1 Introduction

Numerous studies, carried out in different parts of Europe on the basis of observational data, confirm that hot extremes of air temperature increase in frequency (Della-Marta et al. 2007; Fioravanti et al. 2016; Klein Tank and Konnen 2003). At the same time, length and intensity of heat waves have been increasing. In Western Europe, length of heat waves has doubled since the end of the nineteenth century (Della-Marta et al. 2007). This implies not only serious economic consequences, but also considerable health impact. In the twenty-first century, two devastating heat waves have occurred in Europe. The first one, in 2003, may have caused 70,000 deaths in 12 countries in Western Europe (Robine et al. 2008). The second one, in 2010, covered the eastern part of the continent and could have caused 56,000 deaths (Munich Re 2011). Since the second half of the twentieth century, no single natural disaster in Europe, including the earthquakes, has caused such a high number of fatalities. So, one can say that heat waves are currently the most deadly natural disasters in the European continent.

Similarly to other parts of Europe, heat waves occur also in Poland, generating health risk. Increasing frequency of high air temperature, including hot days and very hot days (Graczyk and Kundzewicz 2014; Graczyk et al. 2017; Wibig 2012) has been observed. Also unfavorable demographic changes (aging population) take place in Poland. Advanced age is considered to be one of the most serious health risk factors during heat waves. In 1990, the population aged 65+ accounted for 10.2% of the total population in Poland, while in 2012, it was already 14.2%. Predictions indicate that by 2025 it can reach the level of 21.5%, i.e., more than twice as much, in relative terms, as in 1990 (GUS 2014).

The most intense pan-European heat wave, recorded in 2003, did not affect Poland. Although the heat wave of 2010 was felt almost everywhere in Poland, it did not produce such catastrophic consequences as farther east and in the Russian Federation, in particular. In addition to the aforementioned heat wave 2010, other intense heat waves (1992, 1994, 2006, 2014, 2015) occurred in Poland in last three decades, leading to serious consequences, also in the health sector.

So far, the problem of increased mortality during heat waves was most frequently mentioned in literature in relation to the Mediterranean and Western Europe in the context of the heat wave of 2003. Significantly, fewer references examine health impact of heat waves in Central and Eastern Europe, including Poland. Authors generally restrict their attempts to establishing correlation between the threshold values of temperature or more complex indicators of thermal discomfort and heat stress and the number of deaths. Existing studies for the Polish territory (Kuchcik and Błażejczyk 2001, 2005; Kuchcik and Degórski 2009) examined earlier records, covering the hot summers of 1992 and 1994, but could not include the more recent cases of heat waves of 2006 and 2010.

The aim of this study is to investigate the influence of thermal conditions prevailing during four exceptionally hot summers: in 1992, 1994, 2006, and 2010 on the number of deaths and the risk of death in large cities in Poland. The total number of deaths that could have been caused by adverse weather conditions occurring during heat waves in 10 most populous cities in Poland was estimated. Additionally, the mortality rates for two risk groups: (i) elderly people of 65+ years of age and (ii) people with cardiovascular deficiencies were also examined.

2 Data and methodology

2.1 Meteorological data

Meteorological data used in the analysis are daily values of minimum, maximum, and average daily air temperature, as well as the daily average value of the relative humidity during the four summer periods subject to analysis (1992, 1994, 2006, and 2010). These indices were based on daily meteorological data, and average values of the daily heat discomfort index, humidex, were calculated, linking together air temperature and humidity. This index, introduced in Canada, has also used in works on similar themes in Europe (for example: Bisanti et al. 2004; Conti et al. 2005). The values of the humidex were calculated using the formula:

$$ \mathrm{H}=\mathrm{T}+5/9\left(\mathrm{e}-10\right) $$
(1)

as in Masterton and Richardson (1979) and Conti et al. (2005), where:

T :

is the air temperature in °C,

e :

is the vapor pressure index, calculated after Bisanti et al. (2004), using the empirical formula:

$$ e=6.112\blacktriangledown {10}^{\left[\left(7.5\mathrm{T}\right)/\left(237,7+\mathrm{T}\right)\right]}\blacktriangledown \mathrm{RH}/100 $$
(2)

where

RH:

is the mean daily relative humidity in %.

