Factors associated with recurrent out-of-hospital cardiac arrest after hospital discharge: a population-based study

Graphical abstract

Highlights

• •Recurrent OHCA occurred in 7.2 % of first-time OHCA survivors over five years.
• •Device dependence is an independent risk factor for recurrent OHCA.
• •NG tube and Foley catheter use predict increased recurrence risk.
• •Traditional comorbidities lose significance when adjusting for device use.

Abstract

Background

Survivors of out-of-hospital cardiac arrest (OHCA) remain at significant risk of recurrence; however, the factors associated with a subsequent OHCA episode requiring hospitalisation are not well characterised. Dependence on indwelling devices may indicate a state of chronic vulnerability. This study aimed to identify risk factors for recurrent OHCA using a nationwide cohort.

Methods

We conducted a retrospective cohort study using Taiwan’s National Health Insurance Research Database from 2010 to 2020. Adult non-traumatic OHCA survivors discharged alive were included and followed for up to five years. Demographics, comorbidities, healthcare utilisation, and long-term indwelling device use (nasogastric tube, Foley catheter, tracheostomy with ventilator) were analysed. The primary outcome was a second OHCA requiring hospitalisation. Multivariable logistic regression identified independent risk factors, with adjusted odds ratios (aORs) and 95 % confidence intervals (CIs) reported.

Results

Among 15,617 OHCA survivors, 1,123 (7.2 %) experienced a second OHCA requiring hospitalisation within five years. Patients with recurrence were older (62.6 vs. 61.0 years, p < 0.001), and more frequently used indwelling devices. In multivariable analysis, nasogastric tube (aOR 1.82, 95 % CI 1.50–2.19, p < 0.001), Foley catheter (aOR 1.20, 95 % CI 1.00–1.43, p = 0.05), and tracheostomy with ventilator (aOR 0.85, 95 % CI 0.73–1.00, p = 0.043) were independently associated with recurrent OHCA, while age, sex, and pre-arrest diseases were not.

Conclusion

Functional dependence and healthcare utilisation were associated with recurrent OHCA among survivors hospitalised for a recurrent event. Device dependence may serve as a marker of vulnerability, highlighting the need for ongoing monitoring and follow-up.

Introduction

Out-of-hospital cardiac arrest (OHCA) remains a significant public health burden, with an annual incidence of approximately 41–55 emergency medical services (EMS) – treated cases per 100,000 person-years globally and persistently low survival rates.1, 2, 3, 4 Despite advances in emergency response systems and post-resuscitation care, only about 10 % of OHCA patients survive to hospital discharge.5, 6 While improvements in the chain of survival have contributed to an increasing number of OHCA survivors, emerging data indicate that 6–18 % of these individuals experience recurrent cardiac arrest after discharge, highlighting the need for better risk stratification and post-cardiac arrest care.7, 8, 9, 10

Previous studies have primarily focused on risk factors associated with survival and short-term neurological outcomes following an index OHCA. Although some investigations have identified associations to coronary artery disease, chronic heart failure, and the initial arrest rhythm, the long-term vulnerability of OHCA survivors—particularly the risk of experiencing a second cardiac arrest—remains poorly investigated.11, 12, 13, 14

Recognising this gap, the American Heart Association (AHA) in 2020 introduced a sixth link to the chain of survival: recovery.¹⁵ This update reflects a paradigm shift toward comprehensive post-resuscitation care, emphasising not only acute-phase interventions but also longitudinal monitoring and support. It highlights the need to examine the entire post-discharge trajectory, including patients’ functional capacity, level of care dependency, and ongoing interaction with the healthcare system.¹⁶ The use of life-sustaining indwelling devices—such as nasogastric tubes, Foley catheters, or tracheostomies with mechanical ventilation—may serve as tangible indicators of persistent neurological or physiological compromise. Their presence could potentially signify a higher baseline vulnerability, making them relevant markers for identifying patients at increased risk for recurrent OHCA.

No large-scale cohort study has investigated the risk factors associated with a second OHCA among OHCA survivors of an index arrest. Addressing this knowledge gap is essential for developing targeted interventions to improve long-term outcomes in the high-risk population. This study investigates the risks of recurrent OHCA requiring hospitalisation by utilising a nationwide, population-based database.

