
9 (4.2%)
12 (15%)
2 (5.4%)
0
0
0
Brain stem
6 (2.0%)
4 (1.9%)
2 (2.5%)
1 (2.7%)
0
0
0
Pure IVH
5 (1.7%)
1 (0.5%)
4 (5.0%)
1 (2.7%)
0
0
0
142 (48.1%)
103 (47.9%)
39 (48.8%)
22 (59.5%)
59 (71.1%)
35 (29.7%)
12 (25.5%)
Side, L/R/B
135/150/10 46%/51%/3%
97/113/5 45%/53%/2%
38/37/5 48%/46%/6%
18/18/1 49%/49%/2%
41/42/0 49%/51%/0%
52/65/1 44%/55%/1%
17/29/1 36.2%/61.7%/2.1%
Craniotomy
109 (36.9%)
73 (34.0%)
36 (45.0%)
8 (21.6%)
14 (16.9%)
54 (45.7%)
28 (59.6%)
IVH
EVD
79 (26.8%)
55 (25.6%)
24 (30.0%)
9 (24.3%)
37 (44.6%)
14 (11.9%)
5 (10.6%)
Anticoagulants
39 (13.2%)
37 (17.2%)
2 (2.5%)
37 (100%)
15 (18.1%)
18 (15.3%)
5 (10.6%)
Antiplatelets
52 (17.6%)
46 (21.4%)
6 (7.5%)
8 (21.6%)
17 (20.5%)
27 (22.9%)
17 (36.2%)
CAA
47 (21.9%)
47 (21.9%)
0
5 (13.5%)
1 (1.2%)
46 (39.0%)
47 (100%)
Vascular pathology
80 (27.1%)
0
80 (100%)
0
0
0
0
CAA, cerebral amyloid angiopathy; F, female; M, male; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; L, left; R, right, B, bilateral; EVD, external ventricular drain.
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
7
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
2)
3)
P(>) = 1
150
Relative Humidity [%]
Frequency
100
50 50
−4
−2
0
−2 0 2 Relative Humidity [%]
150
100
50
2)
0
1
2
3
4
5
4 Day relative to bleeding
3)
P(>) = 99
Relative Humidity [%]
P(>) = 1
Frequency
CAA
Frequency
1)
100
100
50 −20
−5 −4 −3 −2 −1
0 −4
2
Relative Humidity [%]
150
0
−40
0
P(>) = 1
20
40
0
200
−0.5
2)
0 0.5 Gust [m/s]
1
P(>) = 4
20 150 0
Frequency
Frequency
VASC
150
100
1) 150
40
Frequency
P(>) = 97
200
anticoagulants
1)
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
−20
100
50 50 −40
0 −4
−2 0 Relative Humidity [%]
2
4
0
−5
0 Relative Humidity [%]
5
−5 −4 −3 −2 −1
0
1
2
3
Day relative to bleeding
4
5
0
−4
−2 0 Pressure [hPa]
2
4
Figure 5. Statistically significant signals found in the analyses 1)-3). CAA, cerebral amyloid angiopathy; VASC, patient group with vascular pathologies.
