Interaction of Extreme Temperature Events and Nitrogen Dioxide on Out‐of‐Hospital Cardiac Arrest in a Subtropical Megacity: A Time‐Series Analysis Authors: Hongli Wan, Siquan Miao, Jingjing Wang, Wenming Shi, Huqiang Dong, Haisheng Wu, Yuqiang Ji Published: 2026 ABSTRACT BackgroundExtreme temperature events and nitrogen dioxide are established risk factors for cardiovascular disease, often co‐occurring during cold spells. However, their additive interactive effects on out‐of‐hospital cardiac arrest (OHCA) in subtropical climates remain unclear.MethodsTime‐series analysis was performed using 16 769 OHCA cases in Shenzhen, China (2015–2019). Extreme temperature events were defined by apparent temperature: cold spells and heatwaves were stratified by percentile thresholds (2.5th–10th for cold, 90th–97.5th for heat) and consecutive durations (2–4 days). Quasi‐Poisson regression with distributed lag nonlinear models was applied, and additive interactions were evaluated using relative excess risk due to interaction, attributable proportion, and synergy index.ResultsCold spells were associated with a significant severity‐dependent increase in OHCA risk, with risk ratios (RRs) rising from 1.19 (95% CI, 1.13–1.26) for mildest definition (≤10th percentile for ≥2 days) to 1.50 (95% CI, 1.26–1.78) for most extreme definition (≤2.5th percentile for ≥2 days), whereas heatwaves showed no significant associations. Cold spells and nitrogen dioxide (lag 0–3 days) showed significant synergy. Under the most extreme cold spell, coexposure yielded the highest risk (RR, 1.543 [95% CI, 1.307–1.822]), with 27.7% of OHCA cases attributable to this interaction. This synergy was most pronounced for shockable rhythms (relative excess risk due to interaction up to 1.57 [95% CI, 0.34–2.79]).ConclusionsExtreme cold and nitrogen dioxide synergistically increase OHCA risk in subtropical megacities, particularly for shockable rhythms—reversible with prompt defibrillation. Integrated early‐warning systems targeting compound cold and pollution exposures are needed to reduce mortality. Nonstandard Abbreviations and Acronyms attributable fraction attributable proportion apparent temperature distributed lag nonlinear model extreme temperature event nitrogen dioxide out‐of‐hospital cardiac arrest relative excess risk due to interaction Clinical PerspectiveWhat Is New? This first subtropical time‐series study shows that cold spells and nitrogen dioxide synergistically increase out‐of‐hospital cardiac arrest risk (up to 27.7% attributable), especially for shockable rhythms. What Are the Clinical Implications? Use cold spell and nitrogen dioxide forecasts to trigger prehospital alerts and implement dynamic emission controls during cold spells.On days with both cold spells and high nitrogen dioxide, advise high‐risk patients to limit outdoor exposure and ensure automated external defibrillator access, as timely defibrillation use reverses shockable rhythms. Clinical Perspective What Is New? This first subtropical time‐series study shows that cold spells and nitrogen dioxide synergistically increase out‐of‐hospital cardiac arrest risk (up to 27.7% attributable), especially for shockable rhythms. This first subtropical time‐series study shows that cold spells and nitrogen dioxide synergistically increase out‐of‐hospital cardiac arrest risk (up to 27.7% attributable), especially for shockable rhythms. This first subtropical time‐series study shows that cold spells and nitrogen dioxide synergistically increase out‐of‐hospital cardiac arrest risk (up to 27.7% attributable), especially for shockable rhythms. What Are the Clinical Implications? Use cold spell and nitrogen dioxide forecasts to trigger prehospital alerts and implement dynamic emission controls during cold spells.On days with both cold spells and high nitrogen dioxide, advise high‐risk patients to limit outdoor exposure and ensure automated external defibrillator access, as timely defibrillation use reverses shockable rhythms. Use cold spell and nitrogen dioxide forecasts to trigger prehospital alerts and implement dynamic emission controls during cold spells. Use cold spell and nitrogen dioxide forecasts to trigger prehospital alerts and implement dynamic emission controls during cold spells. On days with both cold spells and high nitrogen dioxide, advise high‐risk patients to limit outdoor exposure and ensure automated external defibrillator access, as timely defibrillation use reverses shockable rhythms. On days with both cold spells and high nitrogen dioxide, advise high‐risk patients to limit outdoor exposure and ensure automated external defibrillator access, as timely defibrillation use reverses shockable rhythms. Out‐of‐hospital cardiac arrest (OHCA) is a critical medical emergency with sudden onset and severe outcomes. Globally, survival rates following OHCA remain <10%. 1 In China, the age‐standardized incidence is ∼97.1 per 100 000, yet only 1.2% of patients survive to 30 days after the event. 2 The acute and often fatal nature of OHCA makes it a sensitive indicator of environmental triggers, underscoring the importance of identifying modifiable risk factors for public health prevention. Climate change and air pollution pose major environmental threats to cardiovascular health. Extreme temperature events (ETEs), particularly cold spells, are consistently associated with increased cardiovascular morbidity and mortality. 3 , 4 Physiological mechanisms include cold‐induced vasoconstriction, increased blood viscosity, and elevated cardiac workload, which can precipitate acute ischemic events. 5 Populations in subtropical regions are especially vulnerable due to insufficient physiological acclimatization and limited adaptive behaviors, such as the lack of centralized heating. 6 , 7 Concurrently, nitrogen dioxide (NO2), a traffic‐related air pollutant, exerts cardiotoxic effects through pathways including oxidative stress, systemic inflammation, and autonomic dysfunction. 8 Notably, NO2 concentrations often peak during colder periods due to atmospheric stability and increased fossil fuel consumption, creating a temporal overlap between extreme cold and poor air quality. 9 , 10 Growing evidence suggests that extreme temperatures and air pollutants may interact synergistically, producing greater health effects than each exposure alone. For instance, synergistic effects between extreme heat and particulate matter ≤2.5 μm (PM2.5) on myocardial infarction mortality have been reported, 11 and the association between NO2 and cardiovascular hospitalizations is stronger under low‐temperature conditions. 