2.2 Mortality data

The research material is a database containing information about the deaths in Poland for every day in the period 1989–2012, acquired from the Central Statistical Office of Poland (GUS). It includes information on age of the deceased persons and on the cause of death, according to the International Classification of Diseases (ICD). Data for 1989–2000 were organized according to the ninth revision of the international classification (ICD9) and data after 2000 were coded according to the tenth revision (ICD10). All deaths due to natural causes: ICD-9 from 1 to 799; ICD-10 groups from A to D, as well as deaths of cardiovascular and cerebrovascular diseases (CVD), ICD-9 from 390 to 459, ICD-10 group I, were subject to analysis. Also the age of the deceased persons was taken into account as a risk factor. Specifically, the age category 65+ was distinguished.

Socioeconomic changes, which took place in Poland and other countries in the region of Central and Eastern Europe, have rendered it more difficult to estimate the expected number of deaths (and further, to calculate the increase of risk of death and the number of additional deaths during heat waves). Generally, a downward trend in the number of deaths has been observed in such countries of the region as the Czech Republic (Kysely and Huth 2004), accompanying improvements in health care. In ten large Polish cities subject to analysis, considerable changes in the number of inhabitants have been noted (Table 1). Therefore, the expected number of deaths as an average for all the years for which data are available would not be representative. In addition, a protest action of physicians took place in Poland in 1997 and 1998. During this time interval, doctors did not write down the cause of death in official death certificates and this disturbed the homogeneity of records.

Table 1 Information on Polish cities, subject to analysis, ordered after recent (2012 census) population
Full size table

For each of the analyzed heat waves, an individual reference period was assumed, possibly embracing 3 years (without a heat wave and without data gaps due to doctors’ protests) before and after the examined year with a heat wave. The year 2010 constituted an exception, because the available data end in 2012, so that only the years 2007–2009 and 2011, 2012 were considered (i.e., 2 years after the summer with heat wave). In published literature, reference periods are often of similar or even shorter length, e.g., 5 years in Robine et al. (2007), 4 years in Vandentorren et al. (2004), and 1 year in Conti et al. (2005).

For particular summers with occurrence of heat waves in Poland, the following reference periods were examined:

  • For the summer of 1992: 1989–1991 and 1993, 1995, 1996

  • For the summer of 1994: 1990, 1991, 1993, 1995, 1996, 1999

  • For the summer of 2006: 2003–2005 and 2007–2009

  • For the summer of 2010: 2007–2009 and 2011–2012

2.3 Explanation of analytical process

Despite the use of a 6-year reference period, there is a considerable variability of the expected number of deaths between adjacent days. This is due to the presence of shorter hot periods or single days with high temperature that distort regularity. According to Huynen et al. (2001) and Kyselý (2004), even hot spells of moderate length often result in a significant excess mortality. Therefore, comparisons also used 7-day moving average calculated for each reference period. The increase of risk of death during a heat wave is defined as the percentage difference between the observed number of deaths in the analyzed period and the expected number of deaths calculated for corresponding days of the reference years. Also, the number of additional deaths was estimated that might have been caused by the studied heat wave. The analyzed heat waves did not have a similar course. Only in one case (Summer 2006), an uninterrupted period of occurrence of high temperatures was observed in Poland. In other summers subject to analysis (in 1992, 1994, and 2010), hot days were clearly separated by cooler days.

Therefore, all calculations were carried out for two time intervals:

  • Days with increase in mortality during, or shortly after, the occurrence of high values of air temperature and/or humidex.

  • Days between the beginning of the first period of increased mortality accompanied by high values of air temperatures and/or humidex, and the 30th day after the last day of increased mortality period. Longer interval covering monthly period after the end of the heat wave will be used to estimate the number of deaths that is not associated with the “harvesting effect”.

Figure 1 illustrates the process of determination of time intervals with elevated number of deaths, related to the occurrence of high values of air temperature and/or humidex, for the 1994 heat wave in Warszawa. A band of 90% confidence interval was also calculated.

Fig. 1
figure 1

Illustration of the process of determination of time intervals with elevated number of deaths, related to the occurrence of high values of the air temperature and/or the humidex (Warszawa 1994)

Full size image

3 Results

3.1 Mortality in summer months, compared to other seasons

Despite the marked increase in the number of deaths in Poland, during intense heat waves, in an average year, climatic summer is not the season with the highest number of deaths in the country. In fact, in the years 1989–2012, the greatest number of deaths occurred in Poland during the cool part of the year, culminating in the winter months and attaining a minimum value in June, August, or September, depending on the city (Fig. 2). A similar distribution of deaths holds also for the age group 65+, as well as for the category of persons with cardiovascular deficiencies. Typical annual distribution of deaths, as presented in Fig. 2 largely differs from a distribution in a year with a strong summer heat wave. There is a considerable increase in mortality during heat waves, so that the numbers of deaths can be similar to those in winter months and even higher, for some cities. This applies both to the entire population and to the 65+ age group (see Fig. 3 for Poznań and Warszawa).