Taken together, apart from the aforementioned subgroups, no clearly identifiable and meaningful meteorological influence for the ICH events considered could be found in our study. This finding is surprising, because previous reports claim a clear influence of meteorological factors on ICH incidence. Low ambient temperature is a robust risk factor7,9,10,25,26 for ICH that certainly is associated with the increased incidence of ICH in winter,4-7 in line with the result of the seasonality analysis of our data that indicates a cluster within the “cold months” (i.e., February and March). Furthermore, ICH has been linked to high5,27 or low9,28 air pressure, high8 and low8,9 wind speed, high11 and low9,11 relative air humidity, and the occurrence of west/south-west wind and typhoon.9 These partially contradictory findings could be explained as the result of differences in local meteorological conditions as reported studies were conducted in different climate zones and continents. A “contamination” by other factors, however, like seasonal epidemic exposure to viruses, dietary changes,29 hypovitamnosis,30 or changes in daily habits31 also could lead to “pseudo-correlations” between meteorological phenomena and ICH occurrence.12 To overcome this inaccuracy of “seasonality,” we referred all data to the average of monthly climatology derived from a 10-year period as described previously. When this robust statistical approach was used, a potential relationship to ICH to meteorological conditions could be only detected for defined parameters and selected subgroups. We are aware of some limitations of our study. This study focuses on one center and a relatively small catchment area in global context (Switzerland and mostly Zurich region); however, this approach allowed us to collect high-quality local
8
www.SCIENCEDIRECT.com
meteorological data with the greatest possible spatiotemporal resolution. There are intercultural and interindividual differences in the body’s response to weather parameters, especially differences in comfort levels of humidity and temperature might exist. The data collection was performed retrospectively and the analysis is based on the day of symptom onset (not necessarily the admission day). Impacts of very short within-day and spatially confined meteorological fluctuations are theoretically still possible, but are beyond the scope of this study. Such an analysis would be much more complex to perform as it required detailed information about the exact time and location of ICH occurrences and more importantly, meteorological observational data on even shorter temporal and smaller spatial scales are not readily available. Furthermore, false-positive and false-negative results are possible and even likely, considering the number of parameters and statistical tests. However, the focus on results with similar signals in 3 independent statistical approaches adds robustness to the methods applied in this study. In addition, it is possible that other environmental factors, e.g., solar activity or lunar phase, influence ICH occurrence. Given the quality of data and the thorough methodology of this study, it seems highly unlikely that future studies on the parameters considered will show a clear impact of weather on ICH incidence. ACKNOWLEDGMENTS We thank MeteoSwiss for providing access to the observational data.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
REFERENCES 1. Labovitz DL, Halim A, Boden-Albala B, Hauser WA, Sacco RL. The incidence of deep and lobar intracerebral hemorrhage in whites, blacks, and Hispanics. Neurology. 2005;65:518-522. 2. Qureshi AI, Tuhrim S, Broderick JP, Batjer HH, Hondo H, Hanley DF. Spontaneous intracerebral hemorrhage. N Engl J Med. 2001;344:1450-1460. 3. Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993;24:796-800. 4. Jakovljevic D, Salomaa V, Sivenius J, Tamminen M, Sarti C, Salmi K, et al. Seasonal variation in the occurrence of stroke in a finnish adult population. The FINMONICA Stroke Register. Finnish monitoring trends and determinants in cardiovascular disease. Stroke. 1996; 27:1774-1779. 5. Jimenez-Conde J, Ois A, Gomis M, RodriguezCampello A, Cuadrado-Godia E, Subirana I, et al. Weather as a trigger of stroke. Daily meteorological factors and incidence of stroke subtypes. Cerebrovasc Dis. 2008;26:348-354. 6. Wang Y, Levi CR, Attia JR, D’Este CA, Spratt N, Fisher J. Seasonal variation in stroke in the Hunter Region, Australia: a 5-year hospital-based study, 1995e2000. Stroke. 2003;34:1144-1150. 