12 Mechanistically, low temperatures may amplify NO2 impacts through thermal inversions that trap pollutants, 13 and cold stress may enhance pulmonary absorption of NO2 and synergistically worsen autonomic instability. 14 Despite these advances, certain aspects warrant further investigation. Most previous studies have focused on mortality or hospitalization end points, with less attention given to OHCA—a more acute and lethal manifestation that may better reflect immediate environmental triggers. 2 Moreover, the interaction between cold spells and NO2 has not been sufficiently examined, particularly in subtropical megacities where populations are poorly acclimatized to cold and urban microclimates may intensify exposures. Furthermore, it remains unclear whether these risks differ across population subgroups (eg, by age, sex, initial cardiac rhythm, or arrest causes), which could reveal variations in susceptibility or underlying pathological mechanisms. To address these gaps, this time‐series study analyzed 16 769 OHCA cases in Shenzhen, China (2015–2019). As a subtropical metropolis with high population density, substantial traffic‐related pollution, and mild but fluctuating winter temperatures, Shenzhen provides an ideal setting to examine compound environmental effects. This study aims to (1) evaluate the individual and joint effects of ETEs and NO2 exposure on OHCA risk, (2) quantify any additive interactions between these exposures, and (3) examine whether these associations vary across demographic subgroups (by sex and age) or by initial cardiac rhythm (shockable versus nonshockable) or by causes of arrest. The findings are expected to provide epidemiological evidence that can support the development of integrated early‐warning systems and targeted public health interventions aimed at reducing the cardiovascular burden associated with combined environmental exposures. METHODS Data Availability The data sets generated and analyzed during the current study are not publicly available due to privacy restrictions but are available from the corresponding author upon reasonable request. Study Setting and Climate Characteristics This time‐series study included 16 769 OHCA cases occurring in Shenzhen, China, from 2015 to 2019. Shenzhen is a coastal metropolis in southern China (22°27′ N–22°52′ N), with a permanent population of 17.99 million and an area of 1997.47 km2. 2 , 15 The city experiences a subtropical monsoon climate. During the study period, the daily mean temperature averaged 23.7 °C (SD: 5.3 °C); summer (June–August) temperatures ranged from 25 °C to 30 °C, and winter (December–February) temperatures ranged from 10 °C to 17 °C, based on interquartile ranges. Relative humidity remains high year‐round (annual mean: 76%), 15 which can substantially modulate the physiological impact of ambient temperature. Therefore, apparent temperature—which integrates both temperature and humidity—was used to define extreme temperature events. Outcome Variable Following the Utstein‐style registry protocol, cases due to external causes, such as trauma, drowning, poisoning, asphyxiation, or electric shock, were excluded. The initial cardiac rhythm was categorized based on the first rhythm recorded by emergency medical services (EMS) personnel upon arrival. Shockable rhythms included ventricular fibrillation or pulseless ventricular tachycardia, and nonshockable rhythms comprised asystole and pulseless electrical activity. 16 The study was approved by the institutional review board at the Ethics Committee of Shenzhen Center for Prehospital Care (2023–071‐03PJ). As the data are anonymous, the requirement for informed consent was waived, and we have been granted exemption from the institutional during the ethics approval process. Exposure Assessment Meteorological data, including daily mean ambient temperature (°C) and relative humidity (%), were collected from Station 59 493 in Guangdong, China (https://rp5.ru/). Daily mean apparent temperature (AT, °C), which integrates the combined effects of temperature and humidity, was calculated using the weathermetrics package in R. Following established epidemiological approaches for defining temperature extremes where universal thresholds are absent, 6 , 17 this study computed the 10th, 7.5th, 5th, and 2.5th percentiles of daily AT as cold spell thresholds, and the 90th, 92.5th, 95th, and 97.5th percentiles as heatwave thresholds. Cold spells were operationally defined as ≥2, 3, or 4 consecutive days with daily AT at or below specified percentiles (P10, P7.5, P5, or P2.5). Similarly, heatwaves were defined as ≥2, 3, or 4 consecutive days with daily AT at or above the corresponding high percentiles (P90, P92.5, P95, or P97.5). This generated 24 mutually exclusive ETEs definitions. Each day was coded as 1 for a heatwave, 2 for a cold spell, or 0 for a non‐ETE day. The use of consecutive days for temperature extremes reflects the sustained physiological stress associated with prolonged exposure, a common approach in environmental epidemiology. 6 , 17 Daily air pollution data, including 24‐hour average concentrations of PM2.5, sulfur dioxide, NO2, carbon monoxide, and maximum 8‐hour average ozone, were obtained from China's National Urban Air Quality Real‐time Publishing Platform (https://air.cnemc.cn:18007/). For NO2, a 3‐day moving average (lag 0–3 days) was used to characterize cumulative short‐term exposure, as this window was identified as optimal by generalized cross‐validation in the main analysis. This approach captures the delayed and cumulative effects of air pollution on health outcomes. Statistical Analysis The association between ETEs, NO2, and daily OHCA counts was examined using a 2‐stage analytical approach. First, a distributed lag nonlinear model (DLNM) within a quasi‐Poisson regression framework was applied to assess the overall and lagged effects of NO2 as a continuous exposure, while accounting for ETEs and potential confounders. Second, to formally evaluate additive interactions between ETEs and NO2, NO2 was dichotomized using a predefined threshold, and a conventional generalized linear model with categorical variables was employed. Stage 1: Continuous Exposure Analysis Using DLNM The core model was specified as: logEYt=α+β·ETEt+cbNO2,t+nsTimedf+γ·DOWt where EYt denotes the expected daily OHCA count on day t; α is the intercept; ETEt is a categorical variable coded as 0 (nonevent day), 1 (heatwave day), or 2 (cold spell day) according to the definitions in Table S1; cbNO2,t represents the cross‐basis matrix for NO2 exposure; nsTimedf is a natural cubic spline of time with 7 degrees