Background
Cancer remains a leading cause of mortality, even surpassing heart disease in some high-income areas and countries.1–4 Nonetheless, advances in earlier detection, therapeutics, and management have led to an overall accelerated decline in cancer deaths.1 Recent innovations in novel cancer therapies have resulted in an increasing number of patients living with incurable diseases and have transformed certain malignancies to chronic conditions.5,6 As cancer treatments move to an outpatient setting, inpatient wards will care for a higher proportion of more frail and acutely ill patients with cancer who require intensive and potentially higher-risk treatments.7
Since the 1960s, cardiopulmonary resuscitation (CPR) has evolved, striving for improved performance and better outcomes.8–10 Although broadly implemented, CPR is most beneficial for acute, potentially reversible causes of cardiopulmonary arrest and should not be performed on patients with terminal or incurable conditions.9 Overall survival to hospital discharge after CPR has improved, but data for patients with cancer undergoing CPR have generally reported poorer outcomes compared with patients without cancer.11–15 However, a meta-analysis of patients with cancer who required CPR found a temporal improvement in survival.14
Data identifying specific factors or indicators of clinical trajectory to prognosticate survival after CPR in patients with cancer are lacking. Most studies focus mainly on clinical data regarding malignancy type and/or staging.13,15,16 Data that capture the clinical trajectory prior to CPR, treatment options, and overall state throughout their hospital stay are not included. Navigating conversations surrounding potential in-hospital cardiopulmonary arrest (IHCA) is challenging, and identifying which patients with cancer are unlikely to benefit from CPR is equally difficult.17 We hypothesized that in patients with cancer who experience IHCA, detailed and current oncologic focused data, specifically those capturing the overall declining clinical trajectory, may better clarify those who would not survive despite CPR. The aim of this multicenter study was to identify and validate indicators for death despite CPR based on updated oncologic data of patients with cancer requiring in-hospital CPR.
Methods
Study Design and Population
This retrospective study analyzed all patients with cancer aged ≥18 years who required in-hospital CPR from 2012 through 2015 at Memorial Sloan Kettering Cancer Center (MSKCC), MD Anderson Cancer Center (MDACC), and the Cleveland Clinic Health System (CCHS). Further details are provided in supplemental eAppendix 1 (available with this article at JNCCN.org). Only cardiopulmonary arrest events of patients with cancer that occurred at the main hospital campus were included for review. The study was granted a waiver of informed consent by the Institutional Review Board of each institution. All data were maintained in secure local REDCap databases.18 Only exported deidentified data were securely transmitted to MSKCC for final analysis.
Data Sources
Hospital and CPR databases at each institution identified all patients with cancer who underwent in-hospital CPR, defined as receipt of chest compressions and/or defibrillation for a nonperfusing rhythm with or without intubation. Given the differences in local tracking of CPR events, additional data collection methods are provided in supplemental eAppendix 1. In general, each institution collects and tracks all calls for cardiac arrest. Only events in which CPR was performed were included. Isolated respiratory arrest and cardiac events without loss of pulse were excluded.
Demographic and Oncologic Data
Demographic and clinical data included age, gender, service type (medical or surgical), and length of stay. Oncologic data included cancer type and subtype, metastatic disease, receipt of hematopoietic stem cell transplantation (HSCT), chemotherapy within 10 days of CPR, time of last cancer intervention, neutropenia, and cancer-related therapies after CPR. Cancer types were divided into solid or hematologic (leukemia, lymphoma, multiple myeloma, or HSCT), whereas subtype classification included gastrointestinal, genitourinary, head and neck, hematologic, HSCT, thoracic, or other. Last cancer-related treatment or intervention before or after CPR included any chemotherapy, radiation, surgery, HSCT, or invasive procedures that were performed for the treatment (curative or maintenance) or palliation of cancer or its symptoms. Patients without an active cancer diagnosis, basal or squamous cell carcinoma of the skin, or a remote diagnosis of cancer (no cancer-directed therapy in the year prior to CPR) were excluded.
CPR Data and Outcomes
CPR data included initial rhythm, location of arrest, duration of CPR, and targeted temperature management post CPR. Outcome data included immediate (return of spontaneous circulation [ROSC] or death) and hospital discharge status. For patients who underwent repeat CPR after an index event during the same hospitalization, only data from the first event were analyzed. Hospital readmission was defined as any readmission for any reason after receipt of CPR.
