1. Introduction
Heat waves have become one of the most significant health concerns globally as well as in Korea. The increase in extreme temperature events is affecting a variety of conditions relevant to human health, such as ischemic stroke, ischemic heart disease, acute myocardial infarction, angina pectoris, heat-related illness, and mental illness all over the worldwide [
1,
2,
3,
4,
5,
6,
7,
8,
9]. Severe heat waves in Korea have also significantly increased mortality and morbidity, especially in 1994 and 2018 [
10,
11]. Heat waves also have adverse physical impacts: heat-related diseases occur more frequently and the mortality rate increases during extreme heat episodes.
The effects of the global 2018 heat wave highlighted the importance and urgency of having sophisticated heat wave policies. The damage from the 2018 heat wave was reported from all over the world [
3,
10,
12,
13,
14]. In Korea, 48 deaths from heat-related diseases were reported—a figure that was twice that from the last three years. In 2018, all records related to high temperatures, such as daily maximum temperature, daily minimum temperature, sunshine hours, heat wave days, and tropical nights, reached the highest ever documented since 1907. Korea has legislated heat waves as a form of natural disaster [
15].
The impacts of heat waves are not only determined by their severity and frequency, but also by the socioeconomic factors, age, occupation, income, and gender of those it affects. Elderly people are more vulnerable to high temperature [
16,
17,
18] than younger people, and people with low incomes are more vulnerable than people with high incomes [
18,
19,
20,
21]. This is because the elderly and people with low incomes are less able to protect themselves and to respond promptly [
22]. Occupation is also an important factor that affects a heat wave’s impacts. Blue-collar workers (relative risk 1.06) are more vulnerable to high temperatures than white-collar workers (relative risk 1.01) [
23].
The effects of temperature on health are also different depending on the climate characteristics of the affected region. Lowe et al. showed that northern Europe is more sensitive to heat than southern Europe. In Denmark, deaths from temperature start to occur at temperatures 5 °C lower than that at which deaths start to occur in southern Portugal [
24]. Gómez-Martín et al. insisted that response action to avoid heat waves could be influenced by one’s own personal experiences or one’s community’s experiences with heat waves [
25].
Many studies on temperature have revealed the statistical significance of increases in mortality and morbidity due to increasing temperatures (e.g., [
18,
20,
26,
27,
28,
29]. Most studies focus on relative impacts individually by age, occupation, income, and climate conditions [
4,
5,
11,
18,
20,
29]. The impacts of extreme high temperatures are contextualized in specific regions reflecting the multiple factors described above. However, previous studies have not provided enough information for establishing customized heat waves policies for vulnerable groups at the regional level. For this study, we attempted to quantify changes in mortality rate from temperature considering physical factors (temperature), socioeconomic factors (age, occupation, household type, chronic diseases), and temperature distribution in 229 regions in Korean in order to provide customized heat wave policies by region and by group.
2. Methods
This study analyzed the demographic characteristics of 229 basic units of local government in Korea and examined the impacts of mortality from temperature. The study used daily maximum temperature and daily mortality data from 2007 to 2016 (June to August). Mortality data was obtained from Microdata Integrated Service from Statistics Korea [
30]. Meteorological data was obtained from the Korea Meteorological Administration’s meteorological data release portal [
31] (
Table 1).
Mortality data, including cause of death, age at the time of death, occupation, and marital status, were obtained from the Statistics Korea. Total mortality was considered to include all deaths except those from external causes (International Classification of Diseases 10th Revision (ICD-10) codes A to R); demographic characteristics included age, household type, occupation, and chronic diseases. People were divided in those 65 those years of age and older (the “elderly”) and those less than 65 years of age, and occupations were classified as outdoor and others. The former included agricultural, forestry, and fisheries workers; device and machine operators and assemblers; construction workers; and military personnel. Households were divided into one-person households and others. Chronic diseases included heart diseases (ICD-10 I2–I5), liver diseases (ICD-10 K70–K77), diabetes mellitus (ICD-10 E10–E14), and hypertension-related diseases (I10–I15) [
16,
32,
33,
34,
35].
