Impacts of viral respiratory infections on segments of fatal out-of-hospital cardiac arrests

Abstract

Purpose of the study

Viral respiratory infections have been linked to fatal out-of-hospital cardiac arrest (OHCA), yet the specific causes remain unclear, and the impact of individual viral infections is often confounded by local meteorological and environmental factors. This study aimed to investigate the independent effects of prevalent viral respiratory infections on the risk and causes of fatal OHCA in different age groups.

Study design

We conducted negative binomial regression analyses to investigate the association between influenza, respiratory syncytial virus (RSV), and coronavirus disease 2019 (COVID-19) infections, along with temperature, extreme weather alerts, and the air quality health index, with age- and cause-specific fatal out-of-hospital cardiac arrest (OHCA) from the 2nd week of 2014 to the 17th week of 2020. The analysis covered three etiological categories (cardiovascular, respiratory, and non-cardiovascular non-respiratory) across three age groups (≤ 64 years, 65–84 years, ≥ 85 years) in Hong Kong.

Results

During this period, there were 41 548 fatal OHCA cases in Hong Kong. Influenza was consistently associated with fatal OHCA across all etiologies and age groups, significantly impacting cardiovascular and non-cardiovascular non-respiratory causes more than respiratory causes (110.9, 66.8, and 17.4 per 1 million persons, respectively). Reduced healthcare-seeking behaviors during the COVID-19 pandemic was linked to increased fatal OHCA across all ages and etiologies, except for respiratory causes. RSV showed no association with fatal OHCA in our population.

Conclusions

Influenza is a significant independent risk factor for fatal OHCA across variouscauses and age groups, particularly affecting cardiovascular and non-cardiovascular non-respiratory outcomes.

Key messages

What is already known on this topic? 

• Epidemiological studies have shown that the incidence of out-of-hospital cardiac arrest (OHCA) tends to rise during winter months, often linked to seasonal respiratory virus infections.

• Despite improvements in resuscitation techniques and cardiovascular disease management, the overall incidence of OHCA has not decreased in the past decade, indicating that non-cardiovascular causes significantly contribute to the total incidence of OHCA

What this study adds? 

• This study establishes a clear independent association between influenza infection and fatal OHCA from both cardiovascular and non-cardiovascular causes, expanding the understanding of influenza’s impact beyond respiratory complications.

• Our study does not find a consistent relationship between RSV activity and OHCA, highlighting that not all seasonal viral respiratory infections have a similar impact on fatal OHCA.

• Our findings reveal that reduced visits to healthcare facilities can significantly increase the risk of non-respiratory OHCA, thereby emphasizing the need for strategies to encourage timely medical care during pandemic.

How this study might affect research, practice or policy? 

• This study highlights the importance of broadening prevention strategies beyond traditional cardiovascular disease prevention to include other risk factors for OHCA, such as influenza infection. By addressing a more comprehensive range of contributing factors, healthcare systems may more effectively reduce the incidence of fatal OHCA.

• The potential for influenza prevention to reduce fatal OHCA rates underscores the necessity for further investigation into the cost-effectiveness of these strategies. Policymakers and healthcare organizations should conduct comprehensive cost–benefit analyses to evaluate the economic impact of all preventive measures, ensuring that resources are allocated efficiently to optimize health outcomes.

• Our findings emphasize the crucial importance of maintaining healthcare access during pandemics to prevent delays in treatment for non-respiratory conditions, which could lead to an increase in fatal OHCA. This highlights the need for robust healthcare infrastructure and polices that guarantee timely medical care for all conditions, even amidst health crises.

Introduction

Out-of-hospital cardiac arrest (OHCA) is a significant public health challenge worldwide, with survival rates rarely exceeding 10% despite advances in emergency medical services [1–3]. Current prevention strategies primarily focus on optimizing cardiovascular risk factors and managing cardiovascular diseases. Despite these efforts, the decline in the incidence of OHCA has plateaued [4,5], suggesting the need to identify additional preventive approaches.