of freedom per year to control for long‐term trends and seasonality; and DOWt adjusts for day of the week. For the NO2 cross‐basis, natural cubic splines with 3 degrees of freedom were used for both the exposure‐response and lag‐response dimensions. The maximum lag period was selected by minimizing the generalized cross‐validation criterion across candidate lags from 0 to 14 days; the optimal maximum lag was 3 days, which was applied in all subsequent models. This generalized cross‐validation ‐based selection is consistent with established methodological recommendations for DLNM in environmental epidemiology, as it effectively balances model fit and overfitting while avoiding biases from variable sample sizes. To explore effect modification by temperature, stratified analyses were performed by partitioning the data based on specific ETE definitions: cold spell days (AT ≤10th percentile for ≥2 consecutive days) versus noncold spell days, heatwave days (AT ≥90th percentile for ≥2 consecutive days) versus non–heatwave days, and any ETE day versus nonevent day. Separate DLNMs were fitted within each stratum. Stage 2: Dichotomized Exposure Analysis for Additive Interaction To quantify additive interactions between ETEs and NO2, NO2 exposure was dichotomized using a 3‐day moving average (lag 0–3 days) of the daily concentrations, with the cutoff set at the World Health Organization annual guideline value of 25 μg/m3. A 4‐level categorical variable was then created combining ETE status (event versus non‐event) and NO2 status (low versus high), with “non‐ETE and low NO2” as the reference. Quasi‐Poisson regression models were fitted including this categorical variable, time spline, and day of week. From these models, the relative excess risk due to interaction (RERI), attributable proportion (AP), and synergy index were calculated using the standard formulas based on risk ratios (RRs). CIs for RERI, AP, and synergy index were estimated using the Delta method. As a sensitivity analysis, a bootstrap resampling approach (1000 replications) was also applied to obtain 95% CIs, which yielded similar results. Subgroup and Sensitivity Analyses Subgroup analyses were exploratory in nature and conducted to preliminarily examine potential effect heterogeneity of the ETE‐NO2 interaction across key population and clinical subgroups: sex (male versus female), age group (<65 vs ≥65 years), initial cardiac rhythm (shockable versus nonshockable), and cause of arrest (cardiovascular versus other). Differences in RRs or interaction measures between subgroups were tested using a 2‐sample Z‐test: z=β1−β2SEβ12+SEβ22 where β1 and β2 are the coefficients for comparable parameters in 2 subgroup models, and SEβ1 and SEβ2 are their respective SEs. Two‐tailed P values were derived from the standard normal distribution. Model diagnostics, including residual plots, overdispersion tests, and autocorrelation function plots, were conducted for all primary models and are summarized in the Supplementary Materials. Several sensitivity analyses were performed to assess robustness: (1) negative binomial regression—to further address potential overdispersion, all main models were refitted using negative binomial regression, employing the same covariate structure (spline of time, day of week) and cross‐basis for NO2 (with the same maximum lag of 3 days) as in the primary quasi‐Poisson models; (2) copollutant adjustment—the core model was refitted with additional adjustment for PM2.5, ozone, sulfur dioxide, and carbon monoxide individually and simultaneously (PM2.5 and PM10 were highly correlated, see Table S2; only PM2.5 was adjusted in the main sensitivity analysis), using a moving average of lag 0–3 days; (3) alternative ETE definition—ETEs were redefined using daily mean ambient temperature instead of apparent temperature; (4) alternative lag structures—the lag period for NO2 in the cross‐basis was varied (lag0, lag01, lag02); and (5) alternative NO2 threshold—the additive interaction analysis was repeated using the median NO2 concentration as the cutoff. All statistical analyses were performed using R software (version 4.3.0) with the packages dlnm, splines, and tidyverse. Statistical significance was inferred from 95% CIs excluding the null value. RESULTS Descriptive Statistics and Temporal Patterns A total of 16 769 OHCA cases were documented in Shenzhen between 2015 and 2019. Most cases occurred in men (68.29%, n=11 451). By age, 52.64% (n=8828) were individuals <65 years, and 47.36% (n=7941) were aged 65 or older. A clear seasonal variation was observed, with the highest incidence in winter (28.31%) and summer (25.26%), and lower rates in spring (22.97%) and autumn (23.46%) (Table 1). Meteorologically, AT averaged 25.8 °C (SD: 7.7 °C) during the study period. Cold spell thresholds, defined by percentiles of apparent temperature, corresponded to ∼15.6 °C (10th percentile), 14.4 °C (7.5th percentile), 13.3 °C (5th percentile), and 11.1 °C (2.5th percentile). Heatwave thresholds corresponded to ∼36.1 °C (90th percentile), 36.7 °C (92.5th percentile), 37.2 °C (95th percentile), and 37.8 °C (97.5th percentile) (Table S3). Time‐series analysis revealed synchronous patterns: winter months (November to January) featured lower AT, elevated NO2 levels, and higher mean daily OHCA counts (Figure S1). Table 1Characteristics of OHCA Cases (n=16 769)CharacteristicCase (n %)All16 769Age, y<658828 (52.64)≥657941 (47.36)SexMale11 451 (68.29)Female5318 (31.71)Initial rhythmShockable rhythm1160 (6.92)Nonshockable rhythm15 609 (93.08)SeasonSpring (March to May)3852 (22.97)Summer (June to August)4236 (25.26)Autumn (September to November)3934 (23.46)Winter (December to February)4747 (28.31)OHCA indicates out‐of‐hospital cardiac arrest. Characteristics of OHCA Cases (n=16 769) OHCA indicates out‐of‐hospital cardiac arrest. Exposure Distribution and Preliminary Associations The number of ETE days defined by different percentile thresholds and consecutive durations is detailed in Table S4, with corresponding OHCA case counts for each definition shown in Table S5. Mean daily OHCA counts across ETE and NO2 exposure combinations (NO2 by 3‐day moving average, lag 0–3 days, 25 μg/m3 threshold) are presented in Table S6 and visualized in Figure 1. Preliminary results showed a consistent pattern: days with combined ETE and high NO2 exposure had the highest mean daily OHCA counts, followed by ETE exposure alone (low NO2), then high NO2 exposure alone (no ETE), with days of neither exposure recording the lowest burden. Additionally, a clear severity‐response trend was observed: mean daily OHCA