key: cord-322906-zef971xp
authors: Hochman, Assaf; Alpert, Pinhas; Negev, Maya; Abdeen, Ziad; Abdeen, Abdul Mohsen; Pinto, Joaquim G.; Levine, Hagai
title: The relationship between cyclonic weather regimes and seasonal influenza over the Eastern Mediterranean
date: 2020-08-12
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2020.141686
sha: 
doc_id: 322906
cord_uid: zef971xp

Abstract The prediction of the occurrence of infectious diseases is of crucial importance for public health, as clearly seen in the ongoing COVID-19 pandemic. Here, we analyze the relationship between the occurrence of a winter low-pressure weather regime - Cyprus Lows - and the seasonal Influenza in the Eastern Mediterranean. We find that the weekly occurrence of Cyprus Lows is significantly correlated with clinical seasonal Influenza in Israel in recent years (R = 0.91; p < .05). This result remains robust when considering a complementary analysis based on Google Trends data for Israel, the Palestinian Authority and Jordan. The weekly occurrence of Cyprus Lows precedes the onset and maximum of Influenza occurrence by about one to two weeks (R = 0.88; p < .05 for the maximum occurrence), and closely follows their timing in eight out of ten years (2008–2017). Since weather regimes such as Cyprus Lows are more robustly predicted in weather and climate models than individual climate variables, we conclude that the weather regime approach can be used to develop tools for estimating the compatibility of the transmission environment for Influenza occurrence in a warming world. Furthermore, this approach may be applied to other regions and climate sensitive diseases. This study is a new cross-border inter-disciplinary regional collaboration for appropriate adaptation to climate change in the Eastern Mediterranean.

The World Health Organization (WHO) has estimated that in 2012 approximately 12.6 million deaths (23% of all deaths worldwide) were attributed to changeable environmental factors, of which many could be potentially influenced by ongoing climate change (WHO 2016) . In addition, the Lancet Commission on Health and Climate Change determined that "Climate change could be the greatest public health threat of the 21 st century" (Watts et al., 2015; . There is clear evidence that climate change in the last 50 years has affected human health, partly by altering the epidemiology of climate sensitive diseases (e.g., Patz et al., 2005; Mirsaeidi et al., 2016) . Specifically, climate change leads to alterations in the mean, variability, seasonality and/or extremes in one or more climatic variables such as temperature, precipitation, humidity, aerosols etc. These changes influence the dispersal of pathogens, the transmission environment and the host's resilience (Vittecoq et al., 2017) . Health effects related to climate change tend to emerge as seasonal and geographical alterations in the spread of disease (Wu et al., 2016;  Dennis and Fischer 30% of the burden of infectious diseases in Europe in the period 2009-2013 (Cassini et al., 2018) . Caini et al. (2018) analysed the timing of seasonal Influenza maximum occurrence, showing that it was delayed by 2.8d/year in Western Europe from 1996 to 2016, while it was shortened by 3.5d/year in Eastern Europe. Regarding Israel, Caini et al., (2018) identified a progressive delay of maximum incidence by 2.8d/year. The reasons for such changes have not yet been fully explained, but ongoing climate change is among the leading candidates. Moreover, a recent study revealed intense inter-seasonal Influenza activity during 2018/9 ( Barr et al., 2019) , reinforcing the need for year-round surveillance of Influenza, even in areas with strong seasonality patterns like the Eastern Mediterranean.

A plethora of studies have indicated that the timing of seasonal Influenza varies across latitude, thus suggesting that meteorological conditions play an important role in the transmission of the disease (Soebiyanto et al., 2010; 2014; Tang et al., 2010; Baumgartner et al., 2012; Shaman and Karspeck 2012; Yang et al., 2012; Tamerius et al., 2013; Yaari et al., 2013; Chong et al., 2019) . A few studies have demonstrated that absolute humidity modulates the airborne survival and transmission of the Influenza virus (Shaman and Kohn 2009; Shaman et al., 2010; 2011) . Soebiyanto et al. (2015) have investigated the association between climatic variables and seasonal Influenza in temperate and sub-tropical regions, including Israel. They have provided evidence that an increase in Influenza activity is related to a decrease in temperature and specific humidity. Recently, Zhao et al. (2018) has shown that Influenza specific combination of variables that leads to enhanced incidence of infectious diseases, e.g. low temperatures and rainy/moisty or dry conditions in the case of Influenza (Axelsen et al., 2014; Guo et al., 2015; Chong et al., 2020) . Indeed, there is growing evidence that large scale climatic modulations such as the El-Niño or La-Niña may influence the onset and peak of seasonal Influenza in many regions across the globe (Oluwole 2015 (Oluwole , 2017 Chun et al., 2019) .

