Survival Outcomes after Cardiac Arrest in Community-Dwelling Adults Receiving Home Care versus Nursing Home Residents Compared With Unsupported Individuals

Introduction

Although advances in cardiac arrest treatment continue, one-year survival remains low—13% for in-hospital arrests and 8% for out-of-hospital cases [1, 2]. Those who do survive frequently experience long-term physical, cognitive, and psychological challenges, along with reduced quality of life [3-5]. Outcomes are generally poorer for older adults and people living with frailty [6, 7]. Home care clients and nursing home residents represent two aging, medically complex groups with substantial frailty and high use of emergency services [8-10]. Roughly 30% of older Canadians rely on publicly funded home care or live in nursing homes [11, 12]. Home care encompasses a broad set of personal and clinical supports—ranging from personal assistance and nursing care to rehabilitation and mental health services [13]. When needs exceed what home care can offer, individuals often transition to nursing homes, which provide continuous supervision and accommodation. Notably, around 20% of those in long-term care facilities in Canada could theoretically be supported in a home care environment due to their comparable care requirements and clinical characteristics [14].

Given the elevated rates of cardiorespiratory disease and advanced illness in these groups, their risk of cardiac arrest is substantial [15], making early end-of-life planning especially relevant when services begin [16]. Previous work comparing post-arrest survival in nursing home residents versus community-dwelling older adults has shown mixed results [17-21]. Evidence remains limited regarding cardiac arrest outcomes among home care recipients, a population that continues to grow and often faces unmet care needs [22]. Foundational prognostic research is necessary to determine disease burden and identify people who may derive the greatest benefit from advance care planning [23].

The primary aim of this study was to estimate the likelihood of 30-day survival following cardiac arrest among individuals receiving home care and among nursing home residents, compared with community-dwelling adults not using support services. We anticipated lower 30-day survival in both groups. Our secondary aim was to examine one-year survival and the frequency of death occurring at home. Absolute risk estimates are provided for both out-of-hospital and in-hospital cardiac arrest.

Materials and Methods

Study design and data sources

We carried out a retrospective cohort investigation at the population level by linking several anonymized administrative health databases maintained at ICES (formerly the Institute of Clinical and Evaluative Sciences). All individuals who received care for a cardiac arrest in any hospital across Ontario between January 1, 2006, and December 31, 2018, were identified. Arrests that happened after a patient was admitted were labeled in-hospital cases, whereas those occurring before arrival or treated initially in the emergency department were categorized as out-of-hospital events. Although the emergency department is part of the hospital, most cardiac arrests managed there originate outside the facility and are therefore classified as out-of-hospital [24, 25].

Long-term home care recipients (>60 days) and residents of nursing homes were identified using the Home Care Dataset and the Continuing Care Reporting System. Basic demographic variables such as sex and age were retrieved from the Registered Person Data Base. Information related to diagnoses, visit features, clinical procedures, and outcomes for out-of-hospital arrests came from the National Ambulatory Care Reporting System, while similar information for in-hospital arrests was obtained from the Discharge Abstract Database. Use of support devices such as mobility aids or sensory equipment was determined through the Assistive Devices Program dataset. Records of patients admitted for myocardial infarction before their arrest were located via the Ontario Myocardial Infarction Dataset. Details regarding death—timing, location, cause, and autopsy status—were acquired from the Vital Statistics and Death database. All databases included in the analysis undergo routine data quality checks and have been validated for population-based health research in Ontario and Canada [26, 27].

A waiver of ethics review was granted by the Hamilton Integrated Research Ethics Board, as individual consent is not necessary for this type of data according to Section 45 of Ontario’s Personal Health Information Protection Act [28]. Study reporting conformed to the STROBE statement for observational research [29].

Cohort and exposures

We included all patients aged 18 years or older who received emergency department or inpatient care for a cardiac arrest at any hospital in Ontario. Follow-up extended for one year after the arrest. For in-hospital cardiac arrests, only those occurring within 72 hours of ED registration were retained. ED registration was used as the reference point because it marks the first encounter with hospital-based services. The 72-hour limit was intended to reduce the influence of clinical deterioration or other confounding factors that may arise later during hospitalization [30]. Prior evidence indicates that most in-hospital arrests take place within the first two days of admission [31]; accordingly, an additional 24-hour buffer was added for individuals waiting in the ED for an inpatient bed.

