Establishing a Comprehensive Monitoring System for Natural Disasters: The Associations between Wildfire Exposure and Health Outcomes

Abstract

This study examines the health impacts of a 2017 wildfire on residents in both exposed and non-exposed areas, highlighting the increasing importance of health consequences related to natural disasters in contemporary society. The research utilizes retrospective data to compare health outcomes before and after the wildfire, between the two groups. Health impacts are assessed through rates of outpatient visits and hospitalizations, with a focus on conditions such as congestive heart failure, chronic obstructive pulmonary disease, and pneumonia in the exposed area. The study conducted rapid analysis using advanced techniques, including the double-difference method and time series analysis, to identify significant changes and outcomes. Expert consultations and the Delphi method enhanced the accuracy of the analysis, aiding in the identification of vulnerable areas and health conditions. The research emphasizes the importance of collaboration with various agencies to secure data during the pre-preparedness phase and recommends proactive on-site investigations and tracking and observation protocols during retrospective analysis. In the response phase, the focus was on rapid data collection, advanced analysis identification of high-risk populations in exposed areas, and precise tracking and response strategies. Furthermore, the study suggests the potential for proposing legislative measures to support residents in exposed areas. The establishment of a comprehensive and systematic monitoring system based on the findings of this research, is expected to contribute to a deeper understanding of natural disaster exposure and its associated health impacts, as well as to the development of effective response strategies.

Key messages

① What is known previously?

When natural disasters such as bushfires occur, environmental factors contribute to an increase the risk of noncommunicable diseases, including respiratory and cardiovascular diseases.

② What new information is presented?

The monitoring system developed and applied in this study revealed, through data analysis, that outpatient and hospitalization rates for ischemic heart disease and pneumonia increased among residents exposed to the wildfires.

③ What are implications?

The findings from this study highlight the importance of minimizing the health impacts of natural disasters by establishing robust pre-event and post-event preparedness and response monitoring systems. This research is poised to contribute to the formulation of effective prevention and response strategies, aiming to mitigate the health effects of natural disasters and support governments and relevant agencies.

Introduction

According to the World Health Organization (WHO), non-communicable diseases (NCDs) are a prominent cause of global mortality [1], responsible for approximately 71% of all deaths [2]. In contrast to infectious diseases, which result directly from pathogenic agents, NCDs emerge from a complex interplay of lifestyle, genetic, and environmental factors [3]. This category encompasses conditions such as cardiovascular disease, cancer, diabetes, chronic respiratory disease, and more, all of which, over the long term, significantly diminish an individual’s health status and overall quality of life [2]. Consequently, both effective monitoring and management of NCDs are crucial not only for enhancing individuals’ quality of life but also for bolstering public health on a national level and curbing healthcare costs.

The WHO operates various monitoring systems and programs worldwide. It has created the Global NCD Monitoring Framework to oversee the incidence and mortality of NCDs, as well as the policies and actions adopted by countries in addressing them [4]. This framework helps member states gain a better understanding of the current NCD status within their borders and supports the development of essential preventive and response strategies [4]. Furthermore, it facilitates the creation and execution of national and regional strategies concerning NCDs and their risk factors, providing a foundation for the implementation of more effective countermeasures [5]. The STEPwise approach to surveillance (STEP) is a standardized framework for monitoring NCD risk factors, developed by the WHO [6]. This system provides for regular examination of the primary risk factors for NCDs, facilitating the comparison and analysis of these factors and the incidence of NCDs across countries facing similar conditions [6]. The Global Burden of Disease Study (GBD) is a comprehensive assessment of the global burden of disease led by the Institute for Health Metrics and Evaluation (IHME) [7]. The GBD study offers national and regional data on various diseases, causes of death, and NCDs. These data are instrumental for understanding patterns and changes in the global burden of disease [8]. These international monitoring systems and studies play a crucial role in clarifying the transmission and incidence patterns of NCDs. This understanding enables countries to develop and implement measures for reducing the incidence of and mortality from NCDs through comparative monitoring across nations.

In the Republic of Korea (ROK), several surveys and analyses are conducted through the National Health and Nutrition Examination Survey (NHANES) and the Health Insurance Review and Assessment Service. The NHANES is a representative sample survey that plays a crucial role in shaping health policies. It serves as the foundation for initiatives such as national health plan policies and health promotion programs by investigating the health level, health behaviors, and food and nutrition intake of the population. This survey examines key health metrics such as weight, height, blood pressure, blood sugar, and cholesterol levels. Additionally, it gathers information on lifestyle habits in relation to NCDs. The collected data provide valuable insights for assessing and managing the risk of developing NCDs such as obesity, hypertension, diabetes, and hyperlipidemia [9]. On the other hand, the Health Insurance Review and Assessment Service conducts an examination and analysis of health and healthcare utilization patterns of the population using health insurance claims data. This analysis yields detailed healthcare information, including the specific conditions for which people seek medical attention and the treatments they receive. For instance, the Health Insurance Review and Assessment Service analyzes healthcare utilization patterns among patients with NCDs, including prescription drug use. This analysis provides essential information on NCD prevalence, treatment patterns, and the effectiveness of treatments. Such insights are vital for the effective treatment and management of NCDs, informing the development of pertinent health policies. These surveys and analyses play a key role in identifying NCD incidence and associated risk factors. Through use of these data, targeted health policies can be established and implemented, contributing significantly to improving overall population health and reducing the burden of diseases.

Previous studies have revealed the association between natural disasters and NCDs [10]. Specifically, the risk of respiratory diseases increases following events such as wildfires owing to the impact of fine particulate matter [11]. Additionally, heat waves can result in various health issues, including heat stroke, dehydration, and kidney disease [12]. These impacts are especially pronounced in vulnerable populations, necessitating more systematic healthcare policies. However, the transient and unpredictable nature of natural disasters poses challenges to our capacity for immediate and effective response. Despite the global increase in natural disasters driven by extreme weather events [13], there is a notable lack of research on NCDs resulting from these disasters. This underscores the importance of monitoring and managing NCDs induced by natural disasters.

Internationally, and in contrast to domestic efforts, various organizations in the United States and Europe continue to examine the correlation between natural disasters and NCDs, offering diverse guidelines and resources based on their research findings. Notably, entities such as the Centers for Disease Control and Prevention (CDC) and the U.S. Environmental Protection Agency (EPA) furnish guidance and a range of resources aimed at mitigating health risks and facilitating effective responses to natural disasters [14]. In Europe, the European Center for Disease Prevention and Control (ECDC) and the European Environment Agency (EEA) remain actively engaged in research and investigations concerning the connection between disasters and disease, with the goal of enhancing national response capabilities. Drawing from insights gained in previous studies and international experiences, it is clear that establishing a comprehensive natural disaster and NCD monitoring system in the ROK is imperative. This entails a systematic analysis of the correlation between natural disasters, such as wildfires, and NCDs. Such a monitoring system would enable the effective management of health risks during disasters and contribute to the minimization of social and economic losses.

This study seeks to propose an investigation and surveillance system for NCDs by analyzing the correlation between wildfire occurrences and the emergence of NCDs during natural disasters. The aim is to develop a system capable of monitoring and responding to the emergence of NCDs in the aftermath of such disasters. Through the identification of health effects associated with natural disasters and timely preparation and response to vulnerable areas, this study aims to minimize and manage the health impacts of NCDs on both individuals and society as a whole, ultimately reducing social losses.

Methods

1. Target

1) Selection of wildfire-exposed areas

The study utilized Korea Forest Service data to identify significant 2017 wildfires for the exposure group. These wildfires were chosen based on exceeding 100 hectares and lasting more than 24 hours. Selection criteria included wildfire duration, data availability from January 1, 2012, to December 31, 2020, and potential health impacts.

Wildfire-exposed areas were categorized into wildfire-affected areas and expanded wildfire-affected areas, with corresponding control areas selected for each area. The assessment of expanded wildfire-affected regions considered the radius from the wildfire origin and the occurrence of the wildfires. The National Institute of Environmental Science, affiliated with the Ministry of Environment, analyzed footage acquired by Chollian Satellite 2B, the geostationary environmental satellite. The analysis revealed that concentrations of ground air pollutants from the wildfires on the east coast were at least 3.5 times and up to 24 times higher than usual in the observation network. This included elevated levels of up to 385 μg/m³ for ultra-fine particles, and up to 0.028 ppm and 3.8 ppm for nitrogen dioxide and carbon monoxide, respectively [15]. Hence, areas where ambient air pollution level data were accessible were given further consideration. The Korea Forest Service provided data on wildfire occurrences, and air pollution information was available from the Ministry of Environment’s Air Korea, exclusively at the eup, myeon, and dong levels, signifying village and town-scale units. However, recognizing that wildfire impacts may span across multiple villages, and with only one air pollution monitoring station per municipality (or city, county, or district) in mountainous areas, this study established the municipality level as the regional selection unit. The expanded wildfire-exposed areas were selected in the sequence specified below, outlining the wildfire-exposed areas (Table 1).

Table 1. Summary of wildfire exposure areas.

Division	Forest fire area	Expanded forest fire exposure area
		Air pollution measurement station
		Non-installed area	Installation area
Definition	Wildfire origin area	Cities and counties within a 50 km radius of the source	If the source is within a 50 km radius and the maximum PM10 concentration on the day of occurrence is above the standarda)

^a)Baseline value=PM10 24h-maximum concentration the day before the occurrence (2017.05.05)+2SD of PM10 24h-concentration for the month before (2017.04.06–2017.05.05).

(1) Areas within a 50-kilometer radius of the origin were chosen.

(2) Regions with a history of significant wildfires within one year before and after the occurrence date were excluded.

(3) In locations with a monitoring station, wildfire-exposed areas were selected if the 24-hour maximum concentration of PM10 on the day of the event (May 6, 2017) was greater than or equal to the sum of two times the standard deviation (SD) of the 24-hour maximum concentration of PM10 on the day before the event (May 5, 2017) and the 24-hour maximum concentration of PM10 in the month before (i.e., April 6 to May 5, 2017).

(4) An advisory committee comprising industrial and environmental health experts reviewed the validity of the chosen wildfire-exposed areas and extended wildfire-exposed areas.

(5) In areas without monitoring stations, the selection was based on the presence of the wildfire origin and the occurrence of wildfires.

2) Selection of control areas exposed to wildfires

To assess the health effects of wildfires, control areas were chosen based on the wildfire occurrence status provided by the Korea Forest Service and KOSIS data from Statistics Korea. Apart from the exposed areas, this study randomly picked cities, counties, and districts with comparable demographic characteristics to those of the exposed areas. These districts were selected from among those without significant wildfires (damage area of 100 hectares or more) within one year before and after the specified period (May 6, 2017). The demographic characteristics taken into consideration included age, gender, farm population, the number of basic living recipients, and the number of elderly care institutions. The incorporation of additional variables that reflect local characteristics reduces the number of eligible cities, counties, and districts available for matching. Therefore, three key variables representing local characteristics—age distribution, gender distribution, and farm population distribution—were utilized for the matching process. To ensure randomness, the population was divided into five-year increments according to their average age. Additionally, in the random selection process, the average age was categorized into five-year intervals. Furthermore, the gender ratio and the agricultural population’s proportion were calculated relative to the middle-of-the-year population of each city and county. These data were then classified and utilized based on the three quartiles. The same exercise was conducted to determine the ratio of each city and county’s population involved in farming. For both wildfire areas and expanded wildfire-exposed areas, a corresponding control area was selected. Areas without available control candidates due to a lack of suitable matches were excluded from selection (Table 2). Additionally, this study presented a summary of demographic characteristics to provide an overview of both the exposed and control areas (Supplementary Table 1).

Table 2. Wildfire non-exposure candidate area demographic characteristics.

Demographic characteristics	N of regions	Min	Max
Age (yr)
35–39	43	36.1	39.9
40–44	96	40.0	44.8
45–49	53	45.0	49.8
50–54	36	50.1	54.9
55–59	1	55.3	55.3
Gender (%)
Tertile 1	76	47.6	49.6
Tertile 1	77	49.6	50.4
Tertile 1	76	50.4	56.8
Farm population (%)
Tertile 1	76	0.1	5.1
Tertile 1	77	5.2	18.8
Tertile 1	76	19.1	64.3

N=number.

For the 213 control cities, counties, and districts, a 1:10 randomization was conducted for wildfire-exposed areas, and a 1:1 randomization was used for expanded wildfire-exposed areas based on age, gender, and rural population (Table 3).

Table 3. Number of candidate wildfire non-exposure areas.

Wildfire exposure area	Demographic characteristics (age-gender-farm population)	Non-exposure group (N of regions)
Forest fire area
Exposure area 1	2-1-1	11
Exposure area 2	3-2-1	10
Expanded wildfire exposure area
Exposure area 1
Extended exposure area 1	2-0-0	21
Extended exposure area 2	2-2-1	19
Extended exposure area 3	2-2-2	4
⋮	⋮	⋮
Extended exposure area 10	3-2-2	10
Exposure area 2
Extended exposure area 1	2-0-0	19
Extended exposure area 2	3-1-1	4
Extended exposure area 3
⋮	⋮	⋮
Extended exposure area 10	4-1-2	9

N=number.

3) Surveyed disease items

Previous studies have demonstrated that wildfire smoke comprises a complex mixture of gases, including fine particulate matter (PM2.5), carbon monoxide, nitrogen oxides, and volatile organic compounds, which can stimulate respiratory illnesses and trigger immune responses leading to inflammation. Through a review of prior studies and relevant literature, this study identified chronic obstructive pulmonary disease, asthma, pneumonia, ischemic heart disease, and heart failure as the diseases to be observed in the pilot study. These conditions have been cited in the literature as having a significant relationship to wildfires (Table 4) [16].

Table 4. Disease list to be observed in previous research.

Disease	ICD-10 CODE
Chronic obstructive pulmonary disease	J43, J44 (exclude J430)
Heart failure	I50
Asthma	J45
Pneumonia	J10-18
Ischemic heart disease	I20-25

2. Analytical Methods

To assess the acute/sub-acute impacts following the wildfire event, this study examined whether an increase occurred in the utilization of outpatient and inpatient services in both the exposed and unexposed areas. The control group, representing the unexposed area, was defined as an area that did not experience wildfires for one year and had a population similar to that of the exposed area. Utilizing data from the National Health Insurance Service, analyses were performed for examining increased incidences of ischemic heart disease, pneumonia, asthma, heart failure, and chronic obstructive pulmonary disease.

The number of outpatient and inpatient visits before and after the wildfires was classified by disease and summarized for both exposed and unexposed groups. Chi-square analysis was employed to determine the significance of differences in visit numbers before and after the wildfires. For each outpatient and inpatient visit, a 95% confidence interval (95% CI) was calculated using a Poisson distribution. To assess changes in outpatient and inpatient service utilization in the two years preceding and following the wildfires, the ratio and confidence interval of outpatient and inpatient service utilization, both before and after the wildfires were computed for the overall population in 2017 by region per 100,000 people. Subsequently, these results were examined in consultation with experts to identify conditions with notably high rates of utilization of outpatient or inpatient services in the exposed group for further analysis.

For the purpose of further analysis, a difference-in-differences (DID) analysis was initially conducted to examine the change in annual outpatient and inpatient service utilization rates for the exposed and unexposed groups. DID is a method employed to compare the difference between experimental and control groups before and after a specific event. It evaluates the change trend and difference between the predicted utilization rate of outpatient and inpatient services in the exposed group without the event and the actual utilization rate of outpatient and inpatient services after the event. Simultaneously, it compares the ratio of the utilization rates of outpatient and inpatient services of the unexposed group to that of the exposed group. By comparing the before-and-after relationship between the experimental group and a homogeneous control group, unobservable factors that change over time can be removed, allowing causal effect estimates to be identified. To analyze the causal effect of an event with DID analysis, it is necessary to utilize data from before and after the event for both the control and experimental groups. Respectively, and can be classified into the following four groups. The control group before the event can be categorized as CB and the experimental group as TB, and the control group after the event can be categorized as CA and the experimental group as TA. The formula for estimating causality and DID based on the four groups is as follows:

$$\begin{array}{l}\text{Causality}=T_A-T{\text{'}}_A\\ \text{DID estimation method=}\left(T_A-T_B\right)-\left(C_A-C_B\right)\end{array}$$

Second, an interrupted time series analysis, which is a method for examining changes in trends after a specific event in longitudinally recorded data, was employed. This allowed the identification of alterations in the utilization rates of outpatient and inpatient services for specific diseases following the wildfires in both the exposed and unexposed groups. The analysis accounted for age, gender, long-term trends, and seasonal factors based on monthly trends. Interrupted time series analysis methods are designed to discern the distribution and patterns within time-series data, facilitating the prediction of future trends. Through the use of quantitative forecasting methods, historical data of the variable to be predicted reveals different patterns, allowing the design of a model that assumes that such patterns will repeat in the future. The equation for the interrupted time series analysis employed in this study using the GAM model is as follows:

$$g(Heartfailur{e}_t)={\beta }_0+f_0\left(wildfir{e}_{t-1}\right)+f_2\left(Seaso{n}_t\right)f_3\left(Mont{h}_t\right)+\epsilon (t)$$

All analyses were conducted using R (version 4.0.5). This study adhered to the ethical principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Yonsei University Severance Hospital (IRB 4-2023-0433).

Results

The total population in the exposed group and unexposed group areas was 1,116,485, comprising 564,585 males and 551,900 females. The analysis revealed variations in the number of outpatient and hospitalization visits in the two years before and two years after the wildfires, with differences between the exposed and unexposed groups and by disease. Specifically, the number of outpatient visits in the exposed group significantly increased after the wildfires compared to before the wildfires for all conditions. However, the number of hospitalizations increased significantly only for pneumonia and ischemic heart disease (Table 5).

Table 5. Number of outpatient and hospitalization uses by exposed group before and after wildfire outbreak.

Disease	Population	N of outpatient visits per year before forest fire	N of outpatient visits per year after forest fire	p-value	N of hospitalizations per year before wildfire	N of hospitalizations per year after wildfire	p-value
COPD	168,124	12,126	14,228	<0.001	716	704	<0.001
Heart failure	168,124	8,450	11,756	<0.001	839	818	<0.001
Asthma	168,124	28,048	28,884	<0.001	802	784	<0.001
Pneumonia	168,124	15,992	18,628	<0.001	3,524	3,757	<0.001
Ischemic heart disease	168,124	41,479	46,579	<0.001	2,819	3,450	<0.001

N=number; COPD=chronic obstructive pulmonary disease.

In the unexposed group, the number of outpatient visits in the two years before and after wildfires increased significantly after wildfires in cases of heart failure, asthma, pneumonia, and ischemic heart disease. Additionally, the number of inpatient visits before and after wildfires increased significantly after wildfires in cases of chronic obstructive pulmonary disease, pneumonia, and ischemic heart disease (Table 6).

Table 6. Number of outpatient and hospitalization uses in d groups before and after wildfire outbreak.

Disease	Population	N of outpatient visits per year before forest fire	N of outpatient visits per year after forest fire	p-value	N of hospitalizations per year before wildfire	N of hospitalizations per year after wildfire	p-value
COPD	261,321	14,785	14,650	<0.001	1,069	1,108	<0.001
Heart failure	261,321	10,624	12,118	<0.001	839	818	<0.001
Asthma	261,321	51,301	57,541	<0.001	802	784	<0.001
Pneumonia	261,321	22,589	32,487	<0.001	3,524	3,757	<0.001
Ischemic heart disease	261,321	51,269	57,406	<0.001	2,819	3,450	<0.001

N=number; COPD=chronic obstructive pulmonary disease.

