Abstract
The role of early Do Not Resuscitate (DNR) in hospitalized older adults (OAs) with SARS-CoV-2 infection is unknown. The objective of the study was to identify characteristics and outcomes associated with early DNR in hospitalized OAs with SARS-CoV-2. We conducted a retrospective chart review of older adults (65+) hospitalized with COVID-19 in New York, USA, between March 1, 2020, and April 20, 2020. Patient characteristics and hospital outcomes were collected. Early DNR (within 24 hours of admission) was compared to non-early DNR (late DNR, after 24 hours of admission, or no DNR). Outcomes included hospital morbidity and mortality. Of 4961 patients, early DNR prevalence was 5.7% (n = 283). Compared to non-early DNR, the early DNR group was older (85.0 vs 76.8, P < .001), women (51.2% vs 43.6%, P = .012), with higher comorbidity index (3.88 vs 3.36, P < .001), facility-based (49.1% vs 19.1%, P < .001), with dementia (13.3% vs 4.6%, P < .001), and severely ill on presentation (57.9% vs 32.3%, P < .001). In multivariable analyses, the early DNR group had higher mortality risk (OR: 2.94, 95% CI: 2.10–4.11), less hospital delirium (OR: 0.55, 95% CI: 0.40–.77), lower use of invasive mechanical ventilation (IMV, OR: 0.37, 95% CI: .21–.67), and shorter length of stay (LOS, 4.8 vs 10.3 days, P < .001), compared to non-early DNR. Regarding early vs late DNR, while there was no difference in mortality (OR: 1.12, 95% CI: 0.85–1.62), the early DNR group experienced less delirium (OR: 0.55, 95% CI: .40–.75), IMV (OR: 0.53, 95% CI: 0.29–.96), and shorter LOS (4.82 vs 10.63 days, OR: 0.35, 95% CI: 0.30–.41). In conclusion, early DNR prevalence in hospitalized OAs with COVID-19 was low, and compared to non-early DNR is associated with higher mortality but lower morbidity.
Introduction
SARS-CoV-2, the virus responsible for the COVID-19 pandemic, has infected over 66 million people in the United States and resulted in over 850,000 American deaths (at the time of this writing). 1 Older adults (OAs age 65 years and older) have been disproportionately affected by COVID-19, accounting for nearly 50% of hospitalizations, over 50% of ICU admissions, and 80% of deaths. 2 The high risk of COVID-19 in OAs brought the topic of mortality to the forefront of our collective consciousness and highlighted the need to discuss and document life-sustaining treatment preferences. 3
Enacted in 1991, the Patient Self Determination Act (PSDA) required that all health care institutions inform patients about their right to receive preference-congruent care, including care that aligns with their preferences for life-sustaining treatment. 4 A key component of this is a determination of resuscitation preferences during cardiac arrest, or “code status,” which ultimately leads to a “full code” (full resuscitative measures) or a “Do-Not-Resuscitate” (DNR) order that precludes the concerted use of cardiopulmonary resuscitation (CPR), invasive mechanical ventilation (IMV), electric defibrillation, cardioactive drugs and vasopressors in the event of cardiac arrest. Eliciting life-sustaining treatment preferences of patients and, more specifically, eliciting code status was deemed essential in providing preference-congruent care to hospitalized OAs during COVID-19.5,6
Characteristics of patients with a documented DNR have been studied since the enactment of the PSDA in several medical conditions, including septic shock, pneumonia, ischemic stroke out-of-hospital cardiac arrest, as well as in surgical and general medical patient populations.7-14 In the pre–COVID-19 literature, DNR has been differentiated as early (ordered within 24 hours of hospital admission) and late (ordered 24 hours after hospital admission). Early DNR reflects baseline patient resuscitation preferences and pre-hospital disease burden; late DNR reflects poor response to treatment and worsening clinical trajectory.7,15 Late DNR has been associated with higher mortality compared to early DNR, longer hospital and ICU length of stay (LOS), and other indirect markers of hospital complications and morbidity, including delirium.13,14
While early DNR has been studied prior to COVID-19 and in the context of several medical conditions, the prevalence, associated characteristics, and hospital outcomes of early DNR in OAs hospitalized with COVID-19 during the initial pandemic surge are unknown. The frenzied atmosphere early in the pandemic, which was caused by the overwhelming number of hospital admission, disproportionate impact of COVID-19 on OAs, concerns of rationing resources (e.g., mechanical ventilators), and the fear of not being able to see loved ones once admitted, undoubtedly impacted both the medical community’s and public’s perspective on early resuscitation discussions and preferences. Thus, there is an urgent need to evaluate the role of early DNR in this vulnerable population during the initial surge of COVID-19.
Our study aimed to (1) describe the prevalence of early DNR in OAs hospitalized with COVID-19; (2) identify patient characteristics associated with the presence of an early DNR compared to those without an early DNR (non-early DNR comprised of those with a late DNR and those without a DNR); (3) compare early and non-early DNR patients with respect to hospital mortality and morbidity indicators; (4) compare early and late DNR patients with respect to characteristics and outcomes in hospitalized OAs with COVID-19.
Methods
Study Design
An IRB-approved retrospective study was conducted of all OAs, age 65 years and older, hospitalized with COVID-19 across a large health system in the greater NY metropolitan area between March 1, 2020 and April 20, 2020, the height of the initial pandemic. Patients were included if they had a positive nasopharyngeal swab polymerase chain reaction (PCR) assay for SARS-CoV-2 during hospitalization. Patients were excluded if they were transferred to another acute care hospital or military medical station from which data for their complete hospital course could not be obtained. Patients who remained hospitalized at the end of the study period were also excluded.