A total of 535,495 people were considered in the study, which covered over a ten-year period; 402,769 of the people included in the study were over 65 years of age. Those from one-person households and outdoor workers accounted for 276,824 and 90,266 deaths, respectively. The factors considered for this time were age, outdoor workers, one-person households, and chronic diseases (
Table 2).
Meteorological data were obtained from 440 automated weather systems operated by the Korea Meteorological Administration and interpolated into 1 km regular grids and zonal statistics for each region. The Gaussian process regression model was used for the former and only the temperatures of urban and agricultural regions were extracted using zonal statistics.
The mortality change rate (MCR), which is the daily change rate of mortality at a given temperature per average summer mortality, was analyzed after it was calculated using the following Equation (1):
where Mt is the average daily mortality rate above a specific temperature and Ms is the average daily mortality rate during the summer.
MCR is calculated by the following procedure (
Figure 1). Socio-economic conditions to be analyzed are set and mortality data corresponding to socio-economic conditions are distinguished from the data set. Next, the temperature to analyze is set. The specific temperature was applied at intervals of 1 °C from 25 °C to 35 °C. Then, the target municipality and its adjacent areas is selected with temperature and mortality data obtained from 2007 to 2016. The data of adjacent municipalities are considered by using a spatial smoothing method because the confidence interval of the analysis results are widened when only a single sample from the region is used. Daily mortality was converted to daily mortality rate in order to consider the changes in total population. Daily mortality data is used to calculate Ms and Mt. Daily mortality data above a specific temperature is used to calculate Mt for the 10-year period. MCR is calculated by taking the natural logarithm to Ms and Mt. Logarithmic transformation is used to consider offset regional differences in population size, and 1 is added to Ms and Mt to remove errors when the mortality rate is 0. MCR is iteratively calculated by municipalities and 1 °C.
The study area was ranked based on the average daily maximum temperature in the summer over the last 20 years (1997–2016) and classified into A regions (areas with higher temperatures) and B regions (areas with lower temperatures). In Korea, maximum temperatures in summer are high in inland metropolitan and basin areas but low in areas influenced by the ocean and in rural areas, including high mountainous areas.
Figure 2 shows two types of regions categorized based on the average maximum summer temperature. A regions have relatively high temperatures and B regions have relatively low temperatures (
Figure 2a). There is an average difference of 2 °C in the maximum daily temperatures of the two types of region; the distribution of temperatures in the histogram for A regions is shifted to the right because high temperatures are frequent (
Figure 2b).
4. Discussion
This study analyzed the MCRs of outdoor workers by age, household type, pre-existing chronic health condition (a chronic disease), and the distribution of temperatures in their region. Outdoor workers were most sensitive to temperature increases.
Table 3 summarizes the MCRs of outdoor workers.
The impacts of temperature increase when the proportion of time an individual spends working outside is high and when physical exertion levels while working are high [
36,
37]. Outdoor workers who experienced heat-related illness are highly correlated with exposure to extreme high temperature (95.7%) and work after struggling to sleep in tropical nights (78.7%) [
38]. Outdoor workers who are vulnerable to heat waves are 224 difficult to manage their working conditions voluntarily Outdoor workers who are vulnerable to heat waves are difficult to manage their working conditions voluntarily. It is difficult to manage the working conditions of outdoor workers who are vulnerable to heat waves if the implementation of changes to working conditions are made solely on a voluntary basis. Policy management for these workers needs to be enhanced.
The sensitivity of outdoor workers to temperature was higher in one-person households. The influence was relatively higher in young people (aged less than 65 years). In addition, household type and pre-existing chronic disease were important factors not just for the elderly but younger people as well. This indicates that further attention should be given to people living alone and with chronic disease.
Social isolation is a critical risk factor of heat-related mortality [
39]. People who live alone, live on the top floor, and stay mainly in the house showed a high odds ratio for heat-related deaths during the Chicago heat wave of 1995 [
40]. Impacts of heat waves are disproportionately felt by elderly one-person households who have low incomes, weak social networks, and suffer from serious illnesses during the events [
41]. Recently, it was stated that policies to prevent social isolation, such as community-based active monitoring programs, would reduce the impacts from heat waves [
42]. However, more studies on social isolation factor in heat wave research are needed. This study showed that social isolation is a critical factor to determine heat wave mortality.
The MCR of young outdoor workers starts to decrease when heat wave warnings are given (
Table 3). However, the MCR of elderly outdoor workers increased consistently with temperature regardless of age, household type, and regional temperature distributions. This implies that heat wave policies may not have been effective for this group. Elderly outdoor workers mainly work in the fields of agriculture, forestry, and fishing. It is difficult to follow guidelines and policies. Heat wave policies are shared via media, short message service, and apps in Korea, making it difficult for the elderly to obtain heat wave information via electronic devices. Young people are prone to accommodate experience, information, education, and policies about heat waves, all of which is useful for reducing their heat impacts [
25]. If elderly people also receive the appropriate information at the individual level, the heat-evasive behavior increases [
43]. This study indicated that information about the impacts of heat waves should be more actively transmitted to elderly people.
We found that mortality would increase at lower temperatures in the lower temperature regions (B regions). In lower temperature regions, the MCRs of both the elderly and young people showed a steep increase above 31 °C. This means that the heat wave warning system, which is based on daily maximum temperatures (level-1: 33 °C, level-2: 35 °C) may be limited in terms of its ability to reduce the health impacts in those regions. Health impacts vary with different climate zones, as has been shown in studies in Europe [
24], the U.S. [
44], and China [
29]. We found that health impacts vary in the meso-climate zone in Korea. Thus, heat wave policies need to be customized to the temperature distribution of each region, as well as socio-economic factors like age, occupation, and household type.
This study did not consider lag effects. We focused on impacts from temperature distribution and socio-economic conditions. Although previous studies showed the impact of the lag effect varies with region, climate factors, and exposure [
44,
45], lag effects of heat waves occur in a short period of time, often within a few days [
5,
7,
16,
26,
46,
47,
48,
49].
The duration and severity of heat waves are projected to increase in the 21st century [
50]. The aging and polarization trends occurring in Korea are likely to increase the health impacts from future heat waves [
26]. Korea has become an “aged society” because of its low fertility rate and an increasing population of the aged; the proportion of the elderly in the population is expected to reach 25.0% in 2030 and 43.9% in 2060 [
51]. According to Kim et al. mortality from heat waves is expected to increase five times in the 2060s under the representative concentration pathway 4.5 scenario [
18]. It is important to understand the differences in the impacts of heat waves according to socioeconomic characteristics. Carefully designed policies based on contextualized impacts at the local level are required to prevent further heat wave damages in the future.
5. Conclusions
This study analyzed MCRs in 229 municipalities in Korea while considering age, occupation, household type, chronic diseases, and regional temperature distribution. We found that the MCRs for one-person households of the elderly, outdoor workers, and people with chronic diseases are relatively higher compared to other groups. Of these, a significant difference was observed between the MCRs of outdoor workers and others.
The patterns of the MCR with temperature varies with the age of the people affected. The MCR of elderly outdoor workers increased steadily with temperature, but that of young outdoor workers decreased after the heat wave warning level, especially in B regions. These results suggest that young outdoor workers in these regions are not responding adequately to high temperatures before the heat wave warnings are issued. This also means that current heat wave warnings and policies may not be effective for elderly outdoor workers. It implies a need for contextualized heat wave policies considering key factors. This study found that regional temperature distribution is one of the key factors that should be considered to determine heat wave impacts. It is suggested that it is necessary to consider regional temperature distributions when setting heat wave warning levels and building practical and effective policies.