Material and methods

This retrospective study utilised data from Taiwan’s National Health Insurance Research Database (NHIRD), which was established in 1995 alongside the implementation of Taiwan’s single-payer National Health Insurance (NHI) programme. The NHIRD contains claims data for over 23 million beneficiaries, covering more than 99 % of the Taiwanese population.¹⁷ It includes registration details, outpatient and inpatient visits, prescriptions, disease diagnoses, and various healthcare-related variables, providing a robust foundation for population-based research.¹⁸ Over the decades, the comprehensiveness of this database has facilitated numerous epidemiological and clinical investigations, including studies on healthcare utilisation, disease aetiology, and treatment outcomes. To enhance its utility, the NHIRD can be linked to disease-specific registries and other governmental datasets, significantly expanding the potential for real-world evidence generation.¹⁹ Meanwhile, stringent de-identification protocols and ethical review processes ensure patient privacy and data security. Given its extensive coverage and linkage capabilities, the NHIRD supports large-scale longitudinal research, offering robust and comprehensive data for addressing key questions in global public health, clinical practise, and policy development.

This study was approved by the Institutional Review Board of National Taiwan University Hospital (Approval No. 202502001RINC), which waived the requirement for informed consent due to the retrospective design and use of anonymized data. All procedures were conducted in strict accordance with ethical standards and regulatory requirements, including the Declaration of Helsinki.²⁰

Patients: Inclusion and exclusion

Patients diagnosed with OHCA between 2010 and 2020 were identified from the NHIRD. Inclusion criteria required patients to have an Emergency Department (ED) triage level of 1 and at least one relevant International Classification of Diseases (ICD) codes: ICD-9 (427.4, 427.5, 798, 798.1, 798.2, 798.9, 799.9) or ICD-10 (I46, I46.2, I46.8, I46.9, I49.0). Both ICD-9 and ICD-10 codes were utilised because, after 2016, ICD-9 was progressively replaced by ICD-10. Patients were excluded if their OHCA was trauma-related (Supplementary Table 1), if they were younger than 18 years of age, or if their recorded date of death preceded the OHCA event. These criteria ensured the study population comprised solely non-traumatic OHCA cases and allowed for meaningful evaluation of recurrent OHCA events.

Variables

To investigate the risk factors associated with a second episode of OHCA in patients who survived an index OHCA and were subsequently discharged, we categorised variables into four domains: (1) demographic and epidemiological factors, (2) comorbidities, (3) healthcare utilisation, and (4) daily functional status. Demographic and epidemiological factors included sex and age. Comorbidities included 15 specific conditions—diabetes mellitus, hypertension, hepatitis, chronic kidney disease, asthma, peptic ulcer disease, heart disease, cancer, cerebrovascular disease, hyperlipidemia, chronic obstructive pulmonary disease, and liver cirrhosis—each identified annually through a review of each enrollee’s National Health Insurance prescription records. Healthcare utilisation variables included any ICU admission within five years after discharge, renal dialysis lasting more than three consecutive months, at least three outpatient visits within five years, and at least two emergency department or hospitalisation events. The number of outpatient and ED visits was included based on prior literature suggesting that increased healthcare utilisation may reflect clinical instability preceding adverse cardiac events. A recent population-based study from Taiwan found a marked increase in outpatient and emergency visits in the weeks leading up to sudden cardiac death, supporting their role as potential predictors.²¹

Importantly, given that these individuals had survived OHCA and likely experienced prolonged cerebral hypoxia, daily functional status was defined by the long-term need for a nasogastric tube, Foley catheterization, or the presence of a tracheostomy with prolonged mechanical ventilator support. These devices were identified based on post-discharge claims data, and they are most often placed during the index OHCA hospitalisation and persist after discharge. This approach allowed us to use device presence as a surrogate marker of post-acute functional dependency. This framework enabled a systematic evaluation of the multifactorial indices of recurrent OHCA.