7. Rothwell PM, Wroe SJ, Slattery J, Warlow CP. Is stroke incidence related to season or temperature? The Oxfordshire Community Stroke Project. Lancet. 1996;347:934-936. 8. Cevik Y, Dogan NO, Das M, Ahmedali A, Kul S, Bayram H. The association between weather conditions and stroke admissions in Turkey. Int J Biometeorol. 2015;59:899-905. 9. Kim HJ, Kim JH, Kim DR, Kang HI, Moon BG, Kim JS. Age and meteorological factors in the occurrence of spontaneous intracerebral hemorrhage in a metropolitan city. J Cerebrovasc Endovasc Neurosurg. 2014;16:209-215. 10. Magalhaes R, Silva MC, Correia M, Bailey T. Are stroke occurrence and outcome related to weather parameters? Results from a population-based study in northern Portugal. Cerebrovasc Dis. 2011; 32:542-551. 11. Slatina E, Music M, Babic N, Dervisevic A, Mujaric E, Salibasic M, et al. Correlation between
WORLD NEUROSURGERY -: ---, - 2016
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
change in air humidity and the incidence of stroke. Mater Sociomed. 2013;25:242-245. 12. Barnett AJ, Dobson AJ. Analysing Seasonal Health Data. New York, NY: Springer; 2010. 13. Zanobetti A, Luttmann-Gibson H, Horton ES, Cohen A, Coull BA, Hoffmann B, et al. Brachial artery responses to ambient pollution, temperature, and humidity in people with type 2 diabetes: a repeated-measures study. Environ Health Perspect. 2014;122:242-248. 14. Seifert J, McNair M, Declercq P, St Cyr J. A heat and moisture mask attenuates cardiovascular stress during cold air exposure. Ther Adv Cardiovasc Dis. 2013;7:123-129. 15. Rymer MM. Hemorrhagic stroke: intracerebral hemorrhage. Mo Med. 2011;108:50-54. 16. Russin JJ, Spetzler RF, Giannotta S, Meyer FB, Lawton MT, Cohen-Gadol AA. Introduction: strategies for management and outcomes of arteriovenous malformations. Neurosurg Focus. 2014;37. Introduction; http://dx.doi.org/10.3171/ 2014.7.FOCUS14421. 17. Hashiguchi N, Takeda A, Yasuyama Y, Chishaki A, Tochihara Y. Effects of 6-h exposure to low relative humidity and low air pressure on body fluid loss and blood viscosity. Indoor Air. 2013;23:430-436. 18. Morgan M, Winder M. Haemodynamics of arteriovenous malformations of the brain and consequences of resection: a review. J Clin Neurosci. 2001; 8:216-224. 19. Rossitti S, Svendsen P. Shear stress in cerebral arteries supplying arteriovenous malformations. Acta Neurochir (Wien). 1995;137:138-145 [discussion: 145]. 20. Maughan RJ, Shirreffs SM, Ozgunen KT, Kurdak SS, Ersoz G, Binnet MS, et al. Living, training and playing in the heat: challenges to the football player and strategies for coping with environmental extremes. Scand J Med Sci Sports. 2010;20(suppl 3):117-124. 21. Bain AR, Nybo L, Ainslie PN. Cerebral vascular control and metabolism in heat stress. Compr Physiol. 2015;5:1345-1380. 22. Han BH, Zhou ML, Johnson AW, Singh I, Liao F, Vellimana AK, et al. Contribution of reactive oxygen species to cerebral amyloid angiopathy, vasomotor dysfunction, and microhemorrhage in aged tg2576 mice. Proc Natl Acad Sci U S A. 2015;112: E881-E890.
23. Tanifum EA, Starosolski ZA, Fowler SW, Jankowsky JL, Annapragada AV. Cerebral vascular leak in a mouse model of amyloid neuropathology. J Cereb Blood Flow Metab. 2014;34: 1646-1654. 24. Yau MK, Rogers RR. A Short Course in Cloud Physics. Amsterdam, Netherlands: Elsevier; 1996. 25. Passero S, Reale F, Ciacci G, Zei E. Differing temporal patterns of onset in subgroups of patients with intracerebral hemorrhage. Stroke. 2000; 31:1538-1544. 26. Shinkawa A, Ueda K, Hasuo Y, Kiyohara Y, Fujishima M. Seasonal variation in stroke incidence in Hisayama, Japan. Stroke. 1990;21: 1262-1267. 27. Nakaguchi H, Teraoka A. Relationship between the occurrence of spontaneous intracerebral hemorrhage and holidays and traditionally unlucky days in Fukuyama City, Hiroshima Prefecture, Japan. J Stroke Cerebrovasc Dis. 2007;16:194-198. 28. Dawson J, Weir C, Wright F, Bryden C, Aslanyan S, Lees K, et al. Associations between meteorological variables and acute stroke hospital admissions in the west of Scotland. Acta Neurol Scand. 2008;117:85-89. 29. Fares A. Global patterns of seasonal variation in gastrointestinal diseases. J Postgrad Med. 2013;59: 203-207. 30. Karras S, Paschou SA, Kandaraki E, Anagnostis P, Annweiler C, Tarlatzis BC, et al. Hypovitaminosis D in pregnancy in the mediterranean region: a systematic review. Eur J Clin Nutr. 2016;70:979-986. 31. Menz V, Grimm W, Hoffmann J, Maisch B. Alcohol and rhythm disturbance: the holiday heart syndrome. Herz. 1996;21:227-231.
