In this 11-year study of 260 patients with SAH, 533 patients with ICH, and 3,245 patients with IS, we found that some patterns of RH and CC and daily changes in AP and RH were associated with the risk of some types of stroke. In addition, we used NAO, AO, EA/WR, SCA, and ENSO indices as predictors for the evaluation the risk of stroke. For the first time, we detected a protective effect of warmer ENSO (a stronger El Niño) on the risk of SAH, a positive association between the risk of HS and the EA/WR, a negative association between the risk of IS and the AOI, and a negative impact of a strong positive SCA on the risk of IS. Apart from this, during November-March, a higher risk of HS was related to a positive NAO, and a negative association between the risk of IS and NAOI was found. In the analysis, an impact of teleconnection indices was detected, adjusting for seasonal variation, T, and other weather variables.
In our study, the risk of HS was associated with daily change in AP above the threshold of 3.9 hPa for SAH and 9.55 hPa for ICH. These results are in line with those obtained by other authors who found a significant association between SAH and ICH and changing AP. The daily change in AP with a lag of 1 day positively correlated with daily number of SAH in the English Midlands , and the daily change in AP > 10 hPa was association with the risk of SAH in the UK  and in Germany .
To our knowledge, the risk of SAH was associated with a higher RH and CC level, and the risk of ICH – with a higher CC level on the previous day and with a lower daily RH change. Some authors found a significant association between stroke and RH and daily hours of sunshine, which is the opposite variable to cloud cover. A decrease in daily sunlight hours was positively associated with the risk of SAH [17, 39]. A positive association between SAH and RH was found in area of humid subtropical zone  and in the Rhein Main area , while in the southern regions of France and in the area with various climatic condition (41 states of USA), a negative association between SAH and RH was observed [17, 39]. ICH negatively correlated with sunshine hours  and positively correlated with amount of precipitation , which coincident with our results.
We found a positive effect of a strong warm ENSO on human health. In our study, most of these events (NINO3.4 > 1.14) fell into colder months. Therefore, the protective effect of a warmer ENSO on SAH may be explained by the effect of ENSO on the weather pattern during autumn and winter. Studies have shown that the warm and the cold phase of ENSO have different impacts on the pattern of weather regimes during the colder season in Europe . Strong El Niño events were related to a higher sea level pressure, a lower T, and dry air in the Baltic countries [41, 33, 42]. El Niño winters are associated with a significant increase in the prevalence of a positive NAO in November–December and a significant increase in the prevalence of a negative NAO and the Greenland anticyclone in January–March . According to our data, during the second half of autumn, the period of NINO3.4 > 1.14 was characterised by a lower mean AP, a higher amount of precipitation and RH, and a very significantly lower diurnal temperature range (DTR) (by 1.3 oC, p < 0.001). In winter, the period of NINO3.4 > 1.14 was characterised by a lower T, WS, and CC and a higher AP and DTR. These weather patterns may be associated with a lower risk of SAH. Some studies have shown that a higher DTR is significantly associated with higher mortality, and this effect was stronger during autumn [43, 44]. Therefore, a lower DTR during autumn may have a positive effect on human health.
We found a positive association between the EA/WRI and HS and an increase in the risk of IS during a strong negative EA/WR phase. These effects were similar during both the colder and the warmer periods, and the additional inclusion of the NAOI in the model did not reduce the significance of the EA/WRI. In the studied region, in winter, the positive EA/WR produced cold advection from the north and was characterised by a lower air temperature, a lower precipitation level, and stronger atmospheric circulation . In the south-eastern region of Baltic Sea, a positive EA/WR indicated the north-westerly air flow , and the EA/WRI negatively correlated with air temperature in spring , with lake water temperature in spring-autumn , and with precipitation amount in summer . According to our data, during the positive EA/WR phase, a higher mean WS and AP in winter and a lower mean T and a higher mean AP both in spring and summer were observed. Apart from this, a higher variation was found in the daily change of AP both in winter and spring. Thus, it can be said that a higher EA/WRI was related to a stronger variation in AP, colder air flow in winter, and colder air in other seasons. The complex of these weather patterns may be associated with a higher risk of HS. Very low EA/WRI (<-1.81, 3rd percentile) was observed during May, June, and December 2010). The period of May-June is characterised by cyclonic conditions (a lower mean AP and a higher mean RH and precipitation amount) and December – by a lower (by 6.8 oC) mean T and a higher variation in ΔT. These conditions may increase the risk of IS.
For the first time, negative associations between IS and AO and between IS and NAO only during November-March were found. According to the results of studies by other authors, AO was associated not only with tropospheric, but also with stratospheric variability and changes in weather pattern in Lithuania and nearby regions [49, 50]. AO is a main contributor to anomalous winter precipitation in the Northern Europe and Asia . During January-March, positive AO brings a higher surface T and a lower precipitation in middle-latitude regions. Moreover, the boundary of the change in sign of correlation between AO and precipitation lies at ∼ 55N in Eurasia . In the region of the Baltic sea, a positive correlation between T and AO was observed during January-March , March-May , July and October , and September-March . According to our data, the AOI positively correlated with T in all seasons excluding summer and negatively correlated with RH excluding winter. It is possible that this complex of weather patterns (a higher T and a lower RH during the equinox, a lower RH in summer, and warmer winters) related to days of a higher AOI had a protective effect against the risk of IS.