Indicators of Clinical Trajectory
Each case was manually reviewed for the presence of specific indicators that captured the overall declining clinical trajectory of patients. The indicators defined by the study team were based on clinical experience and practicality (Table 1). Each case was also reviewed independently by a second author to confirm the presence of any indicators. Any ambiguity regarding documentation of a particular indicator was adjudicated by the primary author (S. Chawla).
Detailed Indicators of Clinical and Oncologic Trajectory
Statistical Analysis
The main outcome of interest was hospital mortality, which was dichotomized as discharged alive versus died. Secondary outcomes were overall survival, which was measured 2 ways: (1) from CPR to time of death or censored at last follow-up and (2) from a landmark 24 hours after CPR to time of death or censored at last follow-up showing immediate survival of CPR and 30-day, 90-day, and 6-month mortality. Analyses were divided into 2 phases; in the first phase, a set of factors was identified that could classify patients into different risk groups for the primary endpoint using a classification tree. In the second phase, the most important features identified from the classification tree (phase I) were further described by comparing patient characteristics and outcomes between those with and without the highest risk feature.
In phase I, a classification tree based on recursive partitioning model19 was built on a training dataset that included 2 data sources (MSKCC and CCHS) and was validated using external data from MDACC. The model outcome was death before hospital discharge to determine risk-based subgroups of patients and variables that distinguish these risk groups. Variables for consideration for the model included clinical factors (age, gender, cancer type, service type, chemotherapy within 10 days of CPR, neutropenia, metastatic disease, location of care, days from last cancer intervention to CPR, days between hospital admission and CPR), along with indicators of declining clinical and oncologic trajectory as detailed in Table 1. The total number of indicators a patient had (range, 0–8) was also considered. Model performance was assessed in the training and validation set in 2 ways: within each of the final predicted subgroups and using binary cutoff for predicted probability of death before hospital discharge <50% vs ≥50%. Additional details regarding the construction and validation of the classification tree are found in supplemental eAppendix 1.
In phase II, patients were separated into 2 groups defined by the most important feature identified from phase I. Demographic and clinical characteristics were summarized and compared between groups using the Wilcoxon rank sum test for continuous variables and Fisher exact test for categorical variables. Long-term survival (>90 days) was estimated using cumulative incidence curves and compared between the 2 groups with log-rank tests. Analysis of survival endpoints was repeated among predefined subgroups of patients: age (<65 vs ≥65 years), presence of metastatic disease, and type of tumor (solid vs hematologic). Phase II analyses included data from all 3 institutions. SAS version 9.4 (SAS institute Inc) and R version 4.1.1 (R Foundation for Statistical Computing) were used for all analyses. All tests were 2-sided and P<.05 was considered significant.
Results
Patient Characteristics and Outcomes
During the study period, there were 854 patients with cancer who received in-hospital CPR among the 3 institutions (CCHS: n=225; MDACC: n=381; MSKCC: n=248). The median age was 63 years and 372 (43.6%) were female. Solid tumor was the most common cancer type, (n=516; 60.6%), and 426 (68.8%) patients had metastatic disease (Table 2). Asystole and pulseless electrical activity (PEA) were the most common initial rhythms (n=691; 88%). The median duration of CPR was 11 minutes (interquartile range [IQR], 5.0–21.0 minutes) (Table 3). Median time from last cancer intervention to CPR was 12.0 days (IQR, 3.0–32.0 days) and median time from hospital admission to CPR was 5.0 days (IQR, 1.0–14.0 days) (Table 2). Most patients (63.9%) achieved ROSC, 17.6% were discharged alive from the hospital, and 10.4% were alive at 6 months after the date of cardiopulmonary arrest (Table 3). There were no significant differences in outcomes between centers (data not shown).