Conflict of interest statement: The authors declare that the article content was composed in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Received 18 September 2016; accepted 2 December 2016 Citation: World Neurosurg. (2016). http://dx.doi.org/10.1016/j.wneu.2016.12.006 Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
www.WORLDNEUROSURGERY.org
9
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
SUPPLEMENTARY DATA
P(>) = 48
200
10
Frequency
Frequency
mean = 0.16
5
150 100 50
0 −20
0
−10 0 10 Pressure [hPa]
−2
0 Pressure [hPa]
2
12 mean = 0.23
P(>) = 30
150 Frequency
Frequency
10 8 6 4
100
50
2 0
mean = −0.12
−1 0 1 Temperature [deg C] P(>) = 71
150 Frequency
20 Frequency
0
−5 0 5 10 Temperature [deg C]
15 10
100
50 5 0
−2
0
2 4 Gust [m/s]
6
0
8
−0.5
12 mean = −2.51
200 Frequency
Frequency
10 8 6 4
0.5
P(>) = 97
150 100 50
2 0
0 Gust [m/s]
−20 0 Relative Humidity [%]
20
0 −4
−2 0 2 Relative Humidity [%]
Supplementary Figure 1. Patients with vascular pathologies, analysis (i), (ii) and (iii).
9.e1
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
mean = 0.85
15
P(>) = 14
10
20 Pressure [hPa]
150 Frequency
Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
100
5
50
0
0
10 0 −10 −20
0 10 Pressure [hPa] mean = −0.07
10
5
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
100
50
−5
0
0 5 10 Temperature [deg C] mean = 0.09
10
10 5 0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−0.5 0 0.5 1 Temperature [deg C] P(>) = 39
200 Frequency
15
−1
5 Gust [m/s]
0
Frequency
2
P(>) = 59
150 Frequency
Frequency
15
−1 0 1 Pressure [hPa]
Temperature [deg C]
−10
−2
150 100
5
0 −5
50 −10 −5
5
mean = 3.42
15
150 Frequency
Frequency
0 Gust [m/s]
10
−0.5
0 0.5 Gust [m/s]
P(>) = 1
100
50
5
0
0
−20
0 20 Relative Humidity [%]
0 −4
−2 0 2 4 Relative Humidity [%]
1
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding Relative Humidity [%]
0
40 20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 1. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e2
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
P(>) = 28
mean = 0.39
Frequency
20 Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
15 10
100
50
5 0
−20
−10 0 Pressure [hPa]
mean = −0.13
20
10
P(>) = 66
100
50
0 −10
0 −1
−5 0 5 10 Temperature [deg C] mean = 0.11
−0.5 0 0.5 Temperature [deg C] P(>) = 20
150 Frequency
40 Frequency
−1 0 1 Pressure [hPa]
150 Frequency
30 Frequency
0
10
30 20
100
50 10 0
−4
−2
0 2 4 Gust [m/s]
mean = 0.33
30 20
−0.2
0 0.2 Gust [m/s]
0.4
P(>) = 29
200 Frequency
Frequency
40
150 100 50
10 0
0
6
−20 0 20 Relative Humidity [%]
0
−3
−2 −1 0 1 Relative Humidity [%]
Supplementary Figure 2. Patients without vascular pathologies, analysis (i), (ii) and (iii).
9.e3
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
30 mean = −0.31
P(>) = 68
20 15 10
20 Pressure [hPa]
150 Frequency
Frequency
25
100
50
5 0
0 −20 −40
−20
−10 0 10 Pressure [hPa]
0
20
−1
0 1 Pressure [hPa]
2
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 15
20
Temperature [deg C]
30
100
50
10 0 −10
P(>) = 57
150 Frequency
Frequency
40
mean = −0.05
0
−5 0 5 Temperature [deg C]
mean = 0.08
60
−0.5
5 0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
0 0.5 Temperature [deg C]
10
P(>) = 34
200
10
40
Gust [m/s]
Frequency
Frequency
5 150 100
0 −5
20 50
−10 0 Gust [m/s]
5
mean = −0.79
20
10
0
150
−0.5
0 Gust [m/s]
100
0
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 50
P(>) = 81
50
−20 0 20 Relative Humidity [%]
0.5
Relative Humidity [%]
Frequency
30
−5
0
Frequency
0 −10
−2 0 2 Relative Humidity [%]
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 2. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e4
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
mean = −0.99
2
1
100
50
0
−20
−10 0 Pressure [hPa]
0
10
mean = −0.11
−4
−2 0 2 Pressure [hPa]
4
2
4
P(>) = 57
150 Frequency
6 Frequency
P(>) = 69
150 Frequency
3 Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
100
50
0 −10
0
−5 0 5 Temperature [deg C]
−1 0 1 Temperature [deg C]
2
12 mean = 0.22
8 6 4
P(>) = 22
150 Frequency
Frequency
10
100
50
2 0
−2
0 2 Gust [m/s]
0 −1
4
0 0.5 Gust [m/s]
1
mean = 3.12
8
150
6
Frequency
Frequency
−0.5
4
P(>) = 1
100
50
2 0
0 −20
−10 0 10 20 Relative Humidity [%]
−4 −2 0 2 Relative Humidity [%]
4
Supplementary Figure 3. Patients with cerebral, amyloid angiopathy; analysis (i), (ii) and (iii).