Outcome

The primary outcome of this study was the occurrence of a second OHCA that resulted in the patient being brought to the hospital (i.e., hospitalisation). We applied the same criteria used to define the index OHCA event: each episode had to simultaneously meet the highest ED triage level (Triage Level 1) and correspond to the relevant ICD codes for cardiac arrest. Follow-up began from the date of hospital discharge following the index OHCA and continued for up to five years, or until the occurrence of a second OHCA or death—whichever occurred first.

Statistical analysis

In the statistical analysis, continuous variables were compared via Student’s t-test, and categorical variables were analysed using the chi-square test. A univariate logistic regression model was initially employed to determine potential factors of recurrent OHCA, and variables yielding a p-value below 0.1 were included in a multivariable logistic regression model for further adjustment. Before constructing the final model, collinearity among variables was checked. Because ICU admission and hospitalisation were found to frequently co-occur with ventilator use, these factors were excluded from the multivariable regression to reduce the impact of collinearity. We then applied a stepwise selection method (combining forward and backward procedures) based on the Akaike Information Criterion (AIC) to determine the final multivariable model. This iterative process retained variables that improved overall model fit and excluded those that did not contribute meaningfully. The final model includes all variables retained after stepwise selection, regardless of their individual statistical significance. The results are presented as adjusted odds ratios (aORs) with 95 % confidence intervals (CIs), thereby illustrating the risk of a second OHCA event among survivors. All statistical procedures were performed by an independent statistician using SAS software (version 9.4, SAS Institute Inc., Cary, NC, USA), and statistical significance was defined as p < 0.05.

Results

Between 2010 and 2020, a total of 239,929 OHCAs were identified from the NHIRD. After excluding 6,568 cases with trauma-related causes, 4,167 cases involving individuals aged under 18 years, and 147 cases with mismatched data, 229,047 eligible OHCAs remained. Of these, 213,430 patients who died during the index hospitalisation were excluded. The final analytic cohort consisted of 15,617 index OHCA survivors. During the five-year follow-up period, 1,123 patients (7.2 %) experienced a second OHCA event, whereas 14,494 patients (92.8 %) did not (Fig. 1). Among patients who experienced a recurrent OHCA during follow-up, the median time from hospital discharge to second OHCA was 55 days.

Fig. 1.

This flow diagram illustrates the study population selection process from an initial pool of 239,929 OHCA events. After applying exclusion criteria and accounting for mortality, 15,617 index OHCA survivors were analysed and stratified based on the recurrence of OHCA.

Table 1 presents the comparison of characteristics between the OHCA survivors with and without a second OHCA event. Patients with a second OHCA were older compared to those without recurrence (62.63 vs. 60.95 years, p < 0.001), and a slightly higher proportion were male (62.24 % vs. 59.08 %, p = 0.04). Patients with a second OHCA had a significantly higher rate of ICU admission within the previous year (60.55 % vs. 33.79 %, p < 0.001) and more frequent ED or ward hospitalisations more than two times (74.35 % vs. 50.72 %, p < 0.001). Hemodialysis use was also more prevalent among patients with a second OHCA (6.68 % vs. 5.19 %, p = 0.03). No significant difference was observed in outpatient clinic visits between groups. The use of life-sustaining indwelling devices was notably higher in patients with a second OHCA, including nasogastric tube placement (63.22 % vs. 46.58 %, p < 0.001), Foley catheterization (57.52 % vs. 43.74 %, p < 0.001), and tracheostomy with ventilator support (26.63 % vs. 21.97 %, p < 0.001). In terms of comorbidities, the second OHCA group had higher rates of diabetes mellitus (24.31 % vs. 20.22 %, p = 0.001), hypertension (32.15 % vs. 28.75 %, p = 0.02), chronic kidney disease (11.93 % vs. 9.57 %, p = 0.01), and heart disease (22.08 % vs. 18.88 %, p = 0.009). No significant differences between groups were found in the other comorbidities, including hepatitis, asthma, peptic ulcer disease, malignancy, cerebrovascular disease, hyperlipidemia, chronic obstructive pulmonary disease and cirrhosis.

Table 1.

Comparison of characteristics between index OHCA survivors with and without a second OHCA event.