In order to potentially predict the compatibility of the transmission environment for climate sensitive infectious diseases, it is necessary to obtain skilful predictions of several meteorological variables at the same time, as clearly shown in the current COVID-19 pandemic (Ma et al., 2020) . However, one caveat of using individual climatic variables is that seasonal, decadal and multi-decadal weather and climate forecast models struggle in predicting these variables individually, especially in regions distant from the onset of the El-Niño (Weisheimer and Palmer 2014 ).

However, model forecasts are generally more robust in predicting weather regimes occurrences and timing (Weisheimer and Palmer 2014; Grams et al., 2017) . A weather regime can be considered as a capsule containing much of the information on the transmission environment, including the synergistic relations between individual climatic variables like temperature, precipitation, humidity and wind (Lamb 1950; 1972; Stein and Alpert 1993; Alpert and Sholokhman, 2011; Santos et al., 2016) .

Moreover, a weather regime approach also has the advantage of retaining the physical relationship between the individual climatic variables.

south-eastern part of the Eastern Mediterranean region (Alpert et al., 2004a) . The semi-objective synoptic classification describes well the local weather and has many implications (Alpert et al., 2004a, b; Saaroni et al., 2010a, b; Hochman et al., 2018a, b; 2019a, b; 2020a, b) . For example, Hochman et al. (2018b) have provided evidence that by the end of the 21 st century, the duration of the summer is projected to extend by 49% (+ ~ 60 days), while the winter is expected to be shortened by 56% (-~ 60 days) under the "business as usual" greenhouse gas scenario (RCP8.5). The authors (Hochman et al., 2018b) concluded that these alterations may lead to substantial changes in the timing of seasonal health hazards including seasonal Influenza.

High-quality climatic and infectious disease information is sparse in some regions of the world, e.g., in several of the Eastern Mediterranean countries. In this respect, Google Trends is a freely accessible tool that may provide insights into population For this study, we set a cross-border inter-disciplinary regional collaboration, composed of climatologists, epidemiologists and public health professionals from the Palestinian Authority, Israel and Germany (Hochman et al., 2020c) . The purpose of J o u r n a l P r e -p r o o f Journal Pre-proof this study is to investigate the potential link between weather regime occurrences and climate sensitive infectious diseases, and discuss in how far this relationship can help to inform decisions in the health sector. As a case study, the weekly occurrences of an Eastern Mediterranean weather regime -Cyprus Lows -together with precipitation, temperature and humidity, are related to seasonal Influenza in Israel, the Palestinian Authority and Jordan.

Climatic data were acquired from the National Centre for Environmental Prediction/National Centre for Atmospheric Research (NCEP/NCAR) reanalysis archive (Kalnay et al., 1996) . We The student t-test is used to check for statistical significance of the correlation coefficients at the 5% significance level. Finally, we investigate the ability of the weather regime approach in depicting the onset and maximum occurrence of the Influenza season. The onset (maximum) is defined as the first week of the season with above zero (one) normalized incidence rates or weekly occurrence of Cyprus Lows. It should be noted that all datasets are Open Access for reproducibility and transparency and the modeling framework was validated as recommended by Walters et al. (2018) .

Data from the ICDC weekly reports of positive specimens for Influenza at the Trends data (Fig. 2 -4 ; Table S1 ). The reason may be that the individual variables have a strong impact on human behavior, which the Google Trends tool was designed to detect (Nuti et al., 2014) . In terms of seasonality, Google Trends estimates the onset of Influenza at week 35 and the maximum occurrence at week 2 (Fig. 5) . In this case, the Influenza onset and maximum precede all the climatic variables, with closer relation to weekly precipitation amounts ( Fig. 3 and 5) . The observed seasonal occurrence of Influenza in all three regions using Google Trends is very similar (Fig.   5 ). The large similarity between the three countries suggests that indeed the regional environmental factors may play an important role. When analyzing the individual years in terms of ILI and Flu+ relationship to Cyprus Low occurrence, some year-toyear variability is identified (Fig. S1 ). For example, differences in the seasonality of Influenza between years with a relatively low occurrence of Cyprus Lows (e.g., 2013) compared to years with a higher occurrence (e.g., 2009). Still, the tight relationship between Cyprus Lows and Influenza is retained.

Next, we quantify the relationship between the meteorological variables and seasonal Influenza occurrence in Israel. As a first choice, a stepwise multiple linear regression model is adopted for the relation between the predictors, i.e., Cyprus Lows J o u r n a l P r e -p r o o f and four individual meteorological variables (temperature, precipitation, specific humidity and relative humidity) and the predictand (Flu+) for 2008-2017. It is found that the number of Cyprus Lows per week can explain 82% of the variance with a root mean square difference of 0.14. Adding the other meteorological variables only marginally contributes to a higher explained variance (Table 1) occurrence is identified using the student t-test at the 5% significance level (Fig. 6 ).