9.e5
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
30
mean = 1.86 150
4
100
50
2
0 10 Pressure [hPa]
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Temperature [deg C]
Frequency
4
150 100 50
2 0
−5 0 Temperature [deg C]
0
5
mean = −0.39
15
−10
P(>) = 66
200
6
0
−2 0 2 Pressure [hPa]
20
mean = −0.2
8
10
−20
0
0
Frequency
20 Pressure [hPa]
Frequency
Frequency
6
P(>) = 9
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−1 0 1 Temperature [deg C] P(>) = 87
150
10
5
Gust [m/s]
Frequency
Frequency
5 100
0
50 −5
Frequency
8
0 Gust [m/s]
0
5
mean = −4.13 150
6 4
−1
−0.5 0 0.5 Gust [m/s]
1
100
2
−10 0 10 20 Relative Humidity [%]
0
40
P(>) = 99
50
0 −20
−5 −4 −3 −2 −1 0 1 2 3 4 5 Day relative to bleeding
Relative Humidity [%]
−5
Frequency
0
−5 0 5 Relative Humidity [%]
20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 3. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e6
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = 1.51
P(>) = 20 Frequency
Frequency
3
2
100
50
1
0 −10
−5
−2
0 2 Pressure [hPa]
mean = −0.55
8 6 4
4
P(>) = 82
150 Frequency
Frequency
0
0 5 10 Pressure [hPa]
100
50
2 0
0 −2
−5 0 5 10 Temperature [deg C]
−1 0 1 Temperature [deg C]
2
6 mean = 0.76
P(>) = 1
150
4
Frequency
Frequency
5
3 2
100
50
1 0 −2
0
2 Gust [m/s]
Frequency
150
6 4
0 0.5 Gust [m/s]
1
P(>) = 12
100
50
2 0
−0.5
4
mean = 1.7
8 Frequency
0
−30
−20 −10 0 10 Relative Humidity [%]
0
−4
−2 0 2 4 Relative Humidity [%]
Supplementary Figure 4. Patients with anticoagulants, analysis (i), (ii) and (iii).
9.e7
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
20
mean = 2.41
3
P(>) = 4
2
1
10 Pressure [hPa]
200 Frequency
Frequency
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
150 100 50
0 −10 −20 −30 −40
mean = −0.15
−4
6 4
−2 0 2 Pressure [hPa]
4
100
50
2
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 10
P(>) = 62
150 Frequency
8 Frequency
0 10 Pressure [hPa]
Temperature [deg C]
0 −10
0
5 0 −5 −10
0
−5 0 Temperature [deg C]
Gust [m/s]
Frequency
5
100
50
2
0 Gust [m/s]
0
5
3 2
−1 −0.5 0 0.5 Gust [m/s]
1
100
50
1 0 −20
0
−10 0 10 Relative Humidity [%]
−5 −4 −3 −2 −1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 78
150 Frequency
4
0
−5
Relative Humidity [%]
−5
6 mean = −1.31 5 Frequency
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 91
150
4
0
−1 0 1 Temperature [deg C]
mean = −0.52
6 Frequency
0
5
−5
0 5 Relative Humidity [%]
40 20 0 −20 −40 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 4. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e8
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = −0.44
P(>) = 70
150 Frequency
Frequency
8 6 4
100
50
2 0
−10
−5 0 5 Pressure [hPa]
0
10
10
5
2
P(>) = 66
200 Frequency
Frequency
mean = −0.12
−2 0 Pressure [hPa]
150 100 50
0
mean = 0.18
10
5
Frequency
15
−4
−2
0 2 4 Gust [m/s]
100
0
6
mean = −0.49
−0.5
150
10
5
0
P(>) = 20
50
Frequency
0
−1 −0.5 0 0.5 1 Temperature [deg C]
150 Frequency
15 Frequency
0
−5 0 5 10 Temperature [deg C]
0 Gust [m/s]
0.5
P(>) = 68
100
50
−20 0 20 Relative Humidity [%]
0 −4
−2 0 2 Relative Humidity [%]
Supplementary Figure 5. Patients with basal ganglia hemorrhage, analysis (i), (ii) and (iii).