The SCA pattern affects the climate in Eurasia, and a seasonality effect was observed . In the positive SCA phase, warm anomalies rise over the Norwegian Sea and Greenland, especially during autumn . Apart from this, during December-April, due to a weaker stratospheric polar vortex , a lower mean daily surface T and a higher frequency of the occurrence of extreme cold events are probable in northern Europe [55, 56]. The SCAI positively correlated with T in summer and negatively correlated with T and positively correlated with AP over the region of the Baltic Sea in winter [26, 31, 47]; the same associations were found in our study. The positive phase of SCA indicates more likely anti-cyclonic conditions and a lower level of atmospheric circulation over the Baltic Sea region during autumn-spring , and the anticyclonic conditions over Scandinavia substantially suppress westerly zonal airflow in summer . It is possible that cold outbreaks during the colder period and the atmospheric variations related to a stronger positive SCA are associated with the risk of IS.
During the colder period, the positive NAO had a protective effect against IS, but, vice versa, a negative NAO had a protective effect against HS. During wintertime, in the Baltic Sea region, a positive NAOI was associated with a higher T and with altered weather: a higher WS, a lower AP, and a north-eastward shift in the Atlantic storm activity with enhanced activity from Newfoundland into Northern Europe  As a positive NAO during the winter was associated with more changing weather, a positive NAO was risky for HS, whereas a change in T was more relevant for IS. Studies in Northern and Middle Europe have shown a higher risk of HS associated with changing weather but not with T [19, 20, 22, 38]. A study conducted in the UK showed a significant impact of changes in T only on IS, whereas changes in AP had a significant impact only on the risk of HS .
According to literature, both positive and negative NAO phases are associated with worse health outcomes  found a positive association between the daily AOI with a lag of 3 days and the incidence of and mortality from acute myocardial infarction in Northern Sweden. An inverse association between the climate index (which represents winters with a strong negative phase of the NAO) and the level of mortality from ischemic heart disease were found in England . In addition, a negative association between the NAO index and systolic and diastolic blood pressure during spring-autumn was found .
The pathophysiological mechanisms underlying the correlation between stroke and weather conditions have been discussed. Factors that increase the risk of stroke include high blood pressure, some diseases, and the lack of regular exercise. Blood pressure is influenced by cold, stress and physical activity . We found that changes in AP affected the risk of HS. Other researchers presented the explanations of this effect. Donkelaar et al.  hypothesized that AP changes trigger the inflammation process in the aneurysm wall. Variations in AP may influence on vessel walls and their endothelial function by endogenous inflammatory mechanisms . Studies about thrombosis in air travels suggesting that prothrombin fragments and the thrombin-antithrombin complex are activated in hypobaric conditions [62, 63] could be another clue to the underlying mechanism.
The negative impact of a higher CC and RH may be explained by a lower number of sunlight hours and thus low vitamin D levels and fewer physical activity opportunities - especially for the elderly. A lower CC is related to a higher number of sunlight hours and thus higher vitamin D levels during the colder period. Sunlight exposure has been shown to alter blood pressure through the effects of UV light on vitamin D . The lack of vitamin D may induce an inflammatory process in the aneurysm wall that increases the risk of rupture . Studies on the associations between physical activity in the elderly and weather conditions in Europe showed that physical activity decreased with increasing WS, precipitation, humidity, and a shorter duration of sunshine [65, 66]. These weather conditions are associated with a negative AO and NAO excluding winter months – therefore, it can be assumed that negative AOs are associated with fewer physical activity opportunities for the elderly, who are likely to be stressed. This can explain the fact a negative AO increased the risk of IS.
The present study has several strengths: the large number of included patients with various types and subtypes of stroke, the long study period, and standardised methods and criteria used for stroke register. In addition, it analyses daily stroke incidence data by stroke subtypes, daily meteorological data, the variation of these data with respect to the previous day, and uses the teleconnections such as NAO, AO, EA/WR, SCA, and ENSO. Moreover, the fact that the included patients come from a small geographical area (Kaunas city) contributes to the homogeneity of weather conditions. In our study, associations between atmospheric circulation patterns such as NAO, AO EA/WR, SCA, and ENSO and the risk of stroke were found for the first time.
The limitation is that other potential confounders such as air pollution, influenza epidemics or other respiratory infections were not directly considered in this study. In our region infections are strongly related to the season with a highest prevalence during winter . Despite the fact that in our study, the analyses were controlled for the month and T, residual confounding by short-term respiratory epidemics remains a possibility. Moreover, we did not consider weather-related physical activity that may have had an impact on individual exposure to outdoor T and humidity.
In our study, the influence of air pollution, which is a known trigger for cardiovascular diseases, has not been examined. We did not have air pollution data for the entire study period, but the additional inclusion of the daily concentrations of PM10, NO2, or O3 did not change the association between the risk of strokes and teleconnection indices. Based on the results published by other authors , we can assume that the short-term effect of PM10, NO2, or O3 on the risk of stroke was not significant enough to affect the results of our study. First, the level of air pollution in Kaunas is not high. Second, our stroke patients were relatively young (< 65 years of age), whereas in other studies presenting a positive association between air pollution and stroke (except for those performed in subtropics or in regions with high levels of pollutants), the mean age of the patients was over 70 years . Third, we did not evaluate other comorbidities such as acute myocardial infarction, ischemic heart disease, arterial hypertension, heart failure or other risk factors such as atrial fibrillation, diabetes, dyslipidemia, or renal or malignant diseases, which may also be associated with a higher risk of ischemic and hemorrhagic stroke. Fourth, harmful lifestyle factors such as alcohol consumption or smoking, which increase the risk of haemorrhagic stroke, cannot be ruled out either.