Variables	All OHCA survivors (N = 15,617)	Without a second OHCA (N = 14,494)	With a second OHCA (N = 1,123)	P-value
Demographics
Age	61.07 ± 17.28	60.95 ± 17.42	62.63 ± 15.08	<0.001
Sex, male	9262 (59.31)	8563 (59.08)	699 (62.24)	0.04
Utilisation of healthcare facilities
ICU admission*	5577 (35.71)	4897 (33.79)	680 (60.55)	<0.001
Hemodialysis	827 (5.3)	752 (5.19)	75 (6.68)	0.03
Outpatient clinics >3†	9375 (60.03)	8707 (60.07)	668 (59.48)	0.70
ED and/or ward hospitalisation >2†	8186 (52.42)	7351 (50.72)	835 (74.35)	<0.001
Life-sustaining indwelling catheter
Nasogastric tube	7462 (47.78)	6752 (46.58)	710 (63.22)	<0.001
Foley catheter	6985 (44.73)	6339 (43.74)	646 (57.52)	<0.001
Tracheostomy and ventilator support	3484 (22.31)	3185 (21.97)	299 (26.63)	<0.001
Pre-existing comorbidities
Diabetes mellitus	3204 (20.52)	2931 (20.22)	273 (24.31)	0.001
Hypertension	4528 (28.99)	4167 (28.75)	361 (32.15)	0.02
Hepatitis	432 (2.77)	393 (2.71)	39 (3.47)	0.13
Chronic kidney disease	1521 (9.74)	1387 (9.57)	134 (11.93)	0.01
Asthma	460 (2.95)	427 (2.95)	33 (2.94)	0.99
Peptic ulcer disease	1116 (7.15)	1027 (7.09)	89 (7.93)	0.29
Heart disease	2985 (19.11)	2737 (18.88)	248 (22.08)	0.009
Malignancy	545 (3.49)	506 (3.49)	39 (3.47)	0.97
Cerebrovascular disease	1340 (8.58)	1243 (8.58)	97 (8.64)	0.94
Hyperlipidemia	1874 (12)	1724 (11.89)	150 (13.36)	0.15
Chronic obstructive pulmonary disease	425 (2.72)	391 (2.7)	34 (3.03)	0.51
Liver cirrhosis	160 (1.02)	148 (1.02)	12 (1.07)	0.88

ED: Emergency department; ICU: intensive care unit; OHCA, out-of-hospital cardiac arrest.

Data are presented as mean ± standard deviation for continuous variables and n (%) for categorical variables.

^*In previous 365 days.

^†In previous 30 days.

Table 2 presents the factors associated with a second OHCA event among index OHCA survivors. Univariate logistic regression analysis identified older age, male sex, ICU admission, hemodialysis, more than two hospitalisations in the ED or general ward, use of life-sustaining indwelling devices (nasogastric tube, Foley catheter, or tracheostomy with ventilator support), diabetes mellitus, hypertension, chronic kidney disease, and heart disease as factors significantly associated with an increased risk of recurrent OHCA. In the multivariable analysis, after adjusting for potential confounders, nasogastric tube placement (aOR 1.82, 95 % CI 1.50–2.192, p < 0.001), Foley catheterization (aOR 1.20, 95 % CI 1.00–1.43, p = 0.05), and tracheostomy with ventilator support (aOR 0.85, 95 % CI 0.73–1.00, p = 0.04) remained independently associated with the risk of a second OHCA. In contrast, age, sex, and comorbidities—including diabetes mellitus, hypertension, chronic kidney disease, and heart disease—were not independently associated after adjustment.

Table 2.

Factors associated with a second OHCA event in index OHCA survivors.