9.e9
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
12
6 4
Pressure [hPa]
8
20
P(>) = 49
150 Frequency
Frequency
10
mean = 0.03
100
50
−20
0
−10 0 10 Pressure [hPa]
0 −10 −20
2 0
10
−2
−30 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
0 2 Pressure [hPa]
15 P(>) = 60
10
5
Temperature [deg C]
150 Frequency
Frequency
mean = −0.05
100
50
0
−0.5 0 0.5 1 Temperature [deg C]
5 0 −5 −10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 10
P(>) = 44
150
10
−5
0 Gust [m/s]
0
5
mean = 0.57
15 10
−0.5
0 0.5 Gust [m/s]
50
P(>) = 37
150
100
50
−20 0 20 Relative Humidity [%]
0
−10 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
1
5 0
0 −5
Relative Humidity [%]
20
100
50
Frequency
0
Gust [m/s]
Frequency
5
5
Frequency
−1
mean = 0.05
15 Frequency
0
−5 0 5 Temperature [deg C]
10
−4
−2 0 2 Relative Humidity [%]
4
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 5. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e10
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
12 mean = 0.68
P(>) = 23
150 Frequency
Frequency
10 8 6 4
100
50
2 0
−20
−10 0 Pressure [hPa]
mean = 0.1
−2
−1 0 1 Pressure [hPa]
15 10
2
P(>) = 41
150 Frequency
20 Frequency
0
10
100
50 5 0 −10
−5 0 5 Temperature [deg C] mean = 0.11
−0.5 0 0.5 1 Temperature [deg C] P(>) = 27
150 Frequency
20 Frequency
0 −1
10
15 10
100
50 5
Frequency
20
−4
−2
0 2 4 Gust [m/s]
mean = 0.65
150
15 10
−0.5
0 Gust [m/s]
0.5
P(>) = 25
100
50
5 0
0
6
Frequency
0
−20 0 20 Relative Humidity [%]
0
−2 0 2 Relative Humidity [%]
Supplementary Figure 6. Patients with lobar hemorrhage, analysis (i), (ii) and (iii).
9.e11
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
P(>) = 63
10
5
20 Pressure [hPa]
150 Frequency
Frequency
mean = −0.27
100
50
0 −20 −40
−10
20
mean = −0.23
15 10 5
0 2 Pressure [hPa]
10
150 100
0 −1
−5 0 5 Temperature [deg C]
mean = −0.04
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
P(>) = 75
50
0 −10
40
−2
200 Frequency
20 Frequency
0 10 Pressure [hPa]
0
Temperature [deg C]
0
0 −5 −10 −15 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
−0.5 0 0.5 1 Temperature [deg C]
10
P(>) = 62
200
5
20
Gust [m/s]
Frequency
Frequency
5 30
150 100
0 −5
10
50
0 −10
−5
0 Gust [m/s]
0
5
−10 −0.5
0 Gust [m/s]
0.5
−5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding 50
15 10
P(>) = 89 Relative Humidity [%]
150 Frequency
Frequency
20
mean = −1.39
100
50 5 0
−20 0 20 Relative Humidity [%]
0
−2 0 2 Relative Humidity [%]
0
−50 −5−4 −3 −2−1 0 1 2 3 4 5 Day relative to bleeding
Supplementary Figure 6. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e12
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
35 mean = 0.32
P(>) = 33
30
200
Frequency
Frequency
25 20 15
150
100
10 50
5 0
−20
−10 0 Pressure [hPa]
−1
0 Pressure [hPa]
1
2
160
45
mean = 0.01
P(>) = 47
40
140
35
120
30 Frequency
Frequency
0 −2
10
25 20
100 80 60
15 40
10
20
5 0
0
−10
−5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
180
70 mean = 0.19
P(>) = 6
160
60
140
50 Frequency
Frequency
120
40 30
100 80 60
20 40
10 0
20
−2
0
2 4 Gust [m/s]
6
0 −0.3
8
45
mean = −0.75
0 0.1 Gust [m/s]
0.2
0.3
140
35
120
30 Frequency
Frequency
−0.1
P(>) = 82
160
40
25 20
100 80
15
60
10
40
5
20
0
−0.2
180
−30
−20
−10 0 10 20 Relative Humidity [%]
30
0
−2
−1 0 1 Relative Humidity [%]
2
Supplementary Figure 7. All patients with high temporal resolution, analysis (i), (ii) and (iii).