Furthermore, a close inspection of Figure 6 suggests that the seasonal variability of Cyprus Lows closely follows the seasonal variability of Influenza occurrence, especially close to maximum occurrence of Influenza. In addition, we tested other potential regression models for predicting Flu+ from the weekly occurrence of Cyprus Lows. Only marginal improvement is shown for increasing order of polynomial or Sine non-linear models (Table S2 ). 

In this study, we analyze the relationship between weather regimes and seasonal Influenza over the Eastern Mediterranean for the period 2004-2017. We find that Cyprus Low weekly occurrence has the highest significant correlation (R = 0.91; p < 0.05) with weekly clinical Influenza data over Israel with respect to individual climate variables such as precipitation, temperature, specific humidity and relative humidity.

Influenza when performing a complementary analysis for the Palestinian Authority, Jordan and Israel using Google Trends data. The weekly occurrence of Cyprus Lows precedes the onset and peak of Influenza occurrence with a lag of about 1 to 2 weeks, and a correlation for maximum occurrence of R = 0.88 (p < 0.05). The evolution of both curves matches in eight out of ten years. This is an important finding, since the Influenza virus has ~3 days incubation period, and ~3 additional days until a patient visits a clinic and a few more days for positive virus identification.

The role of weather in the spread of Influenza is not yet fully understood. Thus, it is not possible to explicitly determine the biological mechanism relating Cyprus Lows and Influenza occurrence. However, the environmental setting given by the typical cold weather associated with the Cyprus Low may affect directly or indirectly Influenza through the host and/or the pathogen, i.e., the epidemiological triangle (e.g., Fuhrmann 2010). For example, the host susceptibility to infection during Cyprus Low days may increase due to seasonal hormonal changes, which may be related to a reduction in exposure to sun light. For example, some studies suggest that low levels of Vitamin D may weaken the immune response (Cannel et al., 2006) , while high levels may reduce the risk of both Influenza and COVID-19 infections and death (Grant et al., 2020; Merzon et al., 2020) . Furthermore, rapid changes in weather,

Journal Pre-proof which may influence both the host susceptibility and the pathogen survival, can also increase the risk of an Influenza epidemic (Liu et al., 2020) . Indeed, the winter season in the Eastern Mediterranean is typically associated with rapid changes in weather governed by transitions from Cyprus Lows to high-pressure systems (Alpert et al., 2004b) . The strongest variability in weather regime transitions is especially evident in early winter (Hochman et al., 2020b) , which also corresponds to the onset of the Influenza season. In addition, the ability of the Influenza virus to cause infection is higher when the air is sufficiently cold (Polozov et al., 2008) . Finally, studies have provided evidence that fluctuations in the duration of social contacts may relate to weather conditions (e.g., Smieszek 2009; Willem et al., 2012) . Indeed, indoor social contacts during Cyprus Low days may be more frequent and longer lasting.

The present results suggest that the weather regime approach can be used to develop a tool for estimating the compatibility of the Influenza transmission environment in a changing climate. This is particularly important, since weather and climate model forecasts are generally more robust in predicting weather regimes occurrences and timing than individual climatic variables, especially at regions distant from the El-Niño (e.g., Weisheimer and Palmer 2014; Grams et al., 2017) . This can be achieved by applying the methodology to sub-seasonal, seasonal, annual, decadal and even multi-decadal scales using climate model predictions / projections (Marotzke et al., 2016; Vitart and Robertson, 2018) . We envisage that this novel approach could be applied to other regions (Soebiyanto et al., 2015; Chun et al., 2019) and infectious illnesses, such as vector-borne diseases and infectious gastroenteritis or even the SARS-CoV-2 and its associated disease COVID-19.

Unfortunately, climatology tools and data are under-utilized in public health J o u r n a l P r e -p r o o f Journal Pre-proof (Fuhrman 2010; Hochman et al., 2020c) . This study exemplifies the potential for inter-disciplinary collaboration.