Population-based studies show consistent seasonal variations in OHCA, with peaks occurring during the winter months [6–11]. In addition to fluctuations in temperature and air pollution [6,8–10], seasonal outbreaks of viral respiratory infections likely contribute to this trend. Influenza epidemics are linked to increased cardiovascular hospitalizations and mortality [12,13], as well as with OHCA [11]. Respiratory syncytial virus (RSV) infection significantly impacts cardiovascular and respiratory deaths in pediatric and geriatric populations [13–15]. Recently, the coronavirus disease 2019 (COVID-19) pandemic has also been associated with a rise in OHCA [11]. However, the etiologies and thus the mechanisms behind these cases remain unclear. Additionally, viral respiratory infections are often tied to variations in local meteorological and environmental factors, leaving the direct contributions of specific infections to excess OHCA unknown. Hong Kong provides an ideal setting for investigating these relationships due to its comprehensive OHCA registry and robust infectious disease and environmental surveillance systems. In this study, we sought to investigate the independent effects of prevalent viral respiratory infections on the cause- and age-specific fatal OHCA.

Methods

Study design

We conducted a population-based study on all fatal OHCA cases in Hong Kong (population 7.5 million) from the 2nd week of 2014 (Jan 5, 2014) to the 17th week of 2020 (April 25, 2020). The study was approved by the institutional review boards of the Hospital Authority and the Department of Health. In Hong Kong, the Forensic Pathology Service of the Department of Health provides forensic services for all out-of-hospital or emergency room deaths. Patient demographics, time, date, circumstances of discovery, autopsy findings, and causes of death were prospectively recorded in an electronic database. All fatal OHCA cases within this period, except those with obvious unnatural causes, were included. We classified the study population into three age groups (≤64 years, 65–84 years, ≥85 years) and categorized principal causes of death according to the International Classification of Diseases, Tenth Revision (ICD-10) into cardiovascular (OHCA-CV), respiratory (OHCA-Resp), and non-cardiovascular non-respiratory (OHCA-Others). Details of the classification are summarized in Supplemental Table S1.

Influenza and RSV activity proxies

We estimated proxies for age-specific influenza and RSV activity from the following sources: 1) laboratory surveillance data on influenza and RSV infections; [16] 2) proportions of consultations for influenza-like illness in General Out-Patient Clinics; 3) weekly proportions of public hospital admissions with a principal diagnosis of influenza; and 4) outbreaks in institutions other than schools [17]. Methodological details are summarized in Supplemental Materials.

Local COVID-19 cases and reduced healthcare-seeking behavior

We obtained the weekly number of locally contracted COVID-19 cases from the Center for Health Protection in Hong Kong [18]. Despite a low number of locally contracted COVID-19 cases (n = 442) in our study period, we observed a marked reduction in the utilization of emergency departments and outpatient clinics. Therefore, we included a proxy for reduced healthcare-seeking behavior, defined as the percentage of people avoiding visiting healthcare facilities during the COVID-19 outbreak period, i.e. 4^th to 7^th weeks of 2020, compared with the baseline (the 3rd week of 2020) [19]. Methodological details are summarized in Supplemental Materials.

Weather data and pollution data

We obtained weather data from the Hong Kong Observatory [20]. We calculated the weekly average temperature from daily mean temperatures. In addition, we collected data on cold weather warning, very hot weather warning, tropical cyclone warnings of No. 8 or above, and rainstorm warnings (abbreviated as cold warning, very hot warning, typhoon warning, and rainstorm warning hereafter) [20]. We then calculated the number of days in a week when these weather warnings were issued.

We downloaded the Air Quality Health Index (AQHI) from the Environmental Protection Department of Hong Kong [21]. The AQHI is reported on a scale from 1 to 10 and “10+”. We used a score of 12 to replace “10+” to calculate the weekly average AQHI. Since the proportion of days with an AQHI of “10+” is less than 0.25% per year, the replacement should hardly affect our analysis.

Population data

We obtained mid-year population data by age group from the Census and Statistics Department of Hong Kong for the years 2014 to 2020 [22].

Statistical analysis

We first calculated Pearson’s correlation coefficients between each pair of the explanatory variables. Based on these coefficients, we divided the explanatory variables into four groups: influenza and RSV activity proxies; COVID-19 activity proxy and reduced healthcare-seeking behavior proxy; weather and pollution variables for selection; weather variables constantly included in the model (Supplemental Table S2). Since explanatory variable subgroups within the same group were highly correlated, we used only one subgroup from each group in each model. Additionally, we considered the possibility of no lag effect, a 1-week lag, or a 2-week lag between the explanatory variables and the outcome. We allowed different lag effects for respiratory infection proxies and weather and pollution variables. As the reduced healthcare-seeking behavior proxy was primarily used to adjust for the effect of the number of local COVID-19 cases, we assumed its lag time was the same as that of the respiratory infection proxies. All possible combinations of explanatory variables were generated by including one variable subgroup from each group and considering different assumptions of lag effect.