9.e13
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
mean = 0.12
20
140
35
10
30 25 20
100 80
15
60
10
40
5
20
−20
−10
0 Pressure [hPa]
10
0
20
60
Pressure [hPa]
120 Frequency
Frequency
P(>) = 41
160
40
0
30
180
45
0 −10 −20 −30 −40
−1.5
−1
−0.5 0 0.5 Pressure [hPa]
1
1.5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
180
mean = −0.04
15
P(>) = 59
160
50 10
120 Frequency
Frequency
40
Temperature [deg C]
140
30
100 80 60
20
5 0 −5
40
−10
20
−15
10
0
0 −10
−5
0 5 Temperature [deg C]
−0.5
10
0 Temperature [deg C]
0.5
180
70
10
mean = −0.02
P(>) = 56
160
60
8
140
6
120
4
40 30
Gust [m/s]
Frequency
Frequency
50
100 80
0
−4
40
−6 −8
20
−5
0 Gust [m/s]
5
0 −0.4
10
60
−10
−0.2
0 0.2 Gust [m/s]
0.4
180
mean = 0.55
60 P(>) = 31
160
50
40
120 Frequency
Frequency
Relative Humidity [%]
140
40
30
100 80 60
20
0 −2
60
20 10
2
40
20
0
−20
−40
10 20
−60
0
−40
−20 0 20 Relative Humidity [%]
40
60
0
−2
−1 0 1 Relative Humidity [%]
2
3
Supplementary Figure 7. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e14
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.48
14
140
12
120
10 8
80 60
4
40
2
20 0
−15
−10
−5 0 5 Pressure [hPa]
10
15
18
−3
−2
−1 0 Pressure [hPa]
1
2
180
mean = 0.46
16 14
140
12
120
10 8
100 80
6
60
4
40
2
20
0
−5
0 5 Temperature [deg C]
P(>) = 16
160
Frequency
Frequency
100
6
0
P(>) = 33
160
Frequency
Frequency
16
0 −1.5
10
−1
−0.5 0 0.5 Temperature [deg C]
1
180
mean = −0.09
P(>) = 65
160
20
120
15 Frequency
Frequency
140
10
100 80 60 40
5
20
0
−2
0
2 Gust [m/s]
4
0 −0.6
6
12
−0.4
−0.2
0 0.2 Gust [m/s]
0.4
0.6
180
mean = −2.55
160
P(>) = 96
10 140 120 Frequency
Frequency
8
6
100 80 60
4
40
2 20
0 −30
−20
−10 0 Relative Humidity [%]
10
0 −5
0 Relative Humidity [%]
5
Supplementary Figure 8. Patients with vascular pathologies studied with high temporal resolution, analysis (i), (ii) and (iii).