Based on the number of visits before and after the wildfires, the ratio of annual hospitalization utilization before and after the wildfires was analyzed as follows (Figure 1). The ratio of annual hospitalization utilization between the exposed and unexposed groups before and after the wildfires significantly increased in the case of ischemic heart disease (95% CI: 1.26 [1.19–1.33], p-value<0.001; 95% CI: 1.22 [1.16–1.29], p-value<0.001) and pneumonia (95% CI: 1.10 [1.04–1.16], p-value=0.001; 95% CI: 1.07 [1.02–1.12], p-value=0.006). However, the ratios for asthma (95% CI: 1.03 [0.93–1.14], p-value=0.56; 95% CI: 0.98 [0.89–1.08], p-value=0.651) and chronic obstructive pulmonary disease (95% CI: 0.98 [0.89–1.09], p-value=0.76; 95% CI: 1.04 [0.95–1.13], p-value=0.403) were not significant in the exposed and unexposed groups. In contrast, heart failure (95% CI: 1.43 [1.29–1.58], p-value<0.001; 95% CI: 0.97 [0.89–1.07], p-value=0.605) significantly increased in the exposed group but not in the unexposed group.

Figure 1. Ratio of annual hospitalization utilization rates before and after wildfires.

COPD=chronic obstructive pulmonary disease.

The ratio of annual utilization of outpatient services between the exposed and unexposed groups before and after the wildfires was analyzed as follows (Figure 2). Both the exposed and unexposed groups had significantly higher rates of ischemic heart disease (95% CI: 1.17 [1.15–1.20], p-value<0.001; 95% CI: 1.12 [1.11–1.13], p-value<0.001), pneumonia (95% CI: 1.16 [1.14–1.19], p-value<0.001; 95% CI: 1.44 [1.41–1.46], p-value<0.001), asthma (95% CI: 1.03 [1.01–1.05], p-value<0.001; 95% CI: 1.12 [1.11–1.14], p-value<0.001), and heart failure (95% CI: 1.39 [1.35–1.43], p-value<0.001; 95% CI: 1.14 [1.11–1.17], p-value<0.001). However, chronic obstructive pulmonary disease (95% CI: 1.17 [1.15–1.20], p-value<0.001; 95% CI: 0.99 [0.97–1.01], p-value=0.431) significantly increased in the exposed group but not in the unexposed group.

Figure 2. Ratio of annual outpatient utilization rates before and after wildfires.

COPD=chronic obstructive pulmonary disease.

An expert advisory meeting was convened to review the validity of the results for heart failure and to determine whether further analysis was warranted. The review indicated a larger and more significant increase in the exposed group in the above ratio of hospitalization utilization. Drawing upon existing knowledge and expert input, the review team conducted additional DID and interrupted time series analyses. The hospitalization utilization rate was decreasing in both the exposed and unexposed groups before the wildfires. However, after the wildfires, the actual hospitalization utilization rate in the exposed group (annualized hospitalization utilization rate two years after wildfire exposure: 6,992.46/100,000) was 80.05% higher than the rate expected if the wildfires had not occurred (annualized hospitalization utilization rate two years after wildfire exposure: 5,597.76/100,000) (Figure 3). The annual utilization rate of outpatient services for heart failure was increasing in both the exposed and unexposed groups. However, after the wildfires, the actual utilization rate in the exposed group (556.73/100,000) was 68.54% higher than the rate expected if the wildfires had not occurred (annual utilization rate of outpatient services two years post-exposure: 381.56/100,000) (Figure 4).

Figure 3. Changes in heart failure hospitalization rate due to exposure to forest fires.

Figure 4. Changes in heart failure outpatient utilization rate due to exposure to forest fires.

The x-axis of the interrupted time series analysis graph represents the number of months in the two years before and after the event, while the y-axis indicates the annual utilization rate of inpatient and outpatient services per 100,000 people. The vertical red dashed line on the graph signifies when the wildfires occurred, and the solid blue line represents the expected utilization rate of outpatient and inpatient services if the wildfires had not occurred. The blue dashed line illustrates the actual utilization of inpatient and outpatient services after the wildfires, and the red dotted line is a smoothing graph utilized to determine patterns and trends in the time-series data. The change in actual hospitalization utilization for heart failure in the exposed group (pre-interruption slope; estimate, 0.252; p-value<0.001) was increasing before the wildfires. After the wildfires, the change due to the event (level change; estimate, −0.044; p-value=0.986) decreased slightly but was not significant. Hospitalization utilization, reflecting the subacute effect of the event, also showed a downward trend (post-interruption slope; estimate, −0.183; p-value=0.127) but was not significant (Figure 5, Supplementary Table 2).

Figure 5. Changes in hospitalization utilization for heart failure considering monthly trends (2 years).

The interrupted time series analysis revealed an increasing trend in the actual outpatient service utilization rate for heart failure in the exposed group prior to the wildfires (pre-interruption slope; estimate, 0.649; p-value<0.001). Following the wildfires, there was a marked increase in this trend attributable to the event (level change; estimate, 52.525; p-value<0.001), which was then followed by a declining trend in outpatient service utilization, reflecting the subacute effects of the event (post-interruption slope; estimate, −0.866; p-value=0.029) (Figure 6, Supplementary Table 3).

Figure 6. Changes in heart failure outpatient utilization considering monthly trends (2 years).

Discussion

This study investigated the association between wildfires and subacute illnesses in residents of wildfire-exposed and non-exposed areas during the 2017 wildfires in the ROK. From a short-term planning perspective, the findings can help promptly identify health impacts resulting from natural disasters, such as wildfires, and explore strategies for minimizing unnecessary analyses. Respiratory diseases, including heart failure and pneumonia, were selected on the basis of a literature review due to their potential to influence a broad spectrum of illnesses and the likelihood that they would have a significant impact on health outcomes resulting from wildfires [16,17]. Initially, this study sought to determine key changes and outcomes, such as changes in numbers and rates, using relatively expedited analyses. The ratio of utilization of outpatient services before and after the wildfires in the exposed group increased most significantly in cases of heart failure, and hospitalization utilization in ischemic heart disease showed the greatest increase. In the unexposed group, the ratio of utilization of other outpatient services before and after wildfires saw a significant increase in cases of pneumonia, and the ratio of hospitalization utilization saw a significant increase in cases of ischemic heart disease. Subsequently, both a literature review and expert consultation were conducted to identify significant and plausible findings regarding the increased utilization of the aforementioned services due to exposure to wildfires. This helped determine whether further advanced analyses were necessary and which conditions should be included in the advanced analyses. The effects of air pollution from wildfires on heart failure occur regardless of the duration of exposure [18]. DeFlorio-Barker et al. [19] found that a 10 μg/m³ increase in PM2.5 was associated with a 0.61% increase in cardiovascular hospitalizations during the wildfire season and a 0.69% increase in cardiovascular hospitalizations during the rest of the year. To refine the analysis for heart failure, we considered previous research that suggested that wildfires are associated with much higher increases in PM2.5 [19] and, thus, higher overall cardiovascular disease. Advanced analyses using DID analysis and interrupted time series analysis were conducted to compare the utilization rates of outpatient and inpatient services between the exposed and unexposed groups, revealing a significant increase in the rate of utilization of outpatient services in both analyses.

This study has sought to establish the objectivity and validity of researchers’ opinions through expert advisory committees in the selection of exposure areas (disaster-affected areas) using retrospective data, the selection of related diseases, and the derivation of meaningful results from the analysis. In recognition of the limitations of short-term analyses, particularly unmeasured confounders, two rounds of expert review were conducted using the RAND/UCLA appropriateness method (RAM). RAM is employed to determine the appropriate use of technology and it combines a critical review of the literature with input from clinical experts [20]. In health services research, RAM is referred to as a “modified Delphi method” because it adds a discussion round [21]. This method integrates a critical review of the literature and the opinions of clinical experts to determine the appropriate use of technology, thereby helping to overcome the limitations of a short-term analysis. It also underscores the importance of developing a systematic monitoring system to ensure the continuity of the study and address the limitations of the analysis by obtaining mid- to long-term data. This approach is crucial for responding to long-term and significant health risks, such as air pollution from wildfires. Governments and relevant organizations can use this information to raise awareness of the health impacts of natural disasters, including wildfires, and develop effective prevention and response strategies.

Hence, this study proposes a monitoring system for natural disasters, including wildfires. This system can be broadly categorized into the pre-cognition and preparation phase and the response phase. In the pre-cognition and preparation stage, laying the foundation for securing data in advance through cooperation with various ministries is crucial for the response and monitoring system after a natural disaster occurs. If we take wildfires as an example, collaboration with the Korea Forest Service, the Ministry of Environment, local governments, the Ministry of Health and Welfare, and the Ministry of the Interior and Safety can gather appropriate data upon exposure to wildfires as well as data from medical institutions and community health centers. Preliminary retrospective data will be used to inventory and pilot analytical items, identify areas vulnerable to wildfires, and assess potential health impact risks. The results can be employed to establish the areas vulnerable to wildfires in stages and pre-define the prospective field survey collection items for both exposed vulnerable areas and non-exposed (control) areas. Ultimately, protocols should be developed for field surveys and targeted population follow-up in wildfire-vulnerable and control areas. This will enable customized monitoring and effective on-the-ground response strategies. The process outlined in this study can serve as an example of the pre-awareness and preparedness phases of a monitoring system.

In the response phase, the medium-term focus is on rapid data collection through field surveys immediately after the event and response through advanced analytics. Protocols established in the pre-cognition and preparation phases for field surveys and follow-ups are utilized to collect and analyze basic information and the medical records of exposed and non-exposed residents, along with environmental measurements from the exposure site. The analysis helps identify individuals at high risk for health effects, enabling a close follow-up and support system for high-risk individuals. Refinements that emerge from this process can be discussed through ongoing Delphi-style expert advisory meetings to create more sophisticated monitoring systems and follow-up protocols. Ongoing surveillance allows for tracking long-term health impacts and advocating for legislation relevant to residents of exposed areas. Such legislation could encompass medical assistance for high-risk populations facing significant health impacts, a legal foundation for environmental and health monitoring systems, and policies promoting community health and well-being. All these steps contribute to a monitoring system focused on protecting the health and safety of people after a natural disaster and strengthening response capabilities in the long term.

This study has several limitations:

1) Statistical power and reliability could be enhanced with data collected over a more extended period. However, due to data limitations, the study focused on two years before and after the event to correct for seasonal variations. To assess the occurrence of diseases based on various exposure factors, additional studies using long-term data are required to investigate monthly utilization rates of outpatient and inpatient services by disease over four years (48 months) for both the exposed and control groups.

2) Data limitations led to the use of age, gender, and farm population distribution as regional matching variables. Further research is needed to correct and match social variables such as NCD indicators, including smoking and alcohol consumption rates, which show regional variations.

3) Aside from heart failure, diseases did not show significant changes before and after the wildfires, suggesting that disease incidence may have been influenced by other disasters occurring during the study period. Further research is required to explore the impact of other disasters on disease incidence.

4) Due to data limitations in this study, additional research is required to investigate monthly utilization rates of outpatient and inpatient services by disease for four years (48 months) in both the exposed and control groups. This will enable a more thorough examination of disease occurrences based on various exposure factors.

In conclusion, it is expected that the establishment of a comprehensive and systematic monitoring system as a result of this study will contribute to a deeper understanding of the exposure and health effects of natural disasters, fostering the development of effective responses. The systematic approach presented in this study will serve as a crucial foundation for the development of health impact monitoring systems for wildfires and other natural hazards.

Supplementary Materials

Supplementary data are available online.