To account for over-dispersion, we fitted a negative binomial model separately for each age- and cause-specific outcome. We utilized one set of explanatory variable combinations for each model. The model used is summarized below:

where Y_t represents the number of deaths in week t, β₀ is the intercept term, tw and tr account for the time of the lag effect of weather and pollution variables and viral respiratory infection proxies, respectively, β₁, β₂ and β₃ represent coefficients associated with the mean temperature, cold warning and very hot warning or typhoon warning in week tw, β_j represents the coefficient associated with the selected weather and pollution variables in Group 3 in week tw, β_i represents the coefficient associated with the viral respiratory infection proxies and the reduced healthcare-seeking behavior proxy in week tr, and log(Pop_t) is an offset term using the mid-year population. We consistently included one subgroup of variables from Group 4 in the model and conducted backward selection for the other variables based on the Akaike Information Criterion (AIC). For each outcome, AIC differences of each model m (ΔAIC_m), defined as the difference between AIC_m and the lowest AIC among all the models, was calculated. Following Burnham and Anderson [23], models with ΔAIC_m ≤ 10 were first selected as the best-fitting model candidates for each outcome. We then identified the viral respiratory infection proxy that was most frequently included in all the best-fitting model candidates across all outcomes. For each outcome, the model with the lowest AIC that included this most frequent viral respiratory infection proxy was selected as the best-fitting prediction model. If none of the models included the most frequent viral respiratory infection virus proxy, the model with the lowest AIC was directly selected as the best-fitting model for that outcome.

We then calculated the relative risk (RR) and 95% confidence intervals (95% CI) for the predicted numbers, accounting for uncertainty from our model parameter estimates and the negative binomial distribution (referred to as negative binomial 95% CI).

Results

The A total of 49 629 cases of fatal OHCA were identified from the 2nd week of 2014 to the 17th week of 2020. After excluding 8069 unnatural deaths and 12 stillbirths, 41 548 cases were retained for analysis. The age and cause distributions of fatal OHCA during the study period are summarized in Fig. 1. The age- and cause-specific incidences of fatal OHCA are presented in Table 1. Weekly variations in weather and pollution variables, as well as viral respiratory infection proxies, are shown in Supplemental Figs. S1 and S2, respectively.

Influenza activity and fatal OHCA

The weekly incidences of fatal OHCA observed and predicted by the fitted models are presented in Fig. 2. Similar temporal variations in the number of fatal OHCA were observed in regardless of etiology or age. The effects of the explanatory variables on age- and cause-specific fatal OHCA are summarized in Table 2. Influenza activity was closely associated with the incidences of fatal OHCA of all etiologies. Every 1 per 1000 increase in consultations due to influenza was associated with a 6.4% (4.1% - 8.8%) increase in the incidence of OHCA-CV and a 4.7% (1.8% - 7.7%) OHCA-Others in the same week; and a 4.7% (0.1% - 9.6%) increase in the incidence of OHCA-Resp after 2 weeks. Furthermore, influenza activity exerted similar effects on the incidence of fatal OHCA across all age groups. Every 1 per 1000 increase in consultations due to influenza was associated with a 3.6% (1.0% - 6.4%) increase in the incidence of OHCA in individuals aged ≤64 years, 5.3% (2.7% - 8.0%) in those aged 65–84 years and 4.9% (1.8% - 8.0%) increase in those aged ≥85 years.

RSV activity and fatal OHCA

Compare to influenza, the effects of RSV infection on fatal OHCA were less consistent. As shown in Table 2, every 1 per 10 000 consultations due to RSV was associated with a 4.8% (0.8%–8.9%) increase in the incidence of OHCA-Resp after 2 weeks, and a 3.8% (1.6%–5.9%) decrease in the incidence of OHCA-Others in the same week. Overall, RSV activity was not an independent risk factor for fatal OHCA in our population.

COVID-19 and OHCA

There were only 442 locally contracted COVID-19 cases in Hong Kong during our study period. During this time, COVID-19 infection itself was negatively associated with OHCA-Resp, with case COVID-19 case be followed by a 0.4% (0.1–0.9%) reduction in its incidence after 2 weeks. However, reduced healthcare-seeking behavior during the pandemic was positively associated with OHCA-CV (RR 1.009, 95% CI 1.005–1.012) and OHCA-Others (RR 1.020, 95% CI 1.017–1.024), and fatal OHCA across all age groups (age ≤ 64 years: RR 1.011, 95% CI 1.006–1.016; age 65–84 years: RR 1.007, 95% CI 1.003–1.010; age ≥ 85: RR 1.011, 95% CI 1.007–1.016, Table 2).

Sensitivity analysis

Recognizing influenza as a significant viral pathogen causing cardiopulmonary hospitalizations, we conducted a sensitivity analysis by substituting the chosen influenza activity proxies with influenza hospitalization (denoted as Flu hospitalization hereafter) proxies (Supplemental Table S3). The most frequent Flu hospitalization proxy selected in the sensitivity analysis was Flu hospitalization (≥65 years), which was included in sensitivity best-fitting models for fatal OHCAs of all causes and at all ages. The only exception was fatal OHCA in those aged ≤64 years, in which Flu hospitalization (≤64 years) was selected in the best-fitting model. In the sensitivity analysis, influenza activity remains significantly associated with OHCA of all causes and at all ages. In addition, the number of COVID-19 cases and reduced healthcare-seeking behavior retained in the best-fitting models in the same manner as in the main analysis.

Influenza-associated excess fatal OHCA

We estimated weekly (Fig. 3) and yearly (Fig. 4) rates of influenza-associated excess fatal OHCA of all etiologies and at all ages (methodological details are summarized in Supplemental materials). The rates for the complete years 2015, 2016, 2017, 2018, and 2019 were 45.4 (37.6–53.8), 35.7 (29.3–42.7), 34.4 (28.0–41.1), 27.7 (22.3–33.4) and 16.5 (12.8–20.5) per 1 million persons, respectively. Influenza-associated excess fatal OHCA increased markedly with age. Additionally, influenza contributed more to OHCA-CV [110.9 (98.4–124.0) per 1 million persons] and OHCA-Others [66.8 (58.7–75.3) per 1 million persons] than OHCA-Resp (17.4 [14.1–20.8] per 1 million persons) in the study period.

Discussion

Our study found that influenza infection is significantly associated with fatal OHCA across all etiologies and age groups, independent of seasonal meteorological factors. Although primarily a respiratory virus, most influenza-associated fatal OHCA cases were due to cardiovascular and non-cardiovascular non-respiratory causes. Additionally, during periods of controlled COVID-19, reduced healthcare-seeking behavior increased the risk of non-respiratory OHCA. In contrast, RSV activity showed inconsistent effects on fatal OHCA.

Previous study linked severe influenza epidemics to OHCA spikes [7,11], but concurrent viral infections and weather changes complicate estimation of influenza’s true impact. Our study confirms the strong link between influenza activity and fatal OHCA, independent of weather, pollution, RSV, or COVID-19. While earlier research suggested influenza increases OHCA by triggering cardiovascular events in high-risk groups, our findings indicate it also increases fatal OHCA from non-cardiovascular non-respiratory causes. This aligns with previous studies showing influenza is associated with cardiac failure, myocardial infarction, kidney failure, liver failure, encephalopathy, rhabdomyolysis, shock, coagulopathy, multiorgan failure, and death [24].

Although traditionally seen as a respiratory virus, influenza’s damage extends beyond the lungs. Not only has research found viral presence in multiple organs, causing direct cytopathic effects, an uncontrolled systemic inflammatory response in susceptible individuals can trigger a cytokine storm, damaging organs beyond the infection site [24]. Additionally, patients and healthcare providers tend to view influenza as a mild illness, thus overlooking symptoms and delaying the seeking of medical advice and the provision of appropriate treatment, further increasing the risk of fatal OHCA [7].

Current strategies to reduce fatal OHCA emphasize the prevention and treatment of cardiovascular diseases. However, non-cardiovascular causes account for up to 60% of all OHCA cases [25]. Our results suggest that preventing influenza infection may reduce the incidence of fatal OHCA, particularly from non-cardiovascular causes. Interestingly, we observed a decline in influenza-related fatal OHCA from 2015 to 2019, coinciding with increased influenza vaccination rates in Hong Kong [26,27]. Yet, a recent systematic review showed that overall influenza vaccination uptake in Asia remains low, with a median rate of 37.3%, even among high-risk groups [28]. Barriers to vaccine uptake include a low perceived risk of the disease and a lack of recommendations from healthcare workers [28]. Our findings provide important information for healthcare workers and patients to make informed choices about influenza vaccination. Future studies should assess the cost-effectiveness of expanding influenza vaccination coverage to achieve herd immunity in preventing fatal OHCA.