9.e15
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.54
20
P(>) = 28
160
15
140
10
12
120
5
10 8
Pressure [hPa]
14
Frequency
Frequency
16
100 80
0 −5
6
60
−10
4
40
−15
2
20
−20
0
0
−15
−10
−5 0 5 Pressure [hPa]
10
15
12
−25
−3
−2
−1 0 Pressure [hPa]
1
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
4
5
2
180
mean = −0.43
P(>) = 83
160
10
10
120 Frequency
Frequency
8
Temperature [deg C]
140
6
100 80 60
4
5
0
−5
−10
40
2 20 0
0
−5
0 5 Temperature [deg C]
18
−15
−1
−0.5 0 0.5 Temperature [deg C]
1
1.5
180
mean = 0.03
16
8
P(>) = 46
160
6
14
140
12
120
10 8
100 80
6
60
4
40
2
20
0
−6
−4
−2 0 Gust [m/s]
2
Gust [m/s]
Frequency
Frequency
4
0 −2 −4 −6 −8
0 −1
4
2
−0.5
0 Gust [m/s]
0.5
18 180 P(>) = 1
160
14
40 Relative Humidity [%]
140
12
120 Frequency
Frequency
60
mean = 4.22
16
10 8
100 80
6
60
4
40
20
0
−20
20
2 0 −20
0
20 40 Relative Humidity [%]
60
0 −6
−40 −4
−2 0 2 Relative Humidity [%]
4
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
Supplementary Figure 8. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e16
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
18
180
mean = 0.22
140
12
120
10 8
100 80
6
60
4
40
2
20
0
P(>) = 40
160
14
Frequency
Frequency
16
0
−20
−10 0 Pressure [hPa]
−2
10
30
−1
0 Pressure [hPa]
1
2
180
mean = −0.16
P(>) = 70
160
25 140 120 Frequency
Frequency
20
15
100 80 60
10
40
5 20
0
0
−10
−5 0 5 Temperature [deg C]
−0.5 0 Temperature [deg C]
mean = 0.29 160
35
140
30
120 Frequency
40
25 20
0.5
80 60
10
40
5
20
−2
0
2 4 Gust [m/s]
6
P(>) = 2
100
15
0
−1
180
45
Frequency
10
0 −0.4
8
−0.2
0 Gust [m/s]
0.2
0.4
35 mean = −0.03
P(>) = 49
30
200
Frequency
Frequency
25 20 15
150
100
10 50
5 0
0
−30
−20
−10 0 10 20 Relative Humidity [%]
30
−3
−2
−1 0 1 Relative Humidity [%]
2
3
Supplementary Figure 9. Patients without vascular pathologies studied with high temporal resolution, analysis (i), (ii) and (iii).
9.e17
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
180
mean = −0.14
30
P(>) = 58
160
20
20 10
15
10
Pressure [hPa]
120 Frequency
Frequency
140
100 80
0 −10 −20
60 −30
40
5
−40
20
0
−20
−10
45
0 Pressure [hPa]
10
0
20
−2
−1
0 Pressure [hPa]
mean = 0.11
1
2
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
4
5
200 10
Frequency
30 25 20
Temperature [deg C]
35 Frequency
4
15
P(>) = 34
40
−5 −4 −3 −2 −1 0 1 2 3 Day relative to bleeding
150
100
15 50
10
5 0 −5 −10
5 −15
0
−5
0 5 Temperature [deg C]
10
0 0.5 Temperature [deg C]
1
180
45
10
mean = −0.03
P(>) = 60
160
40 35 30 Frequency
Frequency
−0.5
25 20
8
140
6
120
4 Gust [m/s]
0 −10
100 80
2 0 −2
15
60
−4
10
40
−6
5
20
0
−5
0 Gust [m/s]
5
0
10
−8 −10 −0.4
−0.2
0 0.2 Gust [m/s]
0.4
0.6
60
45
mean = −0.9
P(>) = 81
40
200
40
25 20
Relative Humidity [%]
30
Frequency
Frequency
35 150
100
15 50
10
20
0
−20
−40
5 0
−40
−20 0 Relative Humidity [%]
20
40
0 −4
−60
−2
0 Relative Humidity [%]
2
Supplementary Figure 9. (continued)
WORLD NEUROSURGERY -: ---, - 2016
www.WORLDNEUROSURGERY.org
9.e18
ORIGINAL ARTICLE MARIAN CHRISTOPH NEIDERT ET AL.
METEOROLOGICAL INFLUENCES ON THE INCIDENCE OF ICH
600
500
# Cases
400
300
200
100
0 20
40
60 80 Relative Humidity [%]
100
Supplementary Figure 10. Distribution of relative humidity at bleeding day.
Median Number of Cases/Month
6
5
4
3
2
1
0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Supplementary Figure 11. Seasonal distribution between 2006 and 2013 (adjusted time period to ensure whole-year datasets).
9.e19
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2016.12.006