counts increased with higher ETE exposure severity, including cold spells with lower temperature percentiles and longer durations, and heatwaves with higher temperature percentiles and longer durations. Mean daily counts of out‐of‐hospital cardiac arrest under combined exposure to extreme temperature events and nitrogen dioxide across varying ETE definitions in Shenzhen, China (2015–2019). CA indicates cardiac arrest; and NO2, nitrogen dioxide. Exposure‐Response Associations of ETEs and NO2 with OHCA After adjusting for NO2 exposure (lag 0–3 days) and potential confounders using quasi‐Poisson regression with DLNMs, the associations between ETEs and OHCA risk were analyzed, with results presented in Figure 2 and Table S7. All cold spell definitions significantly increased OHCA risk, with RRs exhibiting a monotonic dose–response relationship with cold severity: the RR was 1.19 (95% CI, 1.13–1.26) for the mildest definition (AT ≤P10 for ≥2 consecutive days) and peaked at 1.50 (95% CI, 1.26–1.78) for the most extreme one (AT ≤P2.5 for ≥2 consecutive days). In contrast, heatwave RRs ranged from 1.00 (95% CI, 0.93–1.06) to 1.18 (95% CI, 0.98–1.43) across all definitions, with no statistically significant estimates (Table S7 and Figure 2A). Generalized cross‐validation in the DLNM framework identified the 3‐day moving average (lag 0–3 days) as the optimal exposure window for NO2, with its exposure‐response relationship with OHCA risk shown in Figure 2B. OHCA risk rose with increasing NO2 concentrations on both cold spell and noncold spell days, but the exposure–response curve was markedly steeper during cold spells (AT ≤P10 for ≥2 consecutive days), and the curve for cold spell days lay entirely above that for noncold spell days (Figure 2C), indicating heightened vulnerability to NO2 exposure in cold conditions. A similar but less pronounced effect modification was observed for heatwaves (Figure 2D) and any extreme temperature event (Figure 2E): NO2‐associated OHCA risks on ETE days were slightly higher than on nonevent days, without obvious amplification of the exposure–response relationship. Associations between extreme temperature events, NO2 exposure, and OHCA risk. A, RRs of OHCA for different extreme temperature event definitions. The dashed line indicates the null effect (RR = 1). B, Overall exposure‐response curve of OHCA risk associated with NO2 (lag 0–3 days) from the main model adjusted for ETEs. C–E, Exposure–response curves of OHCA risk associated with lag 0 to 3 days NO2 under different temperature conditions: (C) cold spells (≥2 consecutive days with AT ≤10th percentile) vs noncold spell days, (D) heatwaves (≥2 consecutive days with AT ≥90th percentile) vs non–heatwave days, and (E) any extreme temperature event (cold spell or heatwave) vs nonevent days. ETEs indicates extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; and RR, risk ratio. Interaction Effects Between ETEs and NO2 Significant additive interactions between cold spells and NO2 were observed only for the P2.5_2d and P7.5_2d definitions, whereas no significant interaction was detected for heatwaves across all percentile‐duration definitions (Figure 3, Tables 2 and 3). RR of OHCA associated with combined exposure to ETEs and NO2. RR01, non‐ETE with high NO2 exposure; RR10, ETE with low NO2 exposure; RR11, combined ETE with high NO2 exposure. For example, P10_2d represents a cold spell where daily apparent temperature is ≤10th percentile for at least 2 consecutive days. ETEs indicates extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RERI, relative excess risk due to interaction; and RR, risk ratio. Table 2Risk Ratios (95% CI) for OHCA With Exposure to ETEs and NO2 EventRR01 (95% CI)RR10 (95% CI)RR11 (95% CI)Cold spellP10_2d1.037 (1.001–1.074)1.177 (1.034–1.339)1.277 (1.197–1.361)P10_3d1.039 (1.003–1.075)1.242 (1.077–1.432)1.348 (1.256–1.447)P10_4d1.041 (1.006–1.078)1.257 (1.055–1.497)1.367 (1.261–1.481)P7.5_2d1.039 (1.003–1.075)1.090 (0.943–1.261)1.327 (1.229–1.433)P7.5_3d1.044 (1.008–1.081)1.193 (1.014–1.404)1.405 (1.283–1.539)P7.5_4d1.046 (1.011–1.083)1.228 (1.012–1.489)1.442 (1.297–1.603)P5_2d1.045 (1.010–1.082)1.104 (0.945–1.291)1.353 (1.228–1.491)P5_3d1.050 (1.014–1.087)1.193 (1.008–1.413)1.379 (1.226–1.551)P5_4d1.050 (1.015–1.087)1.220 (0.995–1.496)1.478 (1.286–1.698)P2.5_2d1.046 (1.011–1.083)1.034 (0.830–1.289)1.495 (1.310–1.706)P2.5_3d1.050 (1.014–1.087)1.227 (0.924–1.629)1.543 (1.307–1.822)P2.5_4d1.051 (1.015–1.088)1.322 (0.865–2.020)1.512 (1.247–1.833)HeatwaveP90_2d1.063 (1.025–1.102)1.028 (0.947–1.116)0.932 (0.837–1.036)P90_3d1.060 (1.023–1.099)1.061 (0.967–1.164)1.023 (0.897–1.166)P90_4d1.059 (1.022–1.097)1.068 (0.961–1.187)1.015 (0.843–1.223)P92.5_2d1.059 (1.022–1.098)1.033 (0.941–1.134)0.980 (0.864–1.112)P92.5_3d1.056 (1.020–1.095)1.046 (0.941–1.163)1.074 (0.914–1.262)P92.5_4d1.055 (1.018–1.093)1.044 (0.925–1.178)1.184 (0.948–1.478)P95_2d1.058 (1.022–1.096)1.071 (0.957–1.198)1.034 (0.888–1.204)P95_3d1.055 (1.019–1.093)1.073 (0.935–1.232)1.239 (1.000–1.534)P95_4d1.054 (1.018–1.091)1.045 (0.875–1.249)1.294 (0.903–1.854)P97.5_2d1.056 (1.020–1.094)1.094 (0.948–1.262)1.110 (0.931–1.323)P97.5_3d1.056 (1.020–1.093)1.123 (0.950–1.327)1.176 (0.900–1.535)P97.5_4d1.055 (1.019–1.092)1.138 (0.915–1.415)1.295 (0.839–1.999)ETE indicates extreme temperature event; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RR01, risk ratio for non‐ETE and high NO2 exposure; RR10, risk ratio for ETE and low NO2 exposure and RR11, risk ratio for combined ETE and high NO2 exposure. Risk Ratios (95% CI) for OHCA With Exposure to ETEs and NO2 ETE indicates extreme temperature event; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RR01, risk ratio for non‐ETE and high NO2 exposure; RR10, risk ratio for ETE and low NO2 exposure and RR11, risk ratio for combined ETE and high NO2 exposure. Table 3Measures of Additive Interaction Between ETEs and NO2 on OHCA RiskEventRERI (95% CI)AP (95% CI)SI (95% CI)Cold spellP10_2d0.063 (−0.114 to 0.24)0.049 (−0.084 to 0.200)1.294 (0.269 to 2.320)P10_3d0.068 (−0.137 to 0.272)0.050 (−0.095 to 0.217)1.241 (0.368 to 2.114)P10_4d0.069 (−0.180 to 0.318)0.050 (−0.121 to 0.252)1.231 (0.233 to 2.229)P7.5_2d0.199 (0.007 to 0.390)* 0.150 (0.005 to 0.317)* 2.542 (−0.773 to 5.858)P7.5_3d0.168 (−0.067 to 0.403)0.120 (−0.043 to 0.314)1.710 (0.183 to 3.236)P7.5_4d0.168 (−0.116 to 0.452)0.117 (−0.072 to 0.348)1.614 (0.089 to 3.138)P5_2d0.204 (−0.016 to 0.424)0.151 (−0.011 to 0.345)2.364 (−0.569 to 5.297)P5_3d0.136 (−0.126 to 0.397)0.098 (−0.081 to 0.324)1.557 (0.081 to 3.033)P5_4d0.207 (−0.118 to 0.532)0.140 (−0.069 to 0.414)1.766 (−0.053 to 3.584)P2.5_2d0.415 (0.111 to 0.719)* 0.277 (0.065 to 0.548)* 6.169 (−11.864 to 24.202)P2.5_3d0.266 (−0.168 to 0.699)0.172 (−0.092 to 0.535)1.958 (−0.713 to 4.629)P2.5_4d0.139 (−0.494 to 0.771)0.092 (−0.269 to 0.619)1.372 (−0.896 to 3.639)HeatwaveP90_2d−0.159 (−0.295 to −0.023)−0.171 (−0.285 to −0.028)−0.753 (−2.091 to 0.586)P90_3d−0.099 (−0.269 to 0.071)−0.097 (−0.231 to 0.08)0.185 (−0.933 to 1.304)P90_4d−0.112 (−0.335 to 0.111)−0.110 (−0.274 to 0.132)0.121 (−1.377 to 1.618)P92.5_2d−0.112 (−0.273 to 0.049)−0.114 (−0.245 to 0.057)−0.218 (−1.588 to 1.152)P92.5_3d−0.029 (−0.238 to 0.180)−0.027 (−0.188 to 0.197)0.721 (−1.161 to 2.603)P92.5_4d0.085 (−0.209 to 0.378)0.072 (−0.141 to 0.399)1.857 (−1.785 to 5.498)P95_2d−0.095 (−0.297 to 0.107)−0.092 (−0.247 to 0.120)0.264 (−0.989 to 1.516)P95_3d0.110 (−0.196 to 0.416)0.089 (−0.127 to 0.416)1.861 (−1.185 to 4.906)P95_4d0.194 (−0.308 to 0.697)0.150 (−0.166 to 0.772)2.958 (−4.469 to 10.385)P97.5_2d−0.040 (−0.292 to 0.213)−0.036 (−0.221 to 0.228)0.734 (−0.790 to 2.259)P97.5_3d−0.003 (−0.370 to 0.364)−0.003 (−0.241 to 0.404)0.982 (−1.083 to 3.047)P97.5_4d0.102 (−0.513 to 0.718)0.079 (−0.257 to 0.855)1.530 (−2.086 to 5.146)AP indicates attributable proportion; ETEs, extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RERI, relative excess risk due to interaction; and SI, synergy index.* P<0.05. Measures of Additive Interaction Between ETEs and NO2 on OHCA Risk AP indicates attributable proportion; ETEs, extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RERI, relative excess risk due to interaction; and SI, synergy index. P<0.05. For cold spells, synergistic effects strengthened with increasing exposure severity. The strongest synergy was seen in P2.5_2d: combined exposure yielded the highest risk (RR11, 1.495 [95% CI, 1.310–1.706]), markedly exceeding the individual effects of high NO2 alone (RR01, 1.046 [95% CI, 1.011–1.083]) and cold spells alone (RR10, 1.034 [95% CI, 0.830–1.289]). Formal interaction metrics confirmed this synergy (RERI, 0.415 [95% CI, 0.111–0.719]; AP, 0.277 [95% CI, 0.065–0.548]; synergy index=6.169), with ~27.7% of OHCA cases attributable to the interaction. A significant interaction was also found for P7.5_2d (RERI, 0.199 [95% CI, 0.007–0.390]; AP, 0.150 [95% CI, 0.005–0.317]). All other cold spell definitions showed consistent positive point estimates for RRs and interaction metrics (RERI >0, AP>0, synergy index>1), indicating a similar directional trend without statistical significance. In contrast, no significant additive interaction between heatwaves and NO2 was identified. Although a small number of heatwave‐NO2 coexposure RRs marginally exceeded unity, all interaction metrics (RERI, AP, synergy index) had 95% CIs encompassing the null value across all definitions (Tables 2 and 3). Residual diagnostics (Table S8, Figure S2) confirmed no overdispersion or abnormal residual patterns, validating robust model fits. Attributable Burden Figure 4 and Tables S9 and S10 present the population attributable fraction (AF) and attributable number of OHCA from individual and combined ETE and high NO2 exposure. For cold spells, combined exposure had the highest AF across all definitions, which rose with cold severity and peaked at 35% (95% CI, 23%–45%) for P2.5_3d (P2.5_2d: 33%; P2.5_4d: 34%). High NO2 alone had a modest, stable AF (4%–5%), whereas AF for cold spells alone was variable with 95% CIs often including zero. Population attributable fraction and number of OHCA cases due to individual and combined exposure to extreme temperature events and nitrogen dioxide. A, the attributable fraction (AF); B, the attributable number (AN). AETEs indicates extreme temperature events; and NO2, nitrogen dioxide; and OHCA, out‐of‐hospital cardiac arrest. For heatwaves, combined exposure showed AF values near zero or negative with null‐inclusive CIs, and no severity‐related upward trend was observed. In terms of absolute burden, high NO2 alone accounted for the largest attributable number of excess OHCA cases (344–526 for cold spells, 560–626 for heatwaves), far exceeding ETEs alone or combined exposure. Cold spell combined exposure attributable number declined with increasing severity (287 cases for P10_2d to 42 for P2.5_4d), attributable to fewer extreme cold days in this subtropical region; heatwaves had negligible attributable number across all exposure scenarios. These results confirm that cold spell‐high NO2 coexposure confers the highest individual‐level relative OHCA risk, and chronic high NO2 exposure drives the greatest population‐level absolute attributable burden. Heatwaves contributed minimally to OHCA attributable burden, either alone or in combination with high NO2. Subgroup Analyses Stratified analyses by sex, age, initial cardiac rhythm, and cause of arrest were conducted to explore effect modification of the interaction between ETEs and NO2, with results focused on extreme cold spell (AT ≤ P2.5) and heatwave (AT ≥ P97.5) definitions (Table 4). Overall, the combined effect (RR11) of cold spells and high NO2 was consistently >1 across all subgroups, but the magnitude of additive interaction varied markedly; no significant synergistic effects between heatwaves and NO2 were observed in any stratification. Table 4RR11 and RERI of Exposure to ETEs and NO2 on OHCA by Sex, Age, Initial Rhythm, and CausesDefinitionSexAgeInitial rhythmCausesMaleFemale≥65 y<65 yShockable rhythmNonshockable rhythmCardiovascularOther causesRR11 Cold spellP2.5_2d1.22 (1.01 to 1.47)1.26 (0.98 to 1.62)1.12 (0.90 to 1.39)1.37 (1.10 to 1.70)2.20 (1.30 to 3.72)1.18 (1.00 to 1.38)* 1.24 (0.96 to 1.61)1.23 (1.01 to 1.49)P2.5_3d1.32 (1.06 to 1.64)1.20 (0.88 to 1.62)1.19 (0.93 to 1.53)1.37 (1.05 to 1.78)1.93 (0.98 to 3.78)1.24 (1.03 to 1.49)1.27 (0.94 to 1.72)1.27 (1.01 to 1.60)P2.5_4d1.40 (1.11 to 1.76)1.10 (0.77 to 1.56)1.18 (0.89 to 1.55)1.43 (1.07 to 1.90)1.60 (0.74 to 3.46)1.27 (1.04 to 1.56)1.21 (0.87 to 1.70)1.33 (1.04 to 1.71)Heat waveP97.5_2d1.20 (0.94 to 1.52)1.25 (0.89 to 1.76)1.12 (0.83 to 1.51)1.30 (0.99 to 1.71)1.44 (0.68 to 3.07)1.20 (0.98 to 1.47)1.41 (1.02 to 1.94)1.11 (0.86 to 1.45)P97.5_3d1.29 (0.93 to 1.77)1.42 (0.91 to 2.22)1.12 (0.74 to 1.69)1.53 (1.08 to 2.17)1.34 (0.47 to 3.81)1.33 (1.01 to 1.74)1.54 (1.01 to 2.35)1.22 (0.87 to 1.73)P97.5_4d1.28 (0.64 to 2.58)1.62 (0.67 to 3.91)1.36 (0.60 to 3.11)1.42 (0.66 to 3.08)1.28 (0.16 to 10.12)1.40 (0.79 to 2.50)2.05 (0.95 to 4.42)1.01 (0.44 to 2.34)RERICold spellP2.5_2d0.24 (−0.04 to 0.53)0.41 (0.02 to 0.79)† 0.22 (−0.08 to 0.53)0.39 (0.03 to 0.75)† 1.21 (−0.06 to 2.47)0.25 (0.01 to 0.48)† 0.12 (−0.30 to 0.53)0.40 (0.11 to 0.68)† P2.5_3d0.27 (−0.11 to 0.65)0.19 (−0.32 to 0.70)0.23 (−0.18 to 0.64)0.26 (−0.23 to 0.75)1.25 (−0.21 to 2.71)0.18 (−0.14 to 0.50)−0.05 (−0.62 to 0.52)0.39 (0.01 to 0.77)† P2.5_4d0.40 (−0.10 to 0.91)0.13 (−0.54 to 0.79)0.12 (−0.43 to 0.68)0.53 (−0.08 to 1.14)1.57 (0.34 to 2.79)† 0.23 (−0.20 to 0.65)* 0.10 (−0.61 to 0.81)0.42 (−0.09 to 0.93)HeatwaveP97.5_2d0.14 (−0.19 to 0.46)−0.01 (−0.50 to 0.48)−0.12 (−0.51 to 0.27)0.28 (−0.11 to 0.67)−0.11 (−1.34 to 1.11)0.11 (−0.17 to 0.39)−0.00 (−0.51 to 0.50)0.15 (−0.18 to 0.47)P97.5_3d0.20 (−0.25 to 0.66)0.14 (−0.56 to 0.84)−0.16 (−0.69 to 0.37)0.50 (−0.07 to 1.07)−0.31 (−1.90 to 1.28)0.22 (−0.17 to 0.62)0.03 (−0.70 to 0.75)0.27 (−0.19 to 0.73)P97.5_4d0.13 (−0.80 to 1.06)0.24 (−1.24 to 1.72)−0.03 (−1.20 to 1.14)0.35 (−0.78 to 1.48)−0.78 (−3.65 to 2.09)0.25 (−0.59 to 1.08)0.43 (−1.19 to 2.05)0.01 (−0.88 to 0.89)ETEs indicates extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RR, risk ratio; and RERI, relative excess risk due to interaction.* P difference<0.05, estimated using the 2‐sample Z test. † The 95% CI does not include 0 (statistically significant). RR11 and RERI of Exposure to ETEs and NO2 on OHCA by Sex, Age, Initial Rhythm, and Causes ETEs indicates extreme temperature events; NO2, nitrogen dioxide; OHCA, out‐of‐hospital cardiac arrest; RR, risk ratio; and RERI, relative excess risk due to interaction. P difference<0.05, estimated using the 2‐sample Z test. The 95% CI does not include 0 (statistically significant). The most prominent effect modification was observed for initial cardiac rhythm. RR11 values for shockable rhythms were substantially higher than those for nonshockable rhythms across all P2.5‐based cold spell definitions, peaking at 2.20 (95% CI, 1.30–3.72) for P2.5_2d (nonshockable rhythms: 1.18–1.27). Significant additive interactions were detected in shockable rhythms for P2.5_4d (RERI, 1.57 [95% CI, 0.34–2.79]), with a significant between‐group difference compared with nonshockable rhythms (P=0.037; Table S11). Stratification by cause of arrest revealed distinct patterns: significant additive interactions were observed in the “other causes” subgroup for P2.5_2d (RERI, 0.40 [95% CI, 0.11–0.68]) and P2.5_3d (RERI, 0.39 [95% CI, 0.01–0.77]), whereas all RERI estimates in the cardiovascular cause subgroup were near zero with no statistical significance. In demographic subgroups (sex, age), significant additive interactions were observed only for specific cold spell durations, with no between‐group differences (all P>0.05). Women and individuals <65 years showed significant synergy for P2.5_2d. Although point estimates of RR11 and RERI were higher in men for longer‐duration cold spells (P2.5_3d; P2.5_4d) and higher in women for shorter‐duration cold spells (P2.5_2d), no statistically significant between‐group differences were observed. For heatwaves (AT ≥ P97.5), no significant additive interactions were identified in any subgroup, consistent with the main results. Descriptive analyses of the proportion of daily OHCA cases under combined ETE and high NO2 exposure further supported these patterns (Figure S3). Across all cold spell definitions, proportional burdens showed minimal differences by sex and age, whereas shockable rhythm and “other causes” cases consistently had a higher proportional share under coexposure, aligning with the statistical findings. In brief, the synergistic effect of cold spells and NO2 on OHCA risk was predominantly pronounced in cases with shockable rhythms and “other causes” of arrest, with weak and limited effect modification by sex and age. Caution is warranted in interpreting these findings due to the limited number of events in the shockable rhythm subgroup. Sensitivity Analysis Sensitivity analyses confirmed the robustness of the core findings. Negative binomial regression was applied as a sensitivity analysis. The estimated risk ratios and interaction measures were nearly identical to those from the quasi‐Poisson models (Tables S12–S15; Figures S4–S6). Likelihood ratio tests showed no significant overdispersion (P values 0.44–0.56; Table S16), and residual diagnostics indicated adequate model fit (Ljung‐Box P>0.05; Table S17). When ambient temperature was used instead of AT to define ETEs, RR11 remained >1 across all thresholds and increased with cold severity, and the RERI for the P2.5_2d remaining positive and significant (Table S18). After adjusting for individual or all copollutants (PM2.5, ozone, sulfur dioxide, carbon monoxide), RR11 for combined exposure was consistently >1, and the significant positive RERI for P2.5_2d was maintained (Tables S19–S23). Varying the NO2 exposure lag structure (lag0, lag01, lag02) yielded consistent RR and RERI estimates, with RR11>1 and RERI for P2.5_2d remaining significant across all lag periods (Tables S24–S26). Redefining the NO2 dichotomization threshold as the median concentration also preserved the directional synergy, with P2.5_2d still showing a significant additive interaction (Table S27). Bootstrap resampling (1000 replicates) yielded 95% CIs for interaction measures that were consistent with the Delta method results (Table S28). Collectively, the synergistic effect between extreme cold spells and NO2 on OHCA risk was stable across adjustments in ETE definition, co‐pollutant inclusion, NO2 lag structure and exposure threshold. DISCUSSION This time‐series study analyzed 16 769 OHCA cases in Shenzhen to investigate the interaction between ETEs and NO2 exposure. The findings reveal a significant synergistic effect, with up to 27.7% of OHCA cases under the most extreme cold spell (P2.5_2d) and high NO2 coexposure attributable to this interaction, underscoring its public health significance. Notably, the significant additive interaction was most pronounced for shockable cardiac rhythms, suggesting an acute electrophysiological mechanism may be central to this compound effect. Cold spells exhibited a clear dose–response relationship with OHCA risk, with risk ratios increasing from 1.19 (95% CI, 1.13–1.26) for moderate events (≤10th percentile, lasting ≥2 days) to 1.50 (1.26–1.78) for the most extreme cold spells (≤2.5th percentile, ≥2 days). This pattern is consistent with multicountry studies reporting greater cardiovascular risks from cold than from heat across diverse climates. 18 A nationwide study in China found an additional 9.1 cardiovascular deaths per 1000 cases on cold days, with steeper cold–mortality slopes in subtropical cities—likely due to poor acclimatization and lack of centralized heating. 19 The present results extend these observations by demonstrating that even moderate cold spells significantly increase OHCA risk in Shenzhen, where mild winters and high humidity coexist. 20 Unlike prior studies focused on cardiovascular mortality or hospitalization, this analysis targets OHCA—a more acute end point that better captures immediate environmental triggers—adding evidence for the acute synergistic effect of cold and NO2 in subtropical regions. Populations in warm climates such as Shenzhen may be uniquely vulnerable to cold stress due to limited physiological acclimatization and inadequate adaptive measures (eg, lack of centralized heating), which amplify cold's physiological impacts. 20 Independent of temperature, NO2 exposure at lag 0 to 3 days was associated with a modest increase in OHCA risk, consistent with recent multicenter studies in China showing a 1.0% to 1.2% rise in OHCA per 10 μg/m3 NO2 increment. 21 The consistency of this effect is supported by a global meta‐analysis encompassing 398 cities, which estimated a nearly identical risk increment of 0.9% (95% CI, 0.4%–1.4%) per 10 μg/m3 NO2. 22 Importantly, this exposure–response relationship appears to lack a discernible safety threshold, as significant associations persist even in areas where NO2 concentrations meet the current World Health Organization annual guideline. This suggests a pervasive population‐level risk, albeit modest in magnitude for a single individual. The 3‐day moving average (lag 0–3 days) for NO2 exposure was selected via generalized cross‐validation in the DLNM framework, which identified this window as optimal for capturing cumulative short‐term effects; sensitivity analyses with alternative lag structures (lag0, lag01, lag02) produced consistent results, confirming the robustness of the observed associations. Of note, although cold spell and high NO2 coexposure had the highest relative risk (AF peak at 35% for P2.5_3d), high NO2 alone contributed the largest absolute number of excess OHCA cases, with 344 to 526 cases in cold spell scenarios and 560 to 626 cases in heatwave scenarios. This highlights that public health strategies need to address both short‐term compound environmental exposures and long‐term NO2 pollution control to reduce the overall OHCA burden. Beyond the individual effects of cold spells and NO2, a statistically significant additive interaction was observed between the 2 exposures. Under the most stringent cold spell definition (≤2.5th percentile, ≥2 days), co‐exposure resulted in a RERI of 0.415 (95% CI, 0.111–0.719) and an AP of 27.7%, indicating that more than one quarter of OHCA cases were attributable to this synergistic effect. This finding aligns with previous reports indicating that cold weather amplifies the health effects of NO2. For example, an analysis of daily mortality in Shenzhen revealed that the association between NO2 and cardiovascular deaths was nearly doubled during cold periods, 23 and studies in Ganzhou 24 and Urumqi 25 have similarly reported additive interactions between low temperature and NO2 exposure. A large‐scale study across 291 Chinese cities also documented enhanced effects of NO2 on health outcomes during cold days, 12 although it did not provide a formal quantification of the interaction. In addition, negative binomial regression, conducted as a sensitivity analysis in this study, yielded consistent effect estimates and no evidence of significant overdispersion (likelihood ratio test P>0.05), supporting the robustness of the quasi‐Poisson findings. Notably, both the combined effect and synergistic interaction were more pronounced in cases with shockable rhythms than in those with nonshockable rhythms. This differential vulnerability may stem from distinct pathophysiological mechanisms. Shockable rhythms are typically driven by acute electrical instability, often involving reentry circuits initiated by premature ventricular complexes. 26 Cold stress increases sympathetic activity, prolongs the QT interval, 27 and alters myocardial conduction and refractoriness. 28 Concurrently, NO2 exposure induces systemic inflammation, oxidative stress, and autonomic dysfunction, further destabilizing cardiac electrophysiology. 29 , 30 This combined “double hit” may synergistically lower the arrhythmic threshold and promote the initiation of ventricular fibrillation or tachycardia. This interpretation is further supported by subgroup analyses stratified by arrest cause. The synergistic effect between cold spells and NO2 was more pronounced in cases classified as “other causes” than in those of cardiovascular origin. This pattern may reflect nonischemic pathways, such as respiratory insufficiency induced by NO2 and cold stress. Gaseous pollutants like NO2 can impair pulmonary function, and cold‐induced vasoconstriction may exacerbate hypoxia, precipitating cardiac arrest in individuals with preexisting respiratory conditions or general vulnerability. 31 , 32 In contrast, nonshockable rhythms (asystole and pulseless electrical activity) usually arise from progressive metabolic failure, severe myocardial ischemia, or global circulatory collapse. 26 Although cold and NO2 may contribute to overall clinical deterioration, their synergistic impact on acute electrical disruption is unlikely to represent the primary mechanism in these cases. 31 Additionally, the observed rhythm‐specific association may be shaped by prehospital response dynamics. Witnessed OHCA is often associated with shorter collapse‐to‐EMS arrival time, reducing the risk of rhythm deterioration; unwitnessed cases typically involve longer delays and a higher likelihood of converting from shockable to nonshockable rhythms. If the cold‐NO2 synergy primarily targets initially shockable rhythms, the true association may be more pronounced in witnessed cases but partially masked in unwitnessed cases due to rhythm conversion. This suggests that the current estimate of the synergistic effect for shockable rhythms could be conservative, as misclassification from delayed assessment may dilute the true signal. However, without data on witnessed status, this hypothesis cannot be verified—an important consideration for interpreting the findings. Nevertheless, given the ecological nature of exposure data, these mechanistic inferences remain speculative. Future studies incorporating individual‐level exposure monitoring, continuous cardiac rhythm assessment, detailed pathogenetic classification, and prehospital response timelines (eg, collapse time, call time, arrival time) and witnessed status are warranted to clarify the electrophysiological pathways underlying this compound environmental hazard and disentangle the confounding effects of prehospital delays. If corroborated by future research, the observed rhythm‐specific vulnerability has substantial clinical and policy relevance. Because shockable rhythms (ventricular fibrillation and pulseless ventricular tachycardia) are potentially reversible with early defibrillation and account for a substantial proportion of out‐of‐hospital cardiac arrests with the highest survival rates when treated in a timely manner. 