RSV is a significant cause of morbidity and mortality in pediatric and geriatric populations [13,14]. However, our results showed that RSV activity increased fatal OHCA from respiratory causes but unexpectedly decreased fatal OHCA from other causes, neutralizing its overall effect. We propose that RSV likely impacts mortality in patients with poor health conditions who are at high risk of OHCA. When these patients succumb to respiratory conditions, it reduces the rate of OHCA from other causes. Overall, our findings suggest that the correlation between RSV activity and OHCA is weak, and that not all seasonal viral respiratory infections have a similar impact on fatal OHCA.

Recent studies have shown an increased incidence of mortality from OHCA during the COVID-19 outbreak [29]. In Hong Kong, widespread use of face masks, rigorous contact tracing, and social distancing measures limited the spread of COVID-19 at the early stage [19]. The low number of local COVID-19 cases during our study period may explain the lack of association between COVID-19 activity and fatal OHCA. Additionally, these non-pharmacological measures also reduced the transmission of seasonal viral respiratory infections, including influenza [19], which may account for their “protective effect” against fatal OHCA from respiratory causes. However, fear of COVID-19 infection discouraged visits to healthcare facilities, especially for non-respiratory symptoms [19]. Despite a low number of local COVID-19 cases during our study period, reduced healthcare-seeking behavior increased fatal OHCA across all ages, primarily due to non-respiratory causes. These findings have significant implications for managing future pandemics. Transmissible diseases impact public health not only through their pathogenicity but also by diverting resources from other health conditions and discouraging visits to healthcare facilities. Clear guidelines for triaging patients to home care, designated clinics, and hospitalization, along with increased telemedicine use for other health issues, significantly eased the strain on our healthcare system later in the outbreak. A well-organized plan also reduces patient fear and encourages them to seek help when needed. Our results highlight the necessity of pairing strict measures to prevent transmissible diseases with a well-planned healthcare system that promotes efficient and flexible resource utilization during a pandemic.

Our study has several limitations. First, we focused on the impact of viral respiratory infections on OHCA that did not survive resuscitation. However, with a survival rate of ≤5% in our area [30], and evidence that respiratory infections further decrease OHCA survival by 18–22% [7], including the small survivor cohort is unlikely to alter population-level risk estimates. Thus, our focus on fatal OHCAs is appropriate for assessing mortality prevention strategies from a public health perspective. Second, like other ecological studies, individual-level data on comorbidities are unavailable. However, we used a validated excessive mortality model to assess influenza’s impact by comparing observed mortalityto baseline levels and quantifying lagged effects through distributed lag modeling. This provides a conservative upper-bound estimate of influenza’s impact, aligning with our focus on public health implications. The robustness of our findings is supported by sustained effects after age adjustments and sensitivity analyses. Thus, while ecological designs preclude individual risk inference, they provide critical population-level insights, underscoring the need for complementary individual-level studies. Third, estimates of influenza-like illness activity were based on sentinel reports, and the impact of asymptomatic infections on fatal OHCA was not addressed. Fourth, the absence of comprehensive population-based influenza vaccination data in our locality during the study period prevented analysis of its effect on OHCA incidence, which could have enriched the findings of this study.

Conclusion

Our study demonstrates an independent association between influenza infection and fatal OHCA from both cardiovascular and non-cardiovascular causes, underscoring the need for enhanced influenza prevention efforts. Additionally, our findings suggest that reduced healthcare-seeking behavior during pandemics increases the risk of non-respiratory OHCA, highlighting the importance of a well-organized healthcare response. These insights emphasize the critical role of comprehensive public health strategies in preventing OHCA and ensuring better health outcomes during outbreaks of viral respiratory infections.

Conflict of interest statement

None declared.

Funding

This study is financially supported by Sun Chieh Yeh Heart Foundation. Funder has no role in the conduct of the study.

Data availability

The data underlying this article were provided by Department of Health, Hong Kong by permission. Data will be shared on request to the corresponding author with permission of the Department of Health, Hong Kong.

References

Author notes