Risk factors	OR	95 % CI	P-value	aOR	95 % CI	P-value
Demographics
Age	1.01	1.00–1.01	0.002	1.00	0.996–1.004	0.93
Sex, male	1.14	1.01–1.29	0.04	1.13	0.997–1.283	0.06
Utility of healthcare facilities
ICU admission*	3.01	2.66–3.41	<0.001
Hemodialysis	1.31	1.02–1.67	0.03	1.11	0.849–1.463	0.44
Outpatient clinics >3†	0.98	0.86–1.10	0.70
ED and/or ward hospitalisation >2†	2.82	2.45–3.23	<0.001
Life-sustaining indwelling catheter
Nasogastric tube	1.97	1.74–2.24	<0.001	1.82	1.503–2.192	<0.001
Foley catheter	1.74	1.54–1.97	<0.001	1.20	1.003–1.428	0.05
Tracheostomy and ventilator support	1.29	1.12–1.48	<0.001	0.85	0.729–0.995	0.04
Pre-existing comorbidities
Diabetes mellitus	1.27	1.10–1.46	0.001	1.13	0.956–1.324	0.16
Hypertension	1.17	1.03–1.34	0.02	1.01	0.868–1.172	0.91
Hepatitis	1.29	0.92–1.81	0.13
Chronic kidney disease	1.28	1.06–1.55	0.01	1.10	0.884–1.375	0.39
Asthma	1.00	0.70–1.43	0.99
Peptic ulcer disease	1.13	0.90–1.41	0.29
Heart disease	1.22	1.05–1.41	0.009	1.10	0.938–1.291	0.24
Malignancy	1.00	0.71–1.39	0.97
Cerebrovascular disease	1.01	0.81–1.25	0.94
Hyperlipidemia	1.14	0.96–1.37	0.15
Chronic obstructive pulmonary disease	1.13	0.79–1.61	0.51
Liver cirrhosis	1.05	0.58–1.89	0.88

Variables with a P-value <0.1 would be entered the multivariable logistic regression model.

aOR: adjusted odds ratio; CI: confidence interval; ED: Emergency department; ICU: intensive care unit; OHCA, out-of-hospital cardiac arrest; OR: odds ratio.

^*In previous 365 days.

^†In previous 30 days.

Fig. 2 illustrates the distribution of various combinations of indwelling catheters and ventilator use among patients with and without a second OHCA. Patients who experienced a second OHCA were significantly more likely to have received both a nasogastric tube and a Foley catheter (28.8 % vs. 20.3 %, p < 0.001), or all three devices—nasogastric tube, Foley catheter, and ventilator—combined (23.0 % vs. 18.2 %, p < 0.001). In contrast, patients without a second OHCA were more frequently found to have no indwelling catheters (47.8 % vs. 30.4 %, p < 0.001). The use of a nasogastric tube alone was also more common in the second OHCA group (8.8 % vs. 5.1 %, p < 0.001), while no significant differences were observed for Foley catheter alone (5.4 % vs. 4.9 %, p = 0.39), ventilator alone (0.6 % vs. 0.4 %, p = 0.34), nasogastric tube with ventilator (2.7 % vs. 3.0 %, p = 0.60), or Foley catheter with ventilator (3.6 % vs. 3.8 %, p = 0.90).

Fig. 2.

Bar chart comparing the prevalence of device combinations (NG tube, Foley catheter, and ventilator) among patients with and without a second OHCA.

Discussion

In this nationwide retrospective cohort study, we found that 7.2 % of index OHCA survivors experienced a second OHCA event within five years, capturing those recurrent events during hospitalisation. Notably, our analysis uniquely identified the need for life-sustaining indwelling devices—specifically nasogastric tubes, Foley catheters, and tracheostomy with ventilator support—as significant independent factors of recurrent OHCA, even after adjusting for demographic characteristics and comorbidities. Our findings align with the AHA and Advanced Cardiac Life Support guidelines, which emphasise “recovery” as the sixth link in the chain of survival. This recovery phase extends beyond hospital discharge and includes continuous follow-up, structured rehabilitation programmes, and long-term support services. By implementing tailored rehabilitation protocols and strengthening connexions to long-term care networks, clinicians may substantially improve the functional recovery and quality of life for OHCA survivors, thereby reducing the likelihood of recurrent cardiac arrest and enhancing overall survival outcomes.