Fig. 2
figure 2

A percentage of the monthly contribution to the annual sum of deaths in a year in five Polish cities (average for 1989–2012)

Full size image
Fig. 3
figure 3

Intra-annual distribution of the number of deaths, for all cases and people aged 65+ in Poznan 2006 and Warszawa 1994

Full size image

3.2 Characteristics of the weather conditions and the number of deaths during summers with heat waves

3.2.1 The summer of 1992

During the summer of 1992, temperatures considerably exceeded the long-term mean (for 1961–1990) over a large area of Poland. The highest 7-day anomaly of maximum daily air temperature (Fig. 4a) occurred in the south of Poland, exceeding the long-term mean by more than 10 °C over a large area. In central Poland, the anomaly mostly exceeded the value of 7 °C and only on the coast it was less than 5 °C. A similar geographical distribution at slightly lower values (4–9 °C) was also observed for the 14-day anomaly (Fig. 4b).

Fig. 4
figure 4

The highest anomalies of maximum daily air temperature in the summer of 1992: a 7-day anomaly; b 14-day anomaly

Full size image

The first interval of rising temperatures that lasted longer than 10 days and the accompanying increase in the number of deaths were recorded in the western part of the country in the last 10 days of June and in early July. During this period, the maximum daily temperature approached or exceeded 30 °C, while the value of the humidex exceeded 35°. In comparison with the reference period, the mortality was higher even by more than 40%. The number of deaths of people with cardiovascular deficiencies and in the age group of 65+ years increased by 80% and up to 95%, respectively. The main heat wave commenced in all the studied cities around 20th of July and lasted, interrupted by short periods of temperature drop below 30 °C, for approximately 3 weeks. During the hottest days, depending on the town, temperature reached 34–37 °C and the value of the humidex exceeded 45 °C. Only in Gdansk, located on the Baltic Sea coast, temperature in those days (except for the 10th of August) was below 30 °C. The highest increase in mortality during this heat wave was recorded for most cities in its final phase, which coincided with the occurrence of the highest temperature. The increase in mortality in the total population (all causes, not age restricted) ranged from 41% in Warszawa and Łódź up to 102% in Katowice. The last episode of this hot summer was quite unusual for late August, lasting from 2 days in the north and west to 6 days in the south, with temperatures considerably exceeding 30 °C and the humidex exceeding 45°. A clear mortality increase was observed, which in some cities still remained for about 2 weeks after the end of the heat wave. The number of deaths (all causes and age classes) during this heat wave exceeded 150% of the expected number of deaths for the last days of August and the first half of September. Among people with cardiovascular diseases and circulatory system deficiencies, the increase was up to 200%. The course of temperature (daily maximum and minimum), humidex, and the number of deaths in Warszawa is shown in Fig. 5.

Fig. 5
figure 5

The course of temperature, humidex and the number of deaths in Warszawa in the summer of 1992 (gray thin lines represent 90% confidence interval)

Full size image

3.2.2 The summer of 1994

The highest values for both 7- and 14-day mean maximum temperature anomalies were observed in the summer of 1994. At some places, the maximum temperature during the hottest week of the summer (Fig. 6a) was higher by more than 12 °C than the 1961–1990 average and by more than 10 °C almost everywhere. The mean 14-day anomaly of maximum daily air temperature was only slightly lower (Fig. 6b), exceeding 9 °C for almost 60% area of Poland and being in the range of 8–9 °C in the rest of the country.

Fig. 6
figure 6

The highest anomalies of maximum daily air temperature in the summer of 1994: a 7-day anomaly; b 14-day anomaly

Full size image

The first-time interval during the summer of 1994 with temperatures near or above 30 °C, during which an increase in mortality was noted, took place in the last week of June. For most cities, it began around June 26 and lasted for 3 or 4 days in Warszawa and Bydgoszcz and 8 days in Wroclaw and Katowice. Only in Gdansk, the first heat wave and an accompanying increase in the number of deaths began on July 11, when in other major cities the heat wave already ended. Despite the short duration of the heat wave and a moderate value of the maximum temperature and the humidex, increase in the number of deaths compared to the reference period ranged from 25% in Łódź to 80% in Lublin. The increase in the number of deaths associated with cardiovascular deficiencies was even higher, at around 60% in Poznan and Wroclaw and up to 100% in Lublin. This summer, the main heat wave occurred about 10 days after the end of the first one and lasted more than 3 weeks. On a large area, values of maximum daily air temperature were extremely high for the Polish climate. Lublin was the only city analyzed, where the highest maximum daily air temperature did not exceed 35 °C (reaching 34.2 °C). The highest temperature occurred in Szczecin and Łódź, respectively, 37.8 and 37.6 °C, breaking the previous record-high temperatures in these cities. During the hottest days, the number of deaths doubled in most cities as compared to the reference period. The intra-seasonal distribution of the air temperature, the humidex, and the number of deaths in Łódź in the summer of 1994 is shown in Fig. 7. The increase in the number of deaths ranged from 63% in Krakow to 168% in Bydgoszcz. Even higher values were recorded for deaths of people with cardiovascular problems. In seven out of ten cities, the number of deaths at least doubled, ranging from 205% increase of the expected number of deaths in Katowice to 323% in Bydgoszcz. Even higher numbers were obtained for the age group “65+ years.” For the combined risk groups, the risk of death more than doubled during the hottest days in all cities, reaching 330 and 431% in Łódź and Bydgoszcz.

Fig. 7
figure 7

The course of temperature, humidex, and the number of deaths in Łódź in the summer of 1994 (gray thin lines represent 90% confidence interval)

Full size image

3.2.3 The summer of 2006

The 7-day anomaly of the maximum temperature in the summer of 2006 (Fig. 8a) did not reach as extreme values as in 1994, exceeding 10 °C only locally, but relatively high values in the range of 8–10 °C occurred for most of Poland. The 14-day maximum temperature anomaly (Fig. 8b) exceeded mostly 6–7 °C, while in some places in the south-west, it reached 9 °C.

Fig. 8
figure 8

The highest anomalies of maximum daily air temperature in the summer of 2006: a 7-day anomaly; b 14-day anomaly

Full size image

In the summer of 2006, many days were recorded with maximum temperature exceeding or close to 30 °C, on a large area. In the western and central parts of Poland, days with such high temperatures created a heat wave lasting almost 6 weeks, during which there were 2–3-day intervals with slightly lower temperatures. In contrast to the year of 1994, the level of 35 °C was exceeded only during 1 day and in two cities. The vast majority of days during this period had the maximum temperature in the range of 28–32 °C. The beginning of a heat wave in western and central parts of Poland, and the accompanying increase in number of deaths occurred as early as mid-June. Except for a single day in Wroclaw and Poznan, maximum temperature did not exceed 30 °C, but there were many days with temperatures from 27 to 29 °C. In the end of June, temperatures above 30 °C appeared frequently in all the analyzed cities except for the coastal city Gdansk. The last days of June marked a considerable 2–3-day cooling, when the maximum temperature fell below 25 °C, while in Szczecin, Gdansk, and Bydgoszcz even below 20 °C. In the beginning of July, maximum temperature increased in all analyzed cities, and on the 5th of July, it exceeded 30 °C again in seven out of ten cities. Except for a short and slight cooling around 15th of July, high temperatures lasted until the end of the month. In some cities, temperatures above 30 °C continued for more than 10 days. The highest daily increase in the number of deaths occurred for half the cities already in June when temperatures were not the highest. For the rest of the cities, the largest increase in the number of deaths was recorded during the highest temperatures, and at least a few days after the time interval with temperatures above 30 °C. The highest daily increase in the number of deaths ranged from 33% in Warszawa to 115% in Katowice (for all causes of death). An even greater increase, from 55% in Warszawa to 220% in Bydgoszcz, occurred for deaths of people with cardiovascular diseases. Similarly, a strong risk factor was the age (65+), as the highest daily increase in the number of deaths ranged from 41% in Warszawa to 134% in Bydgoszcz. The course of the air temperature, the humidex, and the number of deaths in Poznan in the summer of 2006 is presented in Fig. 9.

Fig. 9
figure 9

The course of temperature, humidex, and the number of deaths in Poznan in the summer of 2006 (gray thin lines represent 90% confidence interval)

Full size image

3.2.4 The summer of 2010

During the summer of 2010, the highest values (> 10 °C) of 7-day maximum air temperature anomalies (Fig. 10a) were recorded on a small area of north-western and central Poland. The anomaly gradually decreased towards the south-east; however, for more than 65% of the country, the anomaly reached 8–10 °C. The 14-day anomaly (Fig. 10b) also reached a value above 8 °C on a large area of northern and central Poland. Apart from the southern parts, across the whole country, the 14-day maximum temperature anomaly exceeded 7 °C.

Fig. 10
figure 10

The highest anomalies of maximum daily air temperature in the summer of 2010: a 7-day anomaly; b 14-day anomaly

Full size image

In the summer of 2010, abnormally high temperatures occurred in most of Eastern Europe. Poland was outside of the occurrence of the highest temperature anomaly, but in the eastern and northern Poland, temperatures clearly exceeded the long-term average. Also in the west of Poland, a series of days occurred, with maximum temperatures clearly exceeding 30 °C and reaching 35 °C for a few days. The first episode of high temperatures and the accompanying increase in the number of deaths occurred about 10th of June. Despite the short duration (3–5 days) of this heat wave, a marked increase in the number of deaths was recorded. The smallest increase occurred in the cities located in the north of the country (28% in Gdansk and 37% in Bydgoszcz, for all causes). In the south of Poland, days with a doubling of the number of deaths occurred in that interval. The number of deaths increased by 111% in Wroclaw and by 137% Kraków. The longest and the most intense heat wave began in early July and lasted for about 3 weeks. During many days, the maximum temperature remained at around or above 30 °C. In all the cities analyzed, a marked increase in the number of deaths was observed. During the hottest days, growth in the number of deaths in the four largest cities exceeded 50% and in other cities almost doubled. The highest increases in the number of deaths were in Lublin and Bydgoszcz (by 122 and 128%, respectively). High values were observed for mortality associated with cardiovascular deficiencies (up to 156% in Szczecin) and for the age group 65+ (157% in Lublin). In the four cities located in the eastern half of the country, one more (third) sub wave occurred, lasting approximately for 2 weeks. For many days, the maximum daily air temperature fluctuated around 30 °C but did not significantly exceed this threshold even during the hottest hours. The maximum daily increase in the number of deaths during this heat wave ranged from 26% in Warszawa to 142% in Lublin. Deficiencies are 40% in Warszawa and 105% in Lublin. In the age group 65+, daily maximum value of the increase in the risk of death was 162% in Gdansk and the lowest one again in Warszawa (24%). The course of the air temperature, humidex, and the number of deaths in Warszawa in the summer of 2010 is presented in Fig. 11.

Fig. 11
figure 11

The course of temperature, humidex and the number of deaths in Warszawa in the summer of 2010 (gray thin lines represent 90% confidence interval)

Full size image

3.3 Cardiovascular diseases and age as risk factors

Simultaneous occurrence of cardiovascular diseases (CVD) and senior age (65+) were factors responsible for the strongest increase in mortality in comparison to the control, for all surveyed summer seasons with heat waves. This was particularly evident during the most sweltering summer of 1994. Only in one city (Katowice) values for this risk factor were lower than those calculated for all causes and age groups. However, even for Katowice, the highest calculated value can be associated with this risk factor. In the summer of 1994, the increase in the number of deaths for risk factors counted separately was generally higher than for all causes and age groups. For six cities, higher values were found for both factors and for the next three cities—for one factor.

For summers, in which the temperature did not reach such extreme values or heat waves were less permanent, results were not so evident. In 2006, the highest daily increase in the number of deaths for all cities except for Krakow appeared for the combined risk groups (CVD and 65+). For other summers with heat waves, it happened that the essential factor was either related to the age or to the cardiovascular diseases (separately).

Mortality factors during the two summer periods when heat waves differed in duration and amplitudes of air temperature are summarized in Table 2.

Table 2 Age (65+) and cardiovascular deficiencies (CVD) as risk factors during heat waves (numbers in the brackets represent 90% confidence interval)
Full size table

Figure 12 illustrates that, in general, changes of the number of additional deaths during heat waves, both in their milder stages and when temperature attains its peak, are of similar shape for specific risk groups and for all deaths. There are only differences in values. Local maxima are usually highest for the combined risk groups (CVD and 65+), but at the same time one can observe the deepest lows for this risk group. Small differences are apparent during periods of lower air temperatures, separating sequences of hot days. There are, typically, 1–3-day shifts in decrease in the number of deaths and the amount of the decline for individual risk groups.

Fig. 12
figure 12

The course of temperature, humidex, and the number of deaths for all cases (AC), cardiovascular diseases (CVD), age above 65 (65+), and age above 65 with cardiovascular diseases (65+ CVD) in Warszawa in the summer of 1994

Full size image

3.4 The balance of heat waves in selected summer periods

Table 3 summarizes the estimated number of additional deaths during the analyzed four summer seasons with heat waves for the ten largest cities in Poland. Nationwide, the largest increase in the number of deaths occurred in the summer of 1994, when 1076 (90% CI 615 to 1537) additional deaths were recorded during the analyzed period. The highest number of deaths, 397 (90% CI 264–500), occurred in Warszawa, although in relative terms, compared to the population, higher values were recorded during the 1994 heat wave in Poznan and Łódź. These cities are located in the vicinity of the highest maximum temperature anomalies observed. Except for Katowice and Gdansk, differences between the increase in the number of deaths during heat waves and during the longer period including days between the beginning of the first heat wave and the 30th day after the end of the heat wave are small. In other analyzed years, the number of additional deaths in days during heat waves exceeded 700 people and was the highest in cities located in the zone of highest anomalies of air temperature. The sum calculated for individual cities suggest that the most vulnerable are cities with more than 500,000 inhabitants, in which in particularly hot years, the number of additional deaths increased by more than 100 cases. The exception is Wroclaw, the 4th on the population list (Table 1), despite the fact that twice it was located in the zone of highest temperature anomalies, the increase in the number of deaths has not exceeded 100 people, for any of the hot summers.

Table 3 Estimated number of additional deaths (in absolute terms) during four extremely hot summer seasons (numbers in the brackets represent 90% confidence interval)
Full size table

4 Discussion

Many studies on the mortality increase associated with heat waves is focused on threshold values in air temperature or indicators based on a combination of temperature and humidity at which an increase in the number of deaths can be observed. Quantitative values reported in various references differ from each other, often very substantially, depending on geographical location of an analyzed city or country. For instance, Ishigami et al. (2008) examined cities with mean summer temperatures from 20.4 °C in London to 26.3 °C in Milan, while Laaidi et al. (2006) studied different regions of France with mean summer temperatures from 17.8 to 23.6 °C. Ma et al. (2015) researched the issue for different regions of China, with temperature values varying significantly, from 19.5 °C in the north of the country up to 27.4 °C in the south. For England and Wales, Hajat et al. (2007) studied even lower values (17–18 °C). The differences are also seen in the work of analyzing various country-specific thresholds of the maximum daily air temperature, like about 24 °C for Estonia (Oudin Åström et al. 2016), 26 °C in the Czech Republic (Kysely and Huth 2004) and 31 °C in Italy (Zauli Sajani et al. 2002). In Poland, the number of additional deaths in largest cities increased noticeably, depending on the region, when daily maximum air temperature exceeded 26–28 °C. In areas with warmer climates, threshold temperatures at which an increased risk of death can be noticed are also generally higher than in regions with more moderate climate. One explanation may be the lifestyle (e.g., a siesta) and adaptation of housing to the warmer climate. Physiological acclimatization can also play an important role (Keatinge et al. 2000).

During summers investigated in this work, significant difference in change of mortality can be observed, between the cities, despite the relatively small differences in climatic conditions. Similarly to the results of Oudin Åström et al. (2016), the additional mortality caused by heat in the area on the coast is lower than inland. This is also in accordance with previous studies, such as Urban et al. (2017) and Medina-Ramon and Schwartz (2007). The largest number of deaths associated to the occurrence of high air temperature was recorded in cities with the highest population density (and, simultaneously, with the largest population). The urban heat island is probably most intense in these cities, which according to Tan et al. (2010) affects the intensity of heat waves and thus increases the exposure to unfavorable thermal conditions. Among the Polish cities of similar size and population density, such as Wroclaw and Poznan, there were large differences in the number of deaths during heat waves in 1994, in similar weather conditions. Hajat and Kosatsky (2010) and Hondula et al. (2015) explain such a difference in mortality rates by the age of population and economic conditions. Urban et al. (2017) draws attention to the differences in the physical environment, while Son et al. (2016) to the differences in urban vegetation.