We identified cardiopulmonary resuscitation using Canadian Classification of Health Interventions codes (IHZ30JN, IHZ3OJY) and determined cardiac arrest events using ICD-10 codes (I46.1, I46.2, I46.8, I49.0, I49.01, I49.02, R96.0, R96.1, R98, R99) [24, 32, 33]. If a patient experienced more than one arrest during the study period, only the first was included because subsequent arrests are associated with markedly lower survival [34, 35] and could introduce correlated observations. Individuals who were not Ontario residents or who lacked a valid Ontario Health Insurance Plan number or date of birth were excluded. We also removed patients who underwent a surgical procedure within the 72-hour window prior to arrest to focus on arrests primarily driven by medical deterioration rather than postoperative or traumatic causes, which follow different pathways of care [31].

Age was analyzed as a categorical variable (18–49, 50–64, 65–74, 75–84, and 85+ years) due to ICES privacy requirements. Triage urgency at ED presentation was assessed using the Canadian Triage Acuity Scale, which ranges from level one (highest acuity) to level five [36]. Pre-arrest disease burden was assigned using ICD-10 morbidity codes.

Outcomes

The chief endpoint for this work was survival 30 days after a cardiac arrest event. This interval was selected because it aligns closely with discharge survival and is consistent with recommendations outlined in the Utstein guidance documents [37, 38]. We additionally assessed one-year survival and whether death occurred at home.

A death was categorized as “in-home” when it happened outside any medical or institutional care environment (for example, hospitals, rehabilitation facilities, or long-term care homes). Both survival metrics and place of death are emphasized as outcomes that matter to patients and are prioritized in the COSCA framework and by the ICHOM Older Person Working Group [39, 40].

Analysis

Descriptive summaries include counts and measures of central tendency. Logistic regression was used to generate odds ratios and 95% confidence intervals, which were then converted to absolute risk estimates following recommendations from the PROGRESS and GRADE prognosis groups [23, 41].

Age and sex were incorporated into multivariable adjustments because these factors are consistently linked to variations in health trajectories, recovery after arrest, and patterns of bystander action in out-of-hospital arrests [42, 43]. Analyses of death location excluded nursing home residents due to the inability to determine the precise institution where death occurred. Prognostic estimates were produced for the full sample and separately for in-hospital and out-of-hospital arrests. Missing data were minimal (<0.1%) and removed within each relevant analysis. All data handling and statistical procedures were conducted using R version 3.6.0.

Results and Discussion

A total of 86,836 adults with either in-hospital or out-of-hospital cardiac arrest were included. Of these, 47,226 (55.5%) arrests took place in hospitals, while 39,610 (45.5%) occurred out-of-hospital; 7,207 (18.1%) of the latter group showed no signs of life on arrival despite attempted resuscitation. Most individuals were not receiving any support services before arrest (85.9%), were male (60.3%), arrived by ambulance (83.3%), and reached the hospital during daytime periods (60.2%). Figure 1 presents the inclusion pathway and survival distribution. Additional pre-arrest characteristics, proportions, and risk estimates stratified by support status, age, and arrest setting are provided in Supplemental Tables 1–3.

Pre-arrest characteristics associated with support service use

Table 1 summarizes pre-arrest features across those enrolled in home care, residents of nursing homes, and individuals without support. Figure 2 shows the absolute risk differences for requiring these services. Older age strongly correlated with support service use; those aged 85+ had the highest estimated risk (RD = 38.7; 95% CI = 37.9–39.4) compared with adults aged 18–49. Compared with individuals not using services, those in home care or nursing homes more frequently had congestive heart failure (50.8% vs. 29.3%), COPD (49.7% vs. 35.5%), and dementia (37.5% vs. 8.3%). Respiratory illnesses—including COPD exacerbations (RD = 8.1; 95% CI = 6.4–10.2) and pneumonia (RD = 8.8; 95% CI = 6.9–10.9)—were also more commonly listed as causes of death in these groups.

Table 1. Patient Features Compared Between Pre-Arrest Support Status in 86,836 Adults Who Experienced Cardiac Arrest

Daytime presentation: typically 07:00–22:59 CI = Confidence Interval; COPD = Chronic Obstructive Pulmonary Disease; CTAS = Canadian Triage and Acuity Scale

Pre-arrest features across arrest locations

Table 2 contrasts characteristics present before arrest between individuals experiencing out-of-hospital and in-hospital events. Adults aged 85 and above had the highest likelihood of sustaining an arrest outside the hospital when compared with those younger than 50 (RD = −17.7; 95% CI = −18.9 to −16.8). In contrast, people with serious chronic conditions—such as congestive heart failure (RD = 5.6; 95% CI = 4.9–6.3), COPD (RD = 11.2; 95% CI = 10.6–11.7), and dementia (RD = 7.2; 95% CI = 6.3–8.1)—tended to experience arrest in-hospital. Arrests inside the hospital were also more common among women (RD = 8.9; 95% CI = 8.4–9.5) and were more likely to occur during daytime hours between 07:00 and 19:00 (RD = 21.3; 95% CI = 21.3–22.5).

Table 2. Comparison of Pre-Arrest Features Between 39,610 (45.5%) Out-of-Hospital Arrests and 47,226 (55.5%) In-Hospital Arrests in Ontario, Canada

Daytime defined as 07:00–22:59 COPD = Chronic Obstructive Pulmonary Disease CTAS = Canadian Triage and Acuity Scale

Post–cardiac arrest survival

Adjusted odds ratios and confidence intervals for the full sample and for the two arrest settings appear in Supplemental Tables 4–6.

30-day survival

A total of 31,180 individuals (36.0%) remained alive 30 days after arrest. Most deaths occurred beyond the first week (61.9%). After accounting for age and sex, those receiving home care (RD = −6.5; 95% CI = −7.5 to −5.0) and those living in nursing homes (RD = −7.2; 95% CI = −9.3 to −5.3) had lower 30-day survival compared with people without pre-arrest support.

Home care users showed reduced 30-day survival in both out-of-hospital arrests (RD = −6.7; 95% CI = −7.6 to −5.7) and in-hospital arrests (RD = −8.7; 95% CI = −10.6 to −7.3). Nursing home residents demonstrated comparable deficits in out-of-hospital arrests (RD = −8.6; 95% CI = −10.6 to −5.7) and in in-hospital arrests (RD = −5.0; 95% CI = −7.8 to −2.1).

Figure 3 illustrates one-month survival probabilities for the three groups (home care, nursing home, and those without services). Figure 4 presents adjusted absolute risk differences for 30-day and 1-year survival relative to support status.

Figure 3. Probability of Survival to 30 Days Post–Cardiac Arrest in 86,836 Individuals in Ontario, Canada

Figure 4. Forest Plot of Adjusted Risk Differences for 30-Day and 1-Year Survival Across Pre-Arrest Support Categories (N = 86,836)

One-year survival

Roughly one-third (31.6%) were alive at one year. After adjusting for age and sex, home care recipients (RD = −9.6; 95% CI = −10.5 to −8.5) and nursing home residents (RD = −11.8; 95% CI = −13.6 to −9.7) again showed poorer outcomes than those without support needs. For one-year survival, home care users had worse absolute risks in both out-of-hospital (RD = −7.7; 95% CI = −8.6 to −6.9) and in-hospital arrests (RD = −13.2; 95% CI = −14.8 to −11.5). Similar patterns were observed among nursing home residents in out-of-hospital arrests (RD = −8.7; 95% CI = −9.9 to −7.4) and in-hospital arrests (RD = −12.8; 95% CI = −16.1 to −10.2).

In-home death

After removing nursing home residents from this analysis, only 4,586 individuals (7.8%) died in a home setting. Among out-of-hospital arrests, people receiving home care were less likely to die at home within a year (RD = −1.3; 95% CI = −2.2 to −0.05). Older age groups displayed a progressively lower probability of home death. Rates of pre-hospital death were nearly equivalent for those who ultimately died at home (13.9%) and those who did not (12.1%).

We found that individuals relying on home care or residing in nursing homes had lower 30-day survival following cardiac arrest than those without pre-arrest support services. This pattern held both overall and within out-of-hospital and in-hospital arrest subgroups. Younger adults showed more favorable survival trajectories and had a higher likelihood of dying at home; moreover, home care recipients were more likely to die at home if their arrest occurred outside the hospital.

Comparison to prior research

Limited evidence exists regarding cardiac arrest outcomes among home care clients, but our findings align with previous work showing poorer survival among nursing home residents compared with community-dwelling older adults [13, 15, 17]. This study contributes population-level comparisons across three cohorts in Ontario—home care users, nursing home residents, and adults without support services. The results reinforce earlier observations that age is a strong prognostic indicator inversely linked with survival after both out-of-hospital and in-hospital arrests [7, 44].

Clinical and policy implications

Individuals receiving home care or living in nursing homes showed lower chances of surviving to both 30 days and one year after cardiac arrest. Among those who were arrested outside the hospital, home care users were more likely to die in their own homes. These outcomes—survival and place of death—are meaningful to patients and should be incorporated into advance care planning discussions for home care and long-term care populations [40]. Because these groups tend to be older and experience higher burdens of advanced chronic illness, clinicians can quickly identify pre-arrest support needs and use them to guide shared decision-making about end-of-life priorities and directives.

Our analysis showed that enrollment in support services before arrest was linked to poorer 30-day and one-year survival, with comparable absolute risks in the home care and nursing home cohorts (differences <3%). These similar results may indicate overlapping levels of frailty and medical complexity. The chronic shortage of long-term care beds may contribute to worsening health among home care users prior to long-term care admission. Although nursing home residents were older, home care clients more often had high-risk chronic conditions such as congestive heart failure and COPD. Arrests may have occurred among the least frail nursing home residents or, conversely, among the more medically vulnerable home care patients. Another explanation is that nursing home residents might experience more witnessed arrests and receive CPR more often because of constant access to nursing staff [18, 19].

Public expectations of cardiac arrest survival tend to be overly optimistic, partly due to media portrayals, highlighting the ongoing need for realistic conversations led by clinicians [45, 46]. Advance care planning is frequently delayed until severe illness arises [47], which can lead to choices that do not align with a patient’s true preferences, particularly overtreatment [48]. Early and collaborative discussions about advance directives help patients better understand their prognosis and match their care goals to what is achievable [49, 50]. Home care and nursing home settings are well positioned for these conversations because of routine assessments and structured care processes. As these individuals age, the importance of such discussions increases; older adults are less likely to die at home and more likely to die in a hospital, which may contrast with their stated goals or directives [40, 51].

Strengths and limitations

This study offers a unique population-level assessment of survival following cardiac arrest and the likelihood of dying at home across categories of pre-arrest support, age, and sex. Population-wide datasets are suitable for examining cardiac arrest outcomes because the condition is relatively uncommon and requires large samples to inform health system planning [52].

For in-hospital cardiac arrests, our ability to determine overall prognosis was limited because data capture was restricted to events occurring within a 72-hour window. This likely contributed to inflated survival estimates, as clinical deterioration increases with extended hospitalization [53]. A further possible explanation is that many home care or nursing home residents may have do-not-resuscitate orders, meaning those who do undergo CPR may represent a healthier subset [54]. For arrests occurring in the community, we may have overlooked a small number of patients who were resuscitated but never transported to the hospital.

Reliance on ICD-10 administrative coding to identify comorbidities introduces uncertainties due to misclassification and transcription errors [55]. We could not distinguish whether deaths in institutions occurred in hospitals or nursing homes, which required excluding nursing home residents from the analysis of home death. Because of small cell sizes, age could not be treated as a continuous measure and had to be categorized, though multiple cut points were used to refine estimates. Finally, the absence of population-level information on multimorbidity and frailty—both important predictors—limited the depth of the analysis [56].

Conclusion

Home care users and nursing home residents exhibited poorer 30-day and one-year survival after cardiac arrest. Older individuals were also less likely to die at home. These findings reinforce the importance of early, structured discussions about end-of-life care during enrollment and routine follow-up in home care and long-term care environments.

Acknowledgments: We want to acknowledge the Canadian Institute of Health Research (CHIR) for funding this work with a Frederick Banting & Charles Best Canada Graduate Scholarship.

Conflict of interest: None

Financial support: None

Ethics statement: None