26 Identifying environmental triggers that precipitate these arrhythmias is therefore critical for optimizing prehospital emergency responses. This highlights the urgent need for early‐warning systems that integrate low‐temperature forecasts and elevated NO2 levels (>25 μg/m3), 13 , 18 using existing monitoring networks. This study propose establishing a dedicated cross‐agency early‐warning and response mechanism for such compound exposures in subtropical megacities. During periods forecast as “compound high‐risk days” (eg, apparent temperature ≤2.5th percentile for ≥2 consecutive days with concurrent NO2 >25 μg/m3), a coordinated public health response should be activated, including the following. (1) Prehospital resource mobilization–dynamically expanding emergency medical service capacity and issuing public health advisories to high‐risk groups through official channels. (2) AED network optimization–ensuring ready access to automated external defibrillators in high‐risk, high‐traffic locations (eg, transit hubs, outdoor plazas, construction sites) and exploring deployment on mobile platforms such as public buses. Integrating a real‐time “AED map” with emergency dispatch systems would enable rapid retrieval. (3) Targeted training–Intensifying cardiopulmonary resuscitation and AED training for high‐exposure occupational groups (eg, transit workers, police officers, sanitation staff) before high‐risk seasons. Additionally, public health education should emphasize cold adaptation for subtropical populations (eg, avoiding sudden cold exposure, improving indoor thermal comfort) to mitigate physiological vulnerability. This integrated “forewarning–resource allocation–public mobilization” strategy translates epidemiological evidence into practical interventions to reduce the OHCA burden attributable to compound environmental exposures. The stratified analyses revealed variations in susceptibility to the cold‐NO2 synergy across population subgroups. The elevated relative excess risk observed among men and individuals <65 years may be attributable to higher baseline exposure, potentially reflecting greater time spent outdoors or in transit due to occupational or daily activities. 7 , 33 This vulnerability pattern underscores the need for targeted public health measures. During forecast high‐risk periods, practical interventions such as providing heated shelters at construction sites and subsidizing high‐efficiency respirators could mitigate dual exposure for these higher‐risk groups. At the urban scale, these findings reinforce the rationale for dynamic low‐emission zones, which restrict high‐polluting traffic during forecasted cold spells. Although cities like Beijing and Shenzhen have implemented similar zones targeting PM, 34 , 35 integrating a specific cold‐weather trigger could simultaneously reduce NO2 peaks and protect the most vulnerable populations. Aligning emission control strategies with compound‐event forecasts represents a dual‐benefit intervention within existing climate‐adaptation frameworks. Several limitations should be considered. First, exposure assessment relied on fixed‐site monitoring data, which may not capture individual‐level variations from indoor sources, commuting patterns, or microscale environmental gradients, potentially introducing ecological bias. Moreover, the ecological nature of the exposure data limits mechanistic insights. Future studies incorporating land‐use regression models or personal sensors, along with ambulatory ECG monitoring or wearable devices, could improve exposure accuracy and help clarify the physiological pathways linking coexposure to OHCA. Second, the low frequency of extreme cold events in this subtropical region limited statistical power for subgroup and interaction analyses; pooled multi‐city studies could enhance detectability. Third, the lack of collapse‐to‐EMS arrival time and witnessed status data is notable. Rhythms recorded by EMS may not reflect initial rhythms, as shockable rhythms can deteriorate over time; such misclassification would bias rhythm‐specific associations toward the null, potentially underestimating the true synergistic effect in shockable rhythms. Moreover, without witnessed status, the variation of cold–NO2 synergy by response time or collapse circumstances could not be examined. Future studies should incorporate detailed EMS timelines and witnessed status to disentangle the effects of prehospital delays and rhythm conversion. Fourth, despite pathogenetic stratification, detailed subclassification of cardiovascular conditions was precluded. Future studies integrating individual‐level data on preexisting cardiovascular diseases (eg, hypertension, coronary artery disease) could clarify their modifying role in the cold–NO2 synergy, facilitating targeted risk identification. Finally, the extreme cold thresholds derived from a subtropical city may not be directly generalizable to colder regions, where acclimatization, heating practices, and vulnerability differ. CONCLUSIONS In conclusion, the analysis of 16 769 OHCA cases in subtropical Shenzhen provides robust evidence of a synergistic effect between extreme cold spells and NO2 pollution. Notably, approximately one quarter of the cold‐related OHCA burden was attributable to this interaction, which was predominantly observed in cases with shockable cardiac rhythms—a rhythm reversible with prompt defibrillation. These findings directly support the development of compound‐event early‐warning systems that integrate low‐temperature and high‐NO2 forecasts to activate prehospital surge capacity and targeted emission control measures, thereby addressing a critical gap in cardiovascular protection against compound environmental hazards. Sources of Funding This work was supported by the Xi'an Association for Science and Technology 2025 Young Talent Support Program Project (0959202513189), the Xi'an No.1 hospital research fund project (No. 2025005), the Science and Technology Planning Project of Xi'an (25YXYJYB00028) and the Youth Program of the Health Commission of Xi'an (2026qn14). Disclosures None. Supporting information Tables S1–S28 Figures S1–S6 STROBE