Long-term use of nasogastric tubes, urinary catheters, or tracheostomy with mechanical ventilation likely reflects ongoing functional impairment and physiological vulnerability after cardiac arrest. These devices are associated with complications including infection, aspiration, and deconditioning, which may increase susceptibility to recurrent cardiac arrest. As such, their presence could serve as a pragmatic indicator to identify high-risk patients in need of targeted rehabilitation.

Regarding the aetiology of recurrent OHCA, previous studies have consistently shown a predominance of cardiac origins, particularly in individuals with shockable rhythms during the initial event.²² This is because shockable rhythms typically indicate underlying cardiac pathology, making successful resuscitation more likely. However, due to the use of de-identified data from the NHIRD, we were unable to directly ascertain the exact aetiology of second OHCA events in our patient population. The lack of independent association between traditional cardiac comorbidities and recurrent OHCA in our study might suggest that ongoing chronic vulnerability, as indicated by the need for life-sustaining devices, may have a more substantial impact on recurrence risk, potentially diminishing the relative influence of cardiac conditions. Nonetheless, further research explicitly investigating the exact aetiology of second OHCA events would be valuable to clarify these relationships.

Strengths and limitations

This study has several notable strengths, including the use of a comprehensive, nationwide health insurance database that provides a large sample size, robust follow-up data, and integrated measures of both healthcare utilisation and functional dependency. However, several limitations should be acknowledged. First, the dataset lacked detailed clinical parameters—such as cardiac function, the aetiology of the index OHCA, and post-arrest investigations—which limited our ability to explore mechanistic insights or conduct aetiology-specific analyses. Second, reliance on disease classification codes (ICD codes) and procedural coding may have introduced misclassification bias, potentially affecting the accuracy of patient identification and comorbidity assessments. Third, neurological outcomes, such as Cerebral Performance Category scores, were not available, restricting our evaluation of functional recovery and post-discharge cognitive trajectories. Fourth, our dataset only includes patients with documented medical encounters. Those who experienced recurrent OHCA but died in the community without being transported or evaluated in a healthcare facility would not be captured, potentially leading to an underestimation of recurrence rates. Fifth, the NHIRD does not provide information on discharge destination or post-acute rehabilitation services. Therefore, we could not assess whether patients received structured rehabilitation or were transferred to long-term care facilities after hospital discharge. Sixth, the proportion of trauma-related OHCA cases excluded was lower than typically reported.23, 24 This may be due to Taiwan’s emergent medical services policy, which allows prehospital declaration of death for patients with clear signs of non-survivable trauma. Nonetheless, such patients are unlikely to survive to discharge and therefore would not affect our analysis of recurrent OHCA risk.

Conclusion

In summary, we identified several independent factors associated with recurrent OHCA, including indicators of functional dependence and recent healthcare utilisation. The presence of nasogastric tubes and Foley catheters remained significant after adjustment, suggesting that functional status may be a useful marker of vulnerability. These findings support a more comprehensive post-discharge strategy that emphasises functional assessment, longitudinal monitoring, and follow-up among survivors hospitalised for recurrent OHCA in line with current guidelines focusing on recovery in OHCA care.

Data sharing statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to restrictions.

Declaration of AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors utilised ChatGPT (GPT-4o, OpenAI) to assist with grammar, spelling, and language fluency. All content was subsequently reviewed and edited by the authors, who take full responsibility for the final version of the manuscript.

CRediT authorship contribution statement

Sih-Shiang Huang: Writing – original draft, Visualization, Methodology, Investigation, Conceptualization. Cheng-Yi Fan: Validation, Resources, Investigation, Data curation. Yun-Ting Chih: Visualization, Validation, Software, Investigation, Formal analysis. Edward Pei-Chuan Huang: Writing – review & editing, Supervision. Chih-Wei Sung: Writing – review & editing, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Funding

This study was funded by the National Science and Technology Council (Grant no. 113-2314-B-002-182-MY3), and the National Taiwan University Hospital Hsin-Chu Branch (Grant no. 114-HCH102).

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: “This study was funded by the National Science and Technology Council (113-2314-B-002-182-MY3). The funder had no role in the study design, data collection, analysis, interpretation, or writing of the manuscript”.

Appendix A. Supplementary material

The following are the Supplementary material to this article: