Joint Effects of Heat and Fine Particulate Matter on Out‐of‐Hospital Cardiac Arrest in Japan: A Nationwide Time‐Stratified Case‐Crossover Study

Abstract

Background

Out‐of‐hospital cardiac arrest (OHCA) is a leading cause of cardiovascular death. Although ambient temperature and particulate matter with an aerodynamic diameter ≤2.5 μm (PM_2.5) are established risk factors, evidence on their joint effects remains limited. We assessed the joint effects of heat and PM_2.5 on OHCA risk using nationwide data from Japan.

Methods

We collected daily data on cardiac‐origin OHCAs from the All‐Japan Utstein Registry and daily temperature and PM_2.5 from monitoring networks across 47 prefectures during 2012 through 2019. Using a 2‐stage time‐stratified case‐crossover design, we estimated prefecture‐specific associations with conditional quasi‐Poisson models and pooled them using meta‐regression. Interactions between temperature and PM_2.5 were assessed on multiplicative and additive scales using the relative excess risk due to interaction. Attributable cases were also estimated.

Results

A total of 610 613 cardiac‐origin OHCAs were included. Heat (relative risk, 1.04 [95% CI, 1.02–1.06]; 99th versus 80th percentile); and PM_2.5 (relative risk, 1.02 per 10 μg/m³ [95% CI, 1.01–1.03]) were independently associated with increased OHCA risk. Heat effects were amplified at higher PM_2.5 concentrations, with each 10 μg/m³ increase in PM_2.5 associated with an additional 1.8% increase in heat‐related risk. Conversely, PM_2.5 effects increased by 1.2% (95% CI, 1.1%–1.2%) per 25th percentile increase in temperature. Interactions were observed on both scales (relative excess risk due to interaction, 0.01 [95% CI, 0.00–0.02]; P=0.003). Overall, 2638 OHCA cases were attributable to coexposure, exceeding the combined burden attributable to heat (1215) and PM_2.5 (1276) separately.

Conclusions

Heat and PM_2.5 acted synergistically to increase OHCA risk, highlighting the need for integrated public health strategies addressing both extreme heat and air pollution under climate change.

Nonstandard Abbreviations and Acronyms

JCS‐ReSS

Japanese Circulation Society–Resuscitation Science Study

OHCA

out‐of‐hospital cardiac arrest

PM_2.5

particulate matter with an aerodynamic diameter ≤2.5 μm

RERI

relative excess risk due to interaction

Research Perspective.

What Is New?

• This nationwide study from Japan is the first to assess the joint effects of high ambient temperature and particulate matter with an aerodynamic diameter ≤2.5 μm on out‐of‐hospital cardiac arrest across both additive and multiplicative scales. The findings demonstrate a synergistic interaction between heat and particulate matter with an aerodynamic diameter ≤2.5 μm, indicating that combined exposure confers greater cardiovascular risk than either factor alone, particularly among older adults. By quantifying the excess out‐of‐hospital cardiac arrest risk attributable to joint exposure, this study provides clinically relevant evidence on the compounded impact of environmental stressors.

What Question Should Be Addressed Next?

• Future research should evaluate how integrated heat–air pollution warning systems and targeted interventions for vulnerable populations can reduce the incidence of acute cardiac events.

Out‐of‐hospital cardiac arrest (OHCA) is defined as the loss of functional cardiac mechanical activity in association with an absence of systemic circulation, occurring outside a hospital setting. ¹ OHCA is a leading cause of cardiovascular death, with an estimated annual number of 3.8 million and an 8% to 12% survival to hospital discharge rates. ² Despite the substantial public health burden, remarkable strides have been made in the prevention of OHCA over the past 3 decades. ¹ , ³ This progress primarily stemmed from addressing traditional risk factors such as tobacco use, diet, and physical activity. Recently, public health assessments have increasingly focused on nontraditional risk factors, such as environmental exposures, which may provide new insights into the prevention of cardiovascular morbidity. ⁴

Air pollution is a well‐known risk factor for cardiovascular morbidity and death. ⁵ A systematic review ⁶ reported that short‐term exposures to air pollutants were significantly associated with an increase in OHCA risk, with the strongest association being observed for particulate matter with an aerodynamic diameter ≤2.5 μm (PM_2.5). These fine particles can penetrate deep into the lungs, enter the bloodstream, and travel to organs, causing systemic damage to tissues and cells, thereby increasing the risk of cardiovascular diseases. ⁷ , ⁸ In Japan, a study covering 100 Japanese cities showed that daily PM_2.5 could increase the risk of cardiovascular death despite the low average concentration (14.6 μg/m³). ⁹

Ambient temperature is another well‐documented risk factor for cardiovascular health, in addition to PM_2.5. In regions with distinct warm and cold seasons, both high and low temperatures have been linked to an increased risk of cardiovascular morbidity and death, ¹⁰ , ¹¹ indicating that nonoptimal temperatures can trigger acute cardiovascular events. With the ongoing rise in global mean temperature and the projected increase in frequency, intensity, and duration of extreme temperature events, the burden of temperature on cardiovascular health is expected to escalate in the future. ¹² However, despite the urgency, many existing studies have focused on estimating the independent effect of ambient temperature, neglecting to explore how this association may be modified by PM_2.5, and vice versa.

Past studies have observed a positive correlation between ambient temperature and PM_2.5, particularly during the warm season. In Japan, PM_2.5 is positively correlated with temperature in major cities. ¹³ Some evidence further suggests that PM_2.5 may positively modify the short‐term association between heat and death in the urban areas of Japan. ¹⁴ However, current evidence from epidemiological studies does not provide clear insight into whether PM_2.5 and heat interact to inflict additional harm on individuals with compromised cardiovascular functions. A recent review reported inconsistent evidence on interaction between heat and PM_2.5, and its meta‐analysis did not identify a significant modifying effect of PM_2.5 on heat‐related death. ¹⁵ Furthermore, previous studies examining this interaction have primarily focused on death as the health outcome, with very few investigating morbidity. ¹⁶ Existing interaction studies were often conducted in a single city or small region and used different methodologies, making it challenging to consolidate evidence. ¹⁷

Given the uncertainty surrounding the combined effects of temperature and PM_2.5 on cardiovascular health, we conducted a study to assess the joint effects of heat and PM_2.5 on OHCA risk in Japan, on additive and multiplicative scales, including subgroup analyses and evaluation of heterogeneity among the prefectures.

METHODS

Data Collection

Because of the sensitive nature of the OHCA data used in this study, access to the data set may be requested by qualified researchers through the JCS‐ReSS (Japanese Circulation Society–Resuscitation Science Study) Group (https://www.j‐circ.or.jp/english/contact/). The cleaned and aggregated PM_2.5 and meteorological data that support the findings of this study are available from the corresponding author upon reasonable request.

The study included OHCA cases recorded between January 1, 2012, and December 31, 2019, in the All‐Japan Utstein Registry, a nationwide, population‐based prospective database that uses standardized Utstein‐style data collection and is hosted on the Fire and Disaster Management Agency database server. ¹⁸ , ¹⁹ As of 2023, Japan has a population of 124.5 million, served by 726 fire stations with dispatch centers. ²⁰ These centers manage emergency medical services systems that are nearly uniform across the country's 47 prefectures. ²¹ Emergency medical services are available 24 hours a day through the toll‐free 119 emergency number, which directly connects to dispatch centers at regional fire defense headquarters. ²² Upon receiving a report of an OHCA, the nearest available ambulance is dispatched to the scene promptly. ²³ Emergency medical services providers in Japan are not authorized to terminate resuscitation in the field, meaning that most patients with OHCA treated by emergency medical services personnel are transported to the hospital, except in cases of decapitation, incineration, decomposition, rigor mortis, or dependent cyanosis. ²⁴ Physicians at each participating hospital collect in‐hospital data, including information on the cause of arrest, treatments, and outcomes. ²⁴ For this analysis, only OHCA cases of cardiac origin were included, as they are generally more predictable, potentially preventable, and subject to targeted interventions. ²⁵ , ²⁶ Cases of noncardiac origin, such as those resulting from trauma, traffic accidents, drowning, or hypothermia, were excluded from the statistical analysis. All data used in this study were fully and irreversibly anonymized. These data were accessed through the JCS‐ReSS Group. Accordingly, the study was exempt from review by the Institutional Review Board of Hokkaido University Hospital for Clinical Research. We followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline.