The present results and methodology can potentially be helpful for public health in practical terms, i.e., better understanding of the correlation between weather regimes and Influenza may improve vaccination policy and medical resources allocation. For example, health systems may roughly estimate the timing of seasonal Influenza surge and improve the timing of seasonal vaccination campaigns for the general population, as well as for vulnerable populations including the elderly, poor, women, children, disabled, refugees and chronically ill patients. In addition, we suggest that the utilization of Google Trends and other social media information in real-time may be beneficial in sparsely monitored regions, such as several Eastern Mediterranean countries. Improved understanding of the relationship between Internet searches and actual illness patterns may help countries with limited monitoring of diseases to plan timely health promotion, including seasonal vaccination and campaigns for preventive measures, as well as plan allocation of medical resources. However, this type of information should not replace investments in traditional data gathering and analysis, but rather serve as a complement to it (Lazer et al., 2014) . As a caveat, we note that seasonal Influenza occurrence, as most other infectious diseases, may be influenced by: vaccination effectiveness, public awareness, biological and socio-economic factors etc., which are not easily quantified and measured (Caini et al., 2018) . We further note that laboratory confirmed Influenza cases may also suffer from inaccuracies. In fact, a recent review article that tested the quality of different laboratory tests for viral respiratory infections, including Influenza found that the pooled sensitivity of the different tests was 90.9% (95% confidence interval of 88.7% -93.1%) and specificity of 96.1% (95% confidence interval of 94.2% -97.9%; Vos J o u r n a l P r e -p r o o f Journal Pre-proof et al., 2019) . In addition, laboratory tests are indeed just the tip of the iceberg with respect to the actual incidence rates of Influenza or any other infectious disease.

Climatic changes and associated health risks know no borders. The challenges climate change pose to society and especially to public health, can be properly met only with regional collaborations as clearly revealed in the COVID-19 pandemic (Otu et al., 2020) . This manuscript was prepared as part of a Jordanian, Palestinian, Israeli and German collaboration towards the establishment of a Regional Climate Change Adaptation Center (RCCAC). The collaborators are committed to this important regional trans-national cooperation, which will benefit the people of this vulnerable part of the world. 

Epidemic influenza and vitamin D

Impact of infectious diseases on population health using incidence-based disability-adjusted life years (DALYs): results from the Burden of Communicable Diseases in Europe study, European Union and European Economic Area countries

Latitudes mediate the association between influenza activity and meteorological factors: A nationwide modelling analysis in 45 Japanese prefectures from

Association between meteorological variations and activities of influenza A and B across different climate zones: a multi-region modelling analysis across the globe

Weather Regimes and analogues downscaling of Seasonal Precipitation; A case study over Israel

The dynamics of cyclones in the Twentyfirst Century; The Eastern Mediterranean as an Example

Dynamics and predictability of cold spells over the Eastern Mediterranean. Climate Dynamics (under review)

Forecasting seasonal Influenza with a state-space SIR model

Tackling COVID-19: Can the African continent play the long game?

Impact of regional climate change on human health

Progressive ordering with decreasing temperature of the phospholipids of influenza virus

Factors governing the inter-annual variation and the long-term trend of the 850-hPa temperature over Israel

Links between the rainfall regime in Israel and location and intensity of Cyprus lows. International approach

Absolute humidity modulates influenza survival, transmission, and seasonality

Absolute humidity and the seasonal onset of influenza in the continental United States

Absolute humidity and pandemic versus epidemic Influenza

Forecasting seasonal outbreaks of influenza

A mechanistic model of infection: why duration and intensity of contacts should be included in models of disease spread

Modeling and predicting seasonal influenza transmission in warm regions using climatological parameters. PLoS (2017) Modeling the spread of avian influenza viruses in aquatic reservoirs: A novel hydrodynamic approach applied to the Rhône delta (southern France)

Rapid Molecular Tests for Influenza, Respiratory Syncytial Virus, and Other Respiratory Viruses: A Systematic Review of Diagnostic Accuracy and Clinical Impact Studies

Modelling the global spread of diseases: A review of current practice and capability

Health and climate change: policy responses to protect public health

The Lancet countdown: tracking progress on health and climate change

The 2018 report of the Lancet Countdown on health and climate change: shaping the health of nations for centuries to come

On the reliability of seasonal climate forecasts

A nice day for an infection? weather conditions and social contact patterns relevant to Influenza transmission

Impact of climate change on human infectious diseases: Empirical evidence and human adaptation

Modelling seasonal influenza: the role of weather and punctuated antigenic drift

Relationship between humidity and Influenza A viability in droplets and implications for Influenza's seasonality

Developments and prospects in synoptic climatology

The effects of synoptic weather on influenza infection incidences: a retrospective study utilizing digital disease surveillance

This cross-border collaboration is a part of the Track Ⅱ initiative of The AravaInstitute for Environmental Studies, Israel. As such, we would like to thank Robin Twite and Dr. Yara Dahdal for initiating cross-border collaboration. We would like to thank the team of the ICDC and Maccabi Healthcare Services, who routinely collect