Demographic and Clinical Data Among All Patients
CPR and Outcomes Data
Phase I: Classification Tree of Deaths Before Hospital Discharge
A classification tree was built on a training cohort that included patients with available hospital outcomes from MSKCC and CCHS (n=472). MDACC (n=381) was withheld as an external validation set. The training cohort had 375 (79%) deaths before discharge (primary endpoint), whereas the validation cohort had 328 (86%). A recursive partitioning model identified 3 variables to split on and created 4 risk groups (Figure 1). These were (in order of relative importance): number of indicators (<1 vs ≥1), service type (surgical vs medical or other), and duration between last cancer intervention to CPR (<3 vs ≥3 days); the presence of any indicator (or none) was identified as being the most important feature because it represented the first split at the top of the classification tree. None of the individual indicators were subsequently selected, indicating no hierarchy among them. The first group, G1, with the lowest risk of death were patients without any indicator, on surgical service, and had a last cancer intervention <3 days prior to CPR (45/472; 44% died before discharge); G2 included patients without any indicators, on surgical service, and who had a last cancer intervention ≥3 days prior to CPR (65/472; 66% died); G3 included patients without any indicators and on medical or other services (229/472; 79% died); patients in G4 had at least one indicator (133/472; 98% died) (Figure 1). No other factors were determined to add substantial information beyond this final tree. Performance metrics of the classification tree demonstrated good accuracy, high sensitivity, positive predictive value, and good discrimination ability on the training and external validation data (supplemental eFigure 1).
Decision tree analysis for Cleveland Clinic Health System and Memorial Sloan Kettering Cancer Center (n=472a).
G1, patients with 0 indicators, were on surgical service, and had a cancer intervention <3 days prior to CPR; G2, patients with 0 indicators, were on surgical service, and had a cancer intervention ≥3 days prior to CPR; G3, patients with 0 indicators and were on medical or other services; G4, patients with at least 1 indicator.
Abbreviation: CPR, cardiopulmonary resuscitation.
aPatients with available hospital outcomes.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Decision tree analysis for Cleveland Clinic Health System and Memorial Sloan Kettering Cancer Center (n=472a).
G1, patients with 0 indicators, were on surgical service, and had a cancer intervention <3 days prior to CPR; G2, patients with 0 indicators, were on surgical service, and had a cancer intervention ≥3 days prior to CPR; G3, patients with 0 indicators and were on medical or other services; G4, patients with at least 1 indicator.
Abbreviation: CPR, cardiopulmonary resuscitation.
aPatients with available hospital outcomes.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Decision tree analysis for Cleveland Clinic Health System and Memorial Sloan Kettering Cancer Center (n=472a).
G1, patients with 0 indicators, were on surgical service, and had a cancer intervention <3 days prior to CPR; G2, patients with 0 indicators, were on surgical service, and had a cancer intervention ≥3 days prior to CPR; G3, patients with 0 indicators and were on medical or other services; G4, patients with at least 1 indicator.
Abbreviation: CPR, cardiopulmonary resuscitation.
aPatients with available hospital outcomes.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Phase II: Patient Characteristics by Any Versus No Indicators
Based on the classification tree, the most important feature identified was the presence of an indicator (G4; Figure 1). Therefore, patients were divided into 2 groups depending on whether any indicator was present versus no indicator. In total, 249 (29.2%) patients had at least one indicator and were younger (median age, 59 vs 64 years; P<.001), were more often on medical service (94.4% vs 75.5%; P<.001), had a higher incidence of metastatic disease (83% vs 62.9%; P<.001), and were more likely to have CPR in the ICU (55.8% vs 36.5%; P<.001). For patients with any indicator present, the interval from the last therapeutic intervention to CPR (19 vs 9 days; P<.001) and the interval from hospital admission to CPR (8 vs 4 days; P<.001) were both longer than for patients without any indicators (Table 2). There was no significant difference in cancer type or subtype, recent chemotherapy, presence of neutropenia, initial cardiac rhythm, or duration of CPR between the 2 groups (Tables 2 and 3).
Examining post-CPR outcomes, patients with any indicator had higher incidence of death after CPR compared with those with no indicators (30-day cumulative incidence of death estimate: 98% vs 78%; P<.001) (Figure 2). Similar patterns were observed within clinical subgroups, including age ≥65 years (95% vs 78%; P<.001) and <65 years (99% vs 78%; P<.001); presence of metastatic disease (97% vs 80%; P<.001) and no metastatic disease (93% vs 68%; P<.001); and solid (97% vs 75%; P<.001) or hematologic cancer (98% vs 82%; P<.001). These relationships were significant whether analyzing all patients or only those who survived past the first 24 hours after CPR (supplemental eFigures 2–8). Among all patients with an indicator, there was no difference in mortality between CPR location (ICU vs non-ICU: 99% vs 95%; P=.101). A similar result was obtained when analyzing patients who survived past the first 24 hours after CPR (97% vs 85%; P=.504) (supplemental eFigure 9).
Cumulative incidence of death by any or no indicators among (A) all patients from time of CPR and (B) patients who survived past the first 24 hours after CPR.
Abbreviations: Cum_inc, cumulative incidence of death estimate; LCL, lower confidence limit; UCL, upper confidence limit.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Cumulative incidence of death by any or no indicators among (A) all patients from time of CPR and (B) patients who survived past the first 24 hours after CPR.
Abbreviations: Cum_inc, cumulative incidence of death estimate; LCL, lower confidence limit; UCL, upper confidence limit.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Cumulative incidence of death by any or no indicators among (A) all patients from time of CPR and (B) patients who survived past the first 24 hours after CPR.
Abbreviations: Cum_inc, cumulative incidence of death estimate; LCL, lower confidence limit; UCL, upper confidence limit.
Citation: Journal of the National Comprehensive Cancer Network 21, 1; 10.6004/jnccn.2022.7072
Discussion
Our study is the first to describe the application of simple clinical and cancer trajectory indicators to identify hospitalized patients with cancer at higher risk of death following in-hospital CPR. We also show that patients with cancer without indicators who experience IHCA have survival rates comparable to those of the general population.12 Most prior studies on patients with cancer who receive CPR are generally small, single-center, and published at least 2 decades ago.14 Recent estimates report that approximately 1.92 million new cancer cases will be diagnosed in the United States in 2022. Nationally, patients with cancer account for >4.5 million hospital admissions annually.1,20 Cancer therapy and overall cancer survival, as well as outcomes, have improved over time.1,21 As cancer care and therapies evolve, the potential for cardiopulmonary arrest and the need for CPR will likely increase. One study of a large national CPR database found that 14% of the individuals had advanced cancer.13 Therefore, providers must be able to use practical clinical parameters to optimally guide discussions, because cancer prognosis may change throughout the course of treatment.
We found that nearly one-third (29.2%) of patients had at least one indicator of declining trajectory present prior to CPR. Of these, 98% died during hospitalization, with only 5 patients discharged alive. All of these patients were discharged to a long-term facility and died within 6 months of discharge without any opportunity to receive any further cancer interventions. Although patients with indicators had universally poor outcomes, our data also show that survival to hospital discharge for the remaining patients approached that of the general population.12 This finding may be due in part to the high volume of patients with cancer in our institutions and perhaps to a higher rate of do-not-resuscitate (DNR) consents in applicable circumstances. In the United States, the annual incidence of IHCA has steadily increased, paralleled with improving overall survival to discharge.11,12,22 Nonetheless, compared with noncancer patients, the survival rates of IHCA are lower for those with cancer.13–15,23,24
Previous studies of IHCA have used survival to hospital discharge and neurologic sequelae as meaningful metrics of CPR outcomes. Beyond surviving CPR, a key parameter for patients with cancer is the opportunity to receive further treatment. Thus, we propose further cancer intervention and hospital readmission as more relevant measures of long-term outcomes in this population. In our multicenter study, approximately half of patients discharged alive went on to receive further cancer intervention, all of whom did not have an indicator. Therefore, our analysis will help to recognize not only poor short-term survival but also long-term survival and the lack of further cancer-related intervention.
We chose to analyze our data using a classification tree rather than multivariable modeling based on the goal of determining subgroups of patients who are at elevated risk of death prior to discharge rather than estimating the marginal effect of potential factors. Information from the classification tree can provide greater clinical utility and reveal the relative importance of different factors. With more granular data, future studies can explore more sophisticated predictive modeling through latent class analyses or other unsupervised methods.
Our primary focus was to assess risk of death despite CPR; therefore, we did not specifically examine the reasons for the provision of CPR. It is possible there were different reasons to proceed with CPR that were not gleaned by the providers. Not uncommonly, patients and/or caregivers feel the need to “fight to the end,” and pursue CPR for personal reasons. Despite warnings about poor outcomes, 58% of patients with cancer would want CPR, according to one study.25 Additionally, the lay public has generally overestimated survival after CPR.26,27 Clinicians perceive CPR as inappropriate when there are indicators of poor outcome.28 Nonetheless, providers will need to balance their views with laypersons’ understanding.29
Our study has limitations. First, our data are from 2012 through 2015, but we do not feel that there has been any fundamental shift in the management of patients with cancer that would diminish the applicability or relevance of these findings. We hope to analyze more contemporary data and validate our findings. Additionally, noncancer centers could help evaluate the generalizability of these indicators. Second, we could not focus on CPR quality, provider experience, or their potential impact on outcomes. However, our rate of ROSC was similar to what is previously described. Therefore, one could infer that the CPR quality did not play a significant role.13,30 Third, the indicators we chose might be viewed as subjective; however, very similar parameters were found to be associated with poor outcomes when used in the assessment of ICU admission for hematologic malignancy.31–33 Fourth, due to the retrospective nature, an indicator such as last cancer intervention to CPR may have limited utility, because clinicians likely will not know precisely when CPR will occur. Finally, we were unable to collect neurologic outcomes in our study. We invite consideration of the proposed new metric of cancer-directed interventions for cure, disease control, or palliation rather than neurologic outcomes.34
Conclusions
Through the application of clinical and cancer trajectory indicators, our study identifies patients with cancer who will not survive to hospital discharge despite CPR if they experience IHCA. Early recognition and discussion could avoid CPR in circumstances that would only prolong suffering.
Acknowledgments
The authors would like to thank the following contributors for their participation and work in data collection and reviewing individual medical records for indicators: Meaghen Finan, MD; Lisa Canecchia, ACNP; Taylen Richards, BA; Madeleine Hicks, BA; Mary Lou Warren, DNP, RN, CNS-CC; Jason W. Myers, ACNP-BC; Kristen L. Bratcher, ACNP-BC; Jeanne Y. Campbell, MPAS, PA-C; and Virginia V. Schneider, MPAS, PA-C. The authors would also like to thank Marvin Radford at MDACC for sharing the list of all Code Blues in the institution, and the MDACC CPR committee for facilitating access to their database and their continuous work to improve outcomes of our patients.
References
- 2.↑
Hastings KG, Boothroyd DB, Kapphahn K, et al. Socioeconomic differences in the epidemiologic transition from heart disease to cancer as the leading cause of death in the United States, 2003 to 2015: an observational study. Ann Intern Med 2018;169:836–844.
- 3.↑
Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394–424.
- 4.↑
Dagenais GR, Leong DP, Rangarajan S, et al. Variations in common diseases, hospital admissions, and deaths in middle-aged adults in 21 countries from five continents (PURE): a prospective cohort study. Lancet 2020;395:785–794.
- 5.↑
Marks PA, Sterngold J. On the Cancer Frontier: One Man, One Disease, and a Medical Revolution, 1st ed. New York, NY: PublicAffairs; 2014.
- 6.↑
DeVita VT, DeVita-Raeburn E. The Death of Cancer: After Fifty Years on the Front Lines of Medicine, a Pioneering Oncologist Reveals Why the War on Cancer Is Winnable–and How We Can Get There. New York, NY: Sarah Crichton Books/Farrar, Straus and Giroux; 2015.
- 8.↑
Kouwenhoven WB, Jude JR, Knickerbocker GG. Closed-chest cardiac massage. JAMA 1960;173:1064–1067.
- 9.↑
Cardiopulmonary resuscitation: statement by the Ad Hoc Committee on Cardiopulmonary Resuscitation of the Division of Medical Sciences, National Academy of Sciences—National Research Council. JAMA 1966;198:372–379.
- 10.↑
Lavonas EJ, Drennan IR, Gabrielli A, et al. Part 10: special circumstances of resuscitation: 2015 American Heart Association guidelines update for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation 2015;132(Suppl 2):S501–518.
- 11.↑
Girotra S, Nallamothu BK, Spertus JA, et al. Trends in survival after in-hospital cardiac arrest. N Engl J Med 2012;367:1912–1920.
- 12.↑
Andersen LW, Holmberg MJ, Berg KM, et al. In-hospital cardiac arrest: a review. JAMA 2019;321:1200–1210.
- 13.↑
Bruckel JT, Wong SL, Chan PS, et al. Patterns of resuscitation care and survival after in-hospital cardiac arrest in patients with advanced cancer. J Oncol Pract 2017;13:e821–830.
- 14.↑
Reisfield GM, Wallace SK, Munsell MF, et al. Survival in cancer patients undergoing in-hospital cardiopulmonary resuscitation: a meta-analysis. Resuscitation 2006;71:152–160.
- 15.↑
Guha A, Buck B, Biersmith M, et al. Contemporary impacts of a cancer diagnosis on survival following in-hospital cardiac arrest. Resuscitation 2019;142:30–37.
- 16.↑
Ebell MH, Jang W, Shen Y, et al. Development and validation of the Good Outcome Following Attempted Resuscitation (GO-FAR) score to predict neurologically intact survival after in-hospital cardiopulmonary resuscitation. JAMA Intern Med 2013;173:1872–1878.
- 17.↑
Tol I, Cumber E, Nakakande D, et al. Cardiopulmonary resuscitation discussions with patients admitted to acute oncology wards: a national audit of current practice. Eur J Cancer Care (Engl) 2020;29:e13218.
- 18.↑
Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–381.
- 19.↑
Breiman L, Friedman J, Stone CJ, et al. Classification and Regression Trees, 1st ed. New York, NY: Chapman & Hall; 1984.
- 20.↑
Price RA, Stranges E, Elixhauser A. Cancer hospitalizations for adults, 2009. Accessed March 1, 2022. Available at: https://hcup-us.ahrq.gov/reports/statbriefs/sb125.jsp
- 21.↑
Lemiale V, Pons S, Mirouse A, et al. Sepsis and septic shock in patients with malignancies: a Groupe de Recherche Respiratoire en Réanimation Onco-Hématologique study. Crit Care Med 2020;48:822–829.
- 22.↑
Holmberg MJ, Ross CE, Fitzmaurice GM, et al. Annual incidence of adult and pediatric in-hospital cardiac arrest in the United States. Circ Cardiovasc Qual Outcomes 2019;12:e005580.
- 23.↑
Champigneulle B, Merceron S, Lemiale V, et al. What is the outcome of cancer patients admitted to the ICU after cardiac arrest? Results from a multicenter study. Resuscitation 2015;92:38–44.
- 24.↑
Taran A, Guarino M, Kolm P, et al. Cardiopulmonary resuscitation inpatient outcomes in cancer patients in a large community hospital. Del Med J 2012;84:117–121.
- 25.↑
Ackroyd R, Russon L, Newell R. Views of oncology patients, their relatives and oncologists on cardiopulmonary resuscitation (CPR): questionnaire-based study. Palliat Med 2007;21:139–144.
- 26.↑
Jones GK, Brewer KL, Garrison HG. Public expectations of survival following cardiopulmonary resuscitation. Acad Emerg Med 2000;7:48–53.
- 27.↑
Marco CA, Larkin GL. Cardiopulmonary resuscitation: knowledge and opinions among the U.S. general public. State of the science-fiction. Resuscitation 2008;79:490–498.
- 28.↑
Druwé P, Monsieurs KG, Piers R, et al. Perception of inappropriate cardiopulmonary resuscitation by clinicians working in emergency departments and ambulance services: the REAPPROPRIATE international, multi-centre, cross sectional survey. Resuscitation 2018; 132:112–119.
- 29.↑
Neville TH, Tarn DM, Pavlish CL, et al. The community perspective on potentially inappropriate treatment. Ann Am Thorac Soc 2020;17:854–859.
- 30.↑
Ofoma UR, Basnet S, Berger A, et al. Trends in survival after in-hospital cardiac arrest during nights and weekends. J Am Coll Cardiol 2018;71:402–411.
- 31.↑
Azoulay E, Pène F, Darmon M, et al. Managing critically ill hematology patients: time to think differently. Blood Rev 2015;29:359–367.
- 32.↑
Dasch B, Lenz P, Zahn PK. Prevalence of resuscitation in cancer patients at the end of life-a population-based observational study from Germany. Ann Palliat Med 2021;10:1101–1114.
- 33.↑
Potenza L, Scaravaglio M, Fortuna D, et al. Early palliative/supportive care in acute myeloid leukaemia allows low aggression end-of-life interventions: observational outpatient study. BMJ Support Palliat Care. Published online November 8, 2021. doi:10.1136/bmjspcare-2021-002898
- 34.↑
Lee MR, Yu KL, Kuo HY, et al. Outcome of stage IV cancer patients receiving in-hospital cardiopulmonary resuscitation: a population-based cohort study. Sci Rep 2019;9:9478.