Hourly mean temperatures and relative humidity, measured by representative single weather monitoring stations in each prefecture, were acquired each day from the Japan Meteorological Agency and aggregated into daily data. A single weather station situated in the population centers of each prefecture was chosen as a representative of the region. These stations, being synoptic climatological stations, are intended to capture prefecture‐wide weather conditions. The temperature was standardized as prefecture‐specific percentiles.

We acquired the PM_2.5 data from the atmospheric environment database of the National Institute for Environmental Studies in Japan, where hourly measurements of PM_2.5 concentrations were retrieved from automated equipment and standard reference methods. ⁹ We selected the monitoring stations located in the prefectural capital city with the longest length of data period and a high level of correlation (>0.9) within the same prefecture in the case of multiple stations. Extracted hourly measurements were aggregated by calendar date into a city‐specific daily average of PM_2.5 concentrations. Missing PM_2.5 data were replaced by values from a centered 7‐day moving average. If the interval of missing data was >3 days, the missing data were not replaced. Tables S1 through S5 provide a summary of the available data period for PM_2.5.

Statistical Analysis

The study adopted a 2‐stage approach. First, a time‐stratified case‐crossover study design with a conditional quasi‐Poisson model was applied to estimate the nonlinear temperature–OHCA risk and linear PM_2.5–OHCA risk, respectively, in each prefecture. PM_2.5 was modeled as lag 0 to 3 (a 4‐day moving average) based on exploratory results of various single lags and average lag periods (Figure S1). Temperature was modeled using lag 0 to 3 as a natural cubic spline with 3 df. In our time‐stratified case‐crossover study design, we defined strata by grouping days that had the same day‐of‐week within the same calendar month and year into strata. For example, all Mondays in January 2012 were placed in the same stratum. The ambient exposure of each patient on the case day was then compared with the exposures on the control days within the same stratum and this approach eliminated the need to explicitly adjust for long‐term and seasonal time trends. As patients were compared with themselves at different time points, time‐invariant factors, such as age and sex, as well as slowly varying characteristics (eg, smoking habits, socioeconomic status, comorbidities) do not confound the results. When exploring the effects of temperature, PM_2.5 was not adjusted in the model while PM_2.5 effects were estimated with adjustment for temperature. ²⁷ In each prefecture, we estimated heat effect as the relative risk (RR), defined as the risk at the 99th versus the risk at 80th temperature percentile, and the corresponding 95% CIs. The RR of PM_2.5 effect was estimated for per‐10 μg/m³ increase.

To investigate the interaction between temperature and PM_2.5 on OHCA risk in each prefecture, we introduced an interaction term between the natural spline of air temperature and PM_2.5 to the basic confounder model. From this model, we estimated the RRs of heat centering at different levels of PM_2.5 (1 μg/m³ to 40 μg/m³). The range of PM_2.5 was chosen on the basis of the observed prefecture‐specific PM_2.5 distributions (Figure S2). From the same model, we also estimated the RRs of PM_2.5 with temperature centered at different levels (1st to 99th percentile).

In the second stage, we pooled the prefecture‐specific associations between temperature, PM_2.5 concentrations and OHCA risk using a random‐effects meta‐analysis. The estimated PM_2.5‐specific heat effects and temperature‐specific PM_2.5 effects were also pooled using the same approach. Since we modeled the interaction using a product term in a log‐linear model, which implies the interaction on a multiplicative scale, we also pooled the prefecture‐specific estimates of the product terms between temperature and PM_2.5 and examined the significance of the coefficients using Wald test. ¹⁷ To explore the potential regional differences, we conducted additional analysis on the northern prefecture (Hokkaido) and the southern prefectures (Okinawa and Kyushu areas including Fukuoka, Saga, Nagasaki, Kumamoto, Oita, Miyazaki, and Kagoshima).