To the knowledge of the present authors, this work is the first one that seeks to estimate the additional mortality during particularly hot summer periods in Poland. The results indicate that by far the most deadly heat waves occurred, almost throughout the entire Polish territory, in the summer of 1994. In the 10 largest cities, the additional mortality associated with extreme heat almost reached 1100 people, including almost 400 in Warszawa itself (Table 3). In other studies related to individual cities, serious health consequences of the heat wave in Athens in 1988 were recorded, when the number of additional deaths was estimated by Katsouyanni et al. (1988) at the level of about 2000 people. A heat wave in Chicago in 1995, lasting for only a few days, was found responsible for over 700 deaths (Semenza et al. 1996). There were 201 additional deaths spotted in 2012 in the capital of Moldova (Corobov et al. 2013), Chisinau, being of similar sized as the studied cities in Poland.

Nationally, the number of approximately 1100 additional deaths estimated in Poland in 1994 is significantly lower than those recorded in France, Italy, and England in 2003, being 14,700 (Fouillet et al. 2006), 3134 (Conti et al. 2005), and 2234 (Green et al. 2016), respectively. In the neighboring Czech Republic, the largest number of deaths probably also occurred in 1994 and could reach almost 2000 (Urban et al. 2017) and around 1500 in 2015. In Slovakia, the highest mortality rate (540) was recorded in 2015 (Výberči et al. 2017). The present study, however, covers only the ten largest cities inhabited by a total of ca. 16% of the Polish population. Probably, the mortality increase due to heat waves in 1994 was significantly higher although it is not possible to assess this quantitatively, without conducting more research (whereas collection of mortality data would be extremely difficult). The increase in mortality during heat waves is the highest in large and densely populated urban centers. However, there is also a problem of mortality increase in smaller cities and rural areas, where, jointly more than 83% of the Polish population reside. Results of Conti et al. (2005), for the heat wave of 2003 in Italy, show an increase in the number of deaths by 13.8% in cities with less than 100,000 inhabitants. The increase in health risk associated with heat waves in rural areas was also reported by Gabriel and Endlicher (2011) and Hajat et al. (2007). According to Sheridan and Dolney (2003), especially during extreme thermal conditions, an increase in the number of deaths in rural areas and the suburbs may be similar or even higher than in cities.

While analyzing the course of temperature and the number of additional deaths during hot summers in Poland, if can be noticed that the first even relatively mild and brief episode of temperature increase is often associated with the occurrence of a high daily increase of mortality. During the first hot days, mortality of the elderly and those suffering cardiovascular diseases is much higher (by 200% and more) compared to the days of the air temperature close to normal. This agrees with results of previous studies (Hajat et al. 2002; Kyselý 2004; Kyselý and Kim 2009). Similarly, high values of daily numbers of deaths occur later, at much higher temperature values or longer duration of heat waves. Allen and Sheridan (2015) explain reduction in the risk during the summer season by acclimatization. The heat waves studied in this work are characterized by different intensity and duration.

According to D’Ippoliti et al. (2010), heat wave duration plays a greater role than the intensity but the combination of high values of these two factors caused the greatest increase in mortality for most cities in Italy. Long heat waves resulted in the daily number of deaths being 1.5 to 3 times higher than during short heat waves. Similarly, in the case of the analyzed periods of summer in Poland, the largest number of deaths occurred in 1994, when each of the two factors attained high values, i.e., a high maximum daily air temperature during the hottest days (in excess of 35 °C) and the length of the heat wave exceeding 10 days. During the heat wave of 2006, in a part of Poland, a long heat wave was recorded that lasted continuously for about 3 weeks, with temperatures approaching or exceeding 30 °C but not reaching 35 °C. The increase in the number of deaths during this wave was lower than in 1994, which indicates that, for conditions of temperate climate of Poland, the intensity of a heat wave may be more important than its length. Both of these summer seasons have been considered as particularly hot also in the German region of Brandenburg and Berlin, neighboring with Poland. While the temperature values were similar to western Poland, also there the heat wave of 1994 resulted in higher mortality. Also experience from the heat wave in 2003 in France show that the intensity of the heat wave is crucial. From 1 to 20 August 2003, temperatures exceeded the long-term average of the maximum temperature by 11–12 °C and over a large area for nine consecutive days, the Tmax did not fall below 35 °C (Fouillet et al. 2006). A particularly dangerous combination of hot days and tropical nights (Tmax > 35 °C and Tmin > 20 °C) occurred in a few large cities for 9 to 15 days (Vandentorren et al., 2004). A significant increase in the risk of death in elderly people associated with the presence of high value of both the maximum and minimum daily air temperature also was observed by Heudorf and Schade (2014) in Frankfurt am Mein, Germany. Similar conditions, lasting 3–4 days, occurred in 1994 in Poland, which may explain the higher number of victims of heat compared to other years analyzed.

Estimating the number of additional deaths caused by extreme thermal conditions during hot summers is difficult, because in many studies, it is presented as a short-lived effect followed by a compensation in the form of a reduced number of deaths occurring within a few weeks after the end of a heat wave (Keatinge et al. 2000; Huynen et al. 2001; Kyselý 2004). This leads to the interpretation that most of the victims of heat are people whose health could have allowed them to survive the next few days or weeks. Baccini et al. (2013) estimate that 30-day mortality displacement reduces the number of deaths by 75%. In the summers analyzed, the effect of shifting deaths in time did not occur, or was very low in the most intense heat waves and the most vulnerable cities. In such cases, the effect of reducing the number of deaths within 30 days did not occur, while in Warszawa in 1994, there was even a 16% increase. Similarly, a lack of a shift of mortality was also confirmed by Le Tertre et al. (2006) for the heat wave of 2003 in France.

Climate-model based projections indicate a further increase in the number of hot and very hot days in Poland in the future (Błazejczyk et al. 2013; Kuchcik 2013), and the number of days with conditions threatening life and health is likely to increase significantly.

The methodology used in this work should limit the impact of demographic and socioeconomic changes as well as technical progress and health care on the obtained results. When comparing particular hot summer periods, these factors should be taken into account. Since the beginning of the 1990s, there has been a significant progress in health care, especially related to cardiac and cardiovascular diseases. According to the Polish Statistical Office (GUS 2015) data, in 1990 these diseases were responsible for 52% of all deaths and in 2013 for 46%. During the same period, the average life expectancy increased from 66.2 to 72.1 years in men and from 75.2 to 80.1 in women. In Poland, similar studies have not been conducted so far, but in countries with developed economies, regardless of geographic location, a decline in the heat-related mortality has been found in last decades (Gasparrini et al. 2015). In countries affected directly by the 2003 heat wave, this may be the result of improvement of preventive measures and warning systems (Fouillet et al. 2008).

In Poland, among both decision-makers and the broad public, this problem is underestimated and downplayed. Warning systems dedicated to such events exist for a short time only and deliver rather general messages. They are part of a broader crisis management system that was created in the country in 2008 and has so far mainly focused on phenomena related to gale wind and heavy precipitation. Perhaps this is related to the fact that Poland has not yet experienced a large-scale heat wave, similar to that in Western Europe in 2003. The lack of such a system in the Czech Republic was also reported by Urban et al. (2017).

5 Conclusions

An evident increase of mortality, especially during very hot summer periods, can also be observed in countries like Poland, where maximum air temperatures are not as extreme as in southern parts of Europe. In the analyzed interval of 23 years, for at least some cities, mortality increase during heat waves took place during four summers, which proves that the problem does not restrict only to continental-scale, record-beating heat waves.

The increase in the number of fatalities during particularly intense and long-lasting heat waves is not characterized only by a temporal shift of mortality caused by the harvesting effect. Within the most affected urban areas, it is compensated only slightly (up to 25%) or even not at all, by a later decrease of the number of fatalities. This indicates that the majority of additional victims could survive for at least 30 days after the heat wave.

There are large differences of the mortality increase during hot summer periods between analyzed municipalities. Apart from demographic factors, these differences can be explained by the population size that influences the urban heat island effect and by the population density, which implies, inter alia, the size of green areas (parks, lawns, gardens, urban forests).

Increase of mortality risk during heat waves in Poland is particularly high for people older than 65 years and suffering from cardiovascular diseases. During the hottest days, the number of fatalities in some cities was more than three times higher than the mean mortality in the reference period.

For examined cities, the highest rises of mortality risk took place during heat waves that were both intense and long-lasting, i.e., with maximum air temperature exceeding 35 °C and heat wave duration longer than 4 weeks (for all sub-waves together). For long-lasting heat waves in urban conditions in Poland, their intensity seems to be of higher importance than their duration.