Besides the interaction on a multiplicative scale, we assessed the interaction on an additive scale by deriving the relative excess risk due to interaction (RERI) of heat and PM_2.5, calculated as RR₁₁−RR₀₁−RR₁₀+1. ²⁸ We define RR₁₁ as the PM_2.5 effect when temperature is centered at the 99th percentile and RR₁₀ when temperature is centered at the 80th percentile. RR₀₁ refers to the heat effect when PM_2.5 concentration is centered at 5 μg/m³. If RERI is >0, it denotes a positive interaction on an additive scale between heat and PM_2.5. To obtain the 95% CI of RERI, RERI was estimated 1000 times by bootstrap replications (with replacement) of the original sample. ²⁹ , ³⁰

To illustrate the health impact of heat and PM_2.5 in Japan, we first calculated the number of OHCA cases attributable to heat (>80th temperature percentile) and PM_2.5 (>5 μg/m³) using the prefecture‐level time series data, by summing the attributable cases using the attributable fraction (RR‐1)/RR. For joint effects, we calculated the attributable number due to coexposure of heat and PM_2.5 on each heat day using attributable fraction ((RR‐1)/RR)×daily OHCA cases, with the RR for PM_2.5 centered at specific temperature percentile in that day. When calculating the attributable number for PM_2.5, we set a baseline concentration of 5 μg/m³, with RRs estimated on the basis of daily PM_2.5 concentrations (lag 0–3) above the baseline. This decision was based on (1) findings from previous studies ⁸ , ³¹ and (2) our exploratory analysis, in which the nonlinear modeling of PM_2.5 revealed a J‐shaped exposure–response curve, with RR starting to increase at ≈5 μg/m³ (Figure S3).

We conducted several sensitivity analyses to assess the robustness of our primary findings across various modeling choices and definitions. First, we restricted our study months to the 6 warmest consecutive months (May to October). Second, we tested whether the results are sensitive to relative humidity, by adding a natural cubic smooth function of daily mean relative humidity with 3 df to the interaction model. Third, we shortened the moving average of PM_2.5 and heat lags to lag 0 to 1. Additionally, we restricted our analysis to bystander‐witnessed OHCA cases of cardiac origin, as the time of occurrence in these cases was likely more accurately recorded. Finally, we explored whether the effects of cold temperature were modified by PM_2.5 levels.

All analyses were performed using R software version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria), with the packages splines, tsModel, dlnm, and mixmeta.

RESULTS

From January 2012 to December 2019, 1 006 923 OHCA cases were registered in the database. Of these, 610 613 cases were identified as being of cardiac origin (Figure S4). Table 1 summarizes descriptive statistics on OHCA number of cases, air temperature, and PM_2.5 concentrations in Japan, while prefecture‐specific results are in Table S2. The daily mean number of OHCA cases of cardiac origin was 209, with prefecture‐level daily means ranging from 1 (multiple prefectures) to 22 (Tokyo), and the time series of daily OHCA cases in Japan is shown in Figure S5. Japan experiences considerable fluctuations in daily mean temperatures owing to its geographic span and differences in altitude. Most prefectures exhibit a wide range of temperature distributions (Figures S6 and S7). Generally, PM_2.5 levels in Japan are relatively low, within the environmental quality standards (daily average concentrations ≤35 μg/m³). ³² However, air quality varies across prefectures (Figure S2).

Table 1.

Descriptive Statistics of OHCA Cases, Temperature, and PM_2.5 in Japan

Variable	Subgroup	Total	Daily mean	SD	1st	50th	99th
OHCA		610 613	209	57	128	194	364
	Men	347 766	119	33	69	112	209
	Women	262 847	90	26	51	84	161
	Age ≤74 y	209 111	71	17	41	69	120
	Age >74 y	401 502	137	42	77	127	253
Temperature (°C)	Full year		15.7	8.7	−1.9	16.5	30.6
	6 warmest month		22.8	4.6	11.5	23.0	31.0
PM2.5 (μg/m3)	Full year		12.9	7.7	1.7	11.2	38.1

OHCA indicates out‐of‐hospital cardiac arrest; and PM_2.5, particulate matter with an aerodynamic diameter ≤2.5 μm.

The pooled estimates of the heat effects and PM_2.5 effects on OHCA risk in the total population and subpopulations across 47 Japanese prefectures are detailed in Table 2. Overall, both the heat effect (RR, 1.09 [95% CI, 1.01–1.17] per temperature increments between 80th and 99th percentile) and the PM_2.5 effect (RR, 1.02 [95% CI, 1.01–1.03] per 10 μg/m³ increment) in the total population exhibited significant adverse impacts on OHCA risk. The heat effect was more pronounced at higher PM_2.5 concentrations, with a stronger impact observed at 40 μg/m³ PM_2.5 (RR, 1.07 [95% CI, 1.01–1.14], comparing the risks at the 99th versus the 80th percentile of temperature) compared with 1 μg/m³ (RR, 1.00 [95% CI, 0.97–1.03]). Nonlinear temperature–RR curves at varying PM_2.5 concentrations are shown in Figure S8. Similarly, the PM_2.5 effect was amplified with increasing temperature, with PM‐related effects rising from the 1st to the 99th percentile of temperature across all populations and subpopulations.

Table 2.

Pooled Heat Effect (99th Versus 80th Temperature Percentile) and PM_2.5 Effect (Every 10 μg/m³ PM_2.5 Increase) From 47 Japanese Prefectures

Exposure	RR (95% CI)
	All population	Men	Women	Age ≤74 y	Age >74 y
Heat effect (99th vs 80th temperature percentile)	1.04 (1.02–1.06)	1.04 (1.02–1.07)	1.02 (1.00–1.05)	1.04 (1.01–1.07)	1.03 (1.01–1.05)
Modified by PM2.5
At 1 μg/m3	1.00 (0.97–1.03)	1.01 (0.97–1.06)	1.00 (0.96–1.05)	1.00 (0.96–1.04)	1.01 (0.97–1.05)
At 40 μg/m3	1.07 (1.01–1.14)	1.09 (1.02–1.16)	1.08 (1.00–1.16)	1.11 (1.04–1.19)	1.04 (0.97–1.10)
PM2.5 effect (every 10 μg/m3 increase)	1.02 (1.01–1.03)	1.02 (1.01–1.03)	1.01 (1.00–1.02)	1.04 (1.01–1.07)	1.03 (1.01–1.05)
Modified by temperature
At 1st percentile	0.99 (0.98–1.01)	1.00 (0.98–1.02)	0.99 (0.97–1.01)	0.99 (0.97–1.01)	0.99 (0.98–1.01)
At 80th percentile	1.03 (1.02–1.04)	1.03 (1.01–1.04)	1.01 (1.00–1.02)	1.02 (1.01–1.03)	1.03 (1.02–1.04)
At 99th percentile	1.04 (1.02–1.05)	1.04 (1.02–1.05)	1.01 (0.99–1.03)	1.02 (1.00–1.04)	1.04 (1.02–1.05)

PM_2.5 indicates particulate matter with an aerodynamic diameter ≤2.5 μm; and RR, relative risk.

Figure 1 shows the pooled heat effects stratified by PM_2.5 concentration (1–40 μg/m³) in all population (A), men and women (B), and aged ≤74 years and >74 years (C). The heat effect became more pronounced as PM_2.5 concentrations increased, and this increment was nearly linear. Assuming a linear association, each 10 μg/m³ increase in PM_2.5 concentrations was associated with an additional 1.8% (95% CI, 1.7%–1.9%) increase in the heat effect. Subpopulation analyses revealed no significant sex differences, but the enhancement of heat effects with rising PM_2.5 was more prominent among younger individuals (aged ≤74 years). Regionally, results were robust after excluding Hokkaido. However, excluding Kyushu and Okinawa strengthened the effect modification, which then reached statistical significance (Figure S9). In Kyushu and Okinawa, antagonistic modification trends were observed though they were not statistically significant.

Figure 1. Pooled PM_2.5 concentration‐specific heat effects, expressed as the RR comparing the 99th with the 80th percentile of temperature, across 47 Japanese prefectures.

Estimates are shown for the observed PM_2.5 levels ranging from 1 to 40 μg/m³ for (A) all population, (B) men and women, and (C) individuals aged ≤74 years and >74 years. PM_2.5 indicates particulate matter with an aerodynamic diameter ≤2.5 μm; and RR, relative risk.

Figure 2 presents the pooled PM_2.5 effects stratified by temperature percentile (1st–99th) in all population (A), men and women (B), and age ≤74 years and >74 years. The PM_2.5 effect increased almost linearly with rising temperature across the entire range. Assuming a linear association, every 25‐percentile increase in temperature intensified the PM_2.5 effect, resulting in a 1.2% (95% CI, 1.1%–1.2%) increase in RR. Temperature amplified PM_2.5 effects in both sexes, but the amplification was stronger in older individuals (aged >74 years). Temperature modification of the PM_2.5–OHCA association was more evident in northern and central prefectures but not significant in southern Japan (Kyushu and Okinawa; Figure S10).

Figure 2. PM_2.5 effect (RR for every 10 μg/m³ increase in PM_2.5 concentrations) by different temperature percentiles from 47 Japanese prefectures.

A, All population, (B) men and women, and (C) ≤74 years and >74 years. PM_2.5 indicates particulate matter with an aerodynamic diameter ≤2.5 μm; and RR, relative risk.

Table 3 summarizes the interaction analyses. A positive RERI indicated additive interaction between temperature and PM_2.5, while multiplicative interaction was supported by small P values for the interaction term. Full details of the RERI calculations are provided in Table S3. Overall, strong evidence of interaction was found on both additive and multiplicative scales in total population, as well as women and younger individuals (aged ≤74 years).

Table 3.

RERI Index (Interaction on an Additive Scale) and P Values of the Meta‐Analytical Estimates of the Product Terms Between Temperature and PM_2.5 (Interaction on a Multiplicative Scale)

Population	Sensitivity analysis	Additive interaction,* RERI (95% CI)	Multiplicative interaction, P
All population		0.01 (0.00 to 0.02)†	0.003‡
	Adjusted relative humidity	0.01 (0.00 to 0.02)†	<0.001‡
	Lag 0–1	0.00 (−0.01 to 0.01)	<0.001‡
	6 warmest month		0.91
Men		−0.01 (−0.02 to 0.00)	0.02‡
Women		0.01 (0.00 to 0.02)†	0.05
Age ≤74 y		0.02 (0.01 to 0.03)†	<0.001‡
Age >74 y		0.00 (−0.01,0.01)	<0.001‡

PM_2.5 indicates particulate matter with an aerodynamic diameter ≤2.5 μm; and RERI, relative excess risk due to interaction.

^*The details of calculating RERI in each category are demonstrated in Table S3. P values for interactions were obtained from Wald test.

^† 95% CIs>0.

^‡ P value for the interaction term <0.05.

Attributable numbers, attributable fractions, and exposure days for heat and PM_2.5 are summarized in Table S4. A total of 2638 OHCA cases were attributed to coexposure to heat (>80th percentile) and PM_2.5, which exceeded the sum of cases from individual exposures (1276 for heat and 1225 for PM_2.5, totaling 2501).

Sensitivity analyses indicated that the main findings were robust to adjustment for relative humidity and modifications of the lag period (lag 0–1; Table 3; Table S5; Figures S11 and S12). However, restricting the study to the 6 consecutive warmest months weakened the joint effects of temperature and PM_2.5. When the analysis was limited to bystander‐witnessed OHCAs of cardiac origin, similar trends in effect modification were observed, although CIs were wider, likely due to reduced sample size (Figure S13). In contrast, although an opposite pattern was observed for cold exposure, the overlapping confidence intervals suggest the change in risk was not substantially different (Figure S14).

DISCUSSION

To the best of our knowledge, this is the first epidemiological study in Japan to evaluate the interaction between high ambient temperature and PM_2.5 on OHCA risk, using both multiplicative and additive scales. Our findings suggest that PM_2.5 enhances the adverse effects of heat on OHCA risk. Moreover, heat significantly amplifies the effects of PM_2.5 on OHCA. The presence of interaction was evident on both scales.

Our findings on the independent short‐term effects of temperature and PM_2.5 are consistent with earlier studies conducted in Japan. ³³ , ³⁴ , ³⁵ Previous reviews have documented the short‐term impacts of heat ³⁶ and PM_2.5 ³⁷ on OHCA, though with regional variability. For temperature, prior studies often reported V‐, U‐, or J‐shaped exposure–response relationships. While studies in Asia have shown elevated OHCA risk during high temperatures, ³³ , ³⁸ others from Sweden and Australia reported no such effects. ³⁹ , ⁴⁰ Regarding PM_2.5, existing evidence supports a positive association with OHCA risk, though lag structures vary. For example, Zhao et al ³⁴ focused on cardiac‐origin OHCA at lag 0 to 3, while Kojima et al ³⁵ examined bystander‐witnessed OHCA at lag 0 to 1. These inconsistencies, coupled with limited geographic coverage, highlight the need for broader, regionally diverse investigations.

The joint impact of temperature and PM_2.5 exposure is of increasing public health importance, particularly in the context of climate change, which is expected to increase the frequency of extreme heat events alongside persistent air pollution. Our findings show that high PM_2.5 levels significantly exacerbate heat effects on OHCA, consistent with previous reports on cardiovascular death. ¹⁷ , ⁴¹ In contrast, Iranpour et al ⁴² found no such modification in mortality rate, underscoring the limited and inconsistent evidence, especially on acute outcomes such as OHCA.

Some studies suggest PM_2.5 effects are most pronounced during periods of high temperature or in summer months, ¹⁷ , ⁴¹ , ⁴³ , ⁴⁴ although results remain inconsistent. For instance, Li et al ⁴⁵ reported that extremely low temperatures amplified PM_2.5‐related death, while Yitshak‐Sade et al ⁴⁶ observed that higher temperatures strengthened PM_2.5's association with respiratory hospital admissions but attenuated its effect on cardiac admissions. Such inconsistencies may reflect differences in health outcomes, study regions, and exposure metrics. ¹⁵ Given the limited evidence on morbidity and regional variability, further research is needed to validate these findings across diverse settings.

In our study, synergistic effects of temperature and PM_2.5 on OHCA risk were observed on both multiplicative and additive scales. While many prior studies focused on multiplicative interactions, only a few such as the multicountry study in Northeast Asia ¹⁶ examined additive interactions. Additive interaction is particularly important for public health, as it provides estimates of excess risk attributable to joint exposure. In contrast with studies reporting nonlinear interaction patterns, ¹⁶ , ¹⁷ our results showed nearly linear increases in PM_2.5‐related risk across temperature percentiles, reinforcing the presence of synergy.

Subgroup analyses revealed that the 2‐way effect modifications were generally consistent across sexes and age groups, although older adults (aged >74 years) showed stronger modification of PM_2.5 effects by temperature, suggesting increased susceptibility. This finding aligns with 1 earlier mortality study, ¹⁶ but another study reported that PM_2.5 effect on mortality rate was lower during warm days. ⁴³ We observed stronger PM_2.5‐related health effects in men than in women, whereas the effects of high temperature showed little variation by sex. Despite this difference in PM_2.5 susceptibility, interactions between heat and PM_2.5 were evident in both men and women. These findings suggest that while baseline vulnerability to PM_2.5 exposure may differ by sex, the synergistic impacts of concurrent heat and PM_2.5 exposure operate similarly across sexes. Wald test results for interaction terms further supported significant interaction across most strata. Regional analyses indicated that excluding southern prefectures strengthened the observed effect modifications, with more prominent interactions in northern and central regions, possibly due to climatic or behavioral differences. These findings also highlight potential environmental justice concerns, as vulnerable groups such as older adults may face greater risks under climate change.

Mechanistically, the joint effects of heat and PM_2.5 may be explained by overlapping physiological pathways. Heat stress can impair thermoregulation, increase inflammatory responses, and promote blood coagulation. ¹⁶ , ⁴⁷ PM_2.5 exposure can trigger systemic inflammation, autonomic nervous system disruption, and thrombus formation. ⁴⁸ , ⁴⁹ Behavioral factors may also play a role, as some individuals are more likely to spend time outdoors or keep windows open during warm seasons, increasing exposure to PM_2.5. ⁴¹ Regional heterogeneity in behavioral patterns, especially in climates with distinct seasonal variation, may further explain the geographic variability observed.

From a methodological perspective, assessing interaction on the additive scale is critical, as it better informs public health interventions than multiplicative interaction alone. ⁵⁰ Additive measures quantify excess risk from coexposure, which is vital for identifying high‐risk subpopulations. However, additive interaction estimates require careful interpretation, especially for continuous variables like temperature and PM_2.5, due to nonlinear exposure–response relationships and dependence on the reference category. ²⁸ , ⁵¹ In this study, both RR₁₀ and RR₀₁ exceeded 1, reflecting the known risks of heat and PM_2.5, thereby supporting the interpretability of the additive interaction (RERI). While multiplicative interaction is easier to assess statistically (eg, via likelihood ratio tests or Wald tests), our findings on both scales collectively justify the need for joint preventive strategies. Additionally, we estimated the number of OHCA cases attributable to coexposure, which exceeded the sum of cases from individual exposures. This supports the presence of synergy and highlights its public health relevance. Although the excess attributable number due to the synergistic effect was relatively small because PM_2.5 concentrations were not very high during the Japanese summer, the health burden associated with the joint effects of temperature and PM_2.5 should not be overlooked, especially considering the potential rise in temperature due to climate change.

The study has several strengths. First, to our knowledge, it is the first to evaluate the combined effects of temperature and PM_2.5 on OHCA using a large national population data set in Japan. Second, by leveraging Utstein data encompassing all 47 Japanese prefectures and including age and sex information, we were able to compare prefecture‐specific results and identify differences by demographic subgroups. Third, in addition to examining the 2‐way effect modifications, we assessed interaction on both multiplicative and additive scales and estimated attributable numbers due to coexposure, providing robust evidence of synergy.

Several limitations should also be noted. First, OHCA data were recorded at the prefecture level without precise city locations. Consequently, PM_2.5 exposure was based on monitoring stations in prefectural capitals, which may not fully capture intraprefecture variation. However, based on analyses conducted in this study, high correlations of PM_2.5 concentrations between stations within prefectures, suggesting minimal impact. Second, while our approach to assessing interactions has been widely used, results inevitably depend on temperature distributions and definitions of heat effects, and should thus be interpreted cautiously. Beyond the interaction results, stratified analyses of heat effects by PM_2.5 concentration and PM_2.5 effects by temperature percentiles provide additional support for our conclusions. Third, generalizability may be limited. Previous studies have demonstrated variability in temperature–PM_2.5 interactions across regions, warranting further investigations in other settings. Finally, our study focused on the short‐term joint effects of heat and PM_2.5 given the relatively short duration of both exposures. Examining potential joint effects over longer time windows would require a more flexible analytical approach, such as a distributed lag nonlinear model, which is beyond the scope of the current study. Although the association between cold and OHCA has been reported previously, future research is warranted to investigate whether PM_2.5 and low temperature jointly influence this health outcome using an alternative modeling framework.

In summary, this study revealed that the impact of heat is amplified by elevated PM_2.5, while the effects of PM_2.5 are significantly intensified by heat. Interactions on both additive and multiplicative scales were observed, with different patterns across subpopulations. Public health interventions addressing climate change should consider the synergistic health effects of heat and PM_2.5, particularly for OHCA. This includes implementing targeted adaptation actions for vulnerable groups and promoting mitigation measures that reduce both heat and PM_2.5 exposures.

Sources of Funding

This study was supported by the Environment Research and Technology Development Fund (JPMEERF20231007 and S‐20 [2], JPMEERF21S12006) of the Environmental Restoration and Conservation Agency, Ministry of the Environment of Japan, and by JSPS KAKENHI (Grant Nos. JP23K27848, JP25K13560, and JP26H02457).

Disclosures

The authors declare there is no conflict of interest.

Supporting information

STROBE Checklist

Tables S1–S5

Figures S1–S14