Measures
Data were collected from enterprise electronic health record (EHR) (Sunrise Clinical Manager; Allscripts).
Early, non-early, and late DNR
To evaluate the role of early DNR in OA hospitalized with COVID-19, the primary comparison was early DNR (ordered within 24 hours of admission) vs non-early DNR. The non-early DNR group comprised patients with a late DNR (ordered after 24 hours of admission) and patients with no DNR whatsoever. This grouping was because: (1) it is not possible to know at the time of early DNR which remaining patients will eventually receive a late DNR vs no DNR at all, (2) we aimed to compare patients that specifically elected early DNR with all those who did not.
Patient demographics and baseline characteristics
Demographics included age, sex (male or female), self-reported race (Black, White, Asian, Other, or not available), self- reported ethnicity (Hispanic or non-Hispanic) and language (English, Spanish, other, or missing/not available). Residence prior to admission was categorized as home vs a non-home facility such as assisted living, subacute or acute rehabilitation, or long-term skilled nursing facility. Comorbidity index was measured using the Charlson comorbidity index (CCI) excluding the age component of the score. 16 Severity of illness on presentation was represented by initial oxygen delivery requirement (room air, nasal cannula, nonrebreather mask, and, IMV and the Modified Early Warning Score (MEWS), a validated tool scored from 1 to 13, indicating non-severe illness (1–4) and severe illness (5–13). 17
Outcomes
The primary outcome of interest was difference in hospital mortality between early and non-early DNR groups, defined as death during hospitalization from any cause during the study period. Secondary outcomes included indirect hospital morbidity indicators, including: presence of inpatient palliative care consultation, IMV; critical illness (defined as an EHR order for vasopressors or sedative infusions), delirium (defined as any EHR order for benzodiazepines and/or antipsychotics, physical restraints or constant observation used for behavioral symptoms), new renal replacement therapy, index hospital LOS, 30-day readmission and composite of 30-day readmission or inpatient mortality. For those who arrived from home, discharge to a non-home, non-hospice facility as a measure of decreased functional status was also evaluated. Finally, a secondary analysis was conducted comparing characteristics and outcomes of patients with early vs late DNR.
Statistical Analysis
Patient demographics, baseline characteristics and severity of illness were summarized using descriptive statistics. Categorical variables were summarized using frequencies and percentages; continuous variables were summarized using means and standard deviations. Subjects with missing data were excluded from data analyses without imputation. Categorical variables were compared between the DNR groups using Chi-squared or Fisher’s exact test; and continuous variables were compared between the DNR groups using two-sample t-test or Kruskal–Wallis test as appropriate. Logistic or linear regression analyses were performed for binary or continuous outcome variables, respectively.
Statistically significant variables (P < .05) in the univariate analysis were included in the multivariable analyses of outcomes between early vs non-early and early vs late DNR groups. Do-Not-Resuscitate group was forced into the multivariable model as the variable of interest. The final models for multivariable analysis were chosen using a backward selection method among the candidate variables based on the univariable analysis. A secondary analysis compared baseline characteristics and outcomes between early and late DNR groups. All analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).
Results
Overall Sample
Baseline Characteristics of Total Cohort and Comparison Between Early DNR and Non-Early DNR Groups.
a DNR = Do Not Resuscitate.
b MEWS = Modified Early Warning Score.
Patient Characteristics Associated with Early DNR vs Non-Early DNR
Patients in the early DNR group, as compared to non-early DNR, were older (mean age 85.0 vs 76.8, P < .001), more likely to be women (51% vs 44%, P = .01), come from a non-home facility (49% vs 19%, P < .001), and have dementia (13% vs 5%, P < .001). Race was associated with early DNR status at an aggregate level (chi-square test P < .001), with a higher proportion of white patients in the early DNR group vs non-early DNR group (58% vs 46.4%) as compared to Black subjects (11.3% vs 21.5%). There was no significant association between early DNR and ethnicity (P = .42), or language (P = .62). (Table 1)
Regarding severity of illness on presentation, a higher illness severity (MEWS 5-13) was seen in those with early DNR compared with non-early DNR (58% vs 32%, P < .001). There was no association between level of oxygen support on arrival and presence of early DNR (P = .36).
The early DNR group was also more likely to have an inpatient palliative care consultation than those with non-early DNR (26% vs 20%, P = .02).
Multivariable Analysis of Early vs Non-Early DNR and Hospital Outcomes
Outcomes of Total Cohort and Comparisons Between Early and Non-Early DNR Groups.
a DNR = Do Not Resuscitate.
Patient Characteristics Associated with Early vs Late DNR
Patient Characteristics of Early vs Late DNR.
a DNR = Do Not Resuscitate.
b MEWS = Modified Early Warning Score.
Early DNR vs Late DNR and Hospital Outcomes
Patient Outcomes of Early vs Late DNR.
a DNR = Do Not Resuscitate.
b LOS = Length of Stay.
Discussion
To our knowledge, this is the first study examining the associations of DNR status with mortality and indirect morbidity indicators in OAs hospitalized with SARS-CoV-2 infection. The prevalence of early DNR was low (5.7%) and associated with higher hospital mortality when compared to non-early DNR patients, even when controlling for patient demographics and severity of illness. Hospital mortality was similarly high among late DNR patients, as late DNR may be a more likely request in the setting of deteriorating course and longer LOS. The lower relative mortality in the non-early DNR group was likely driven by factors associated with those subjects that had no DNR orders whatsoever. Importantly, patients with early DNR experienced less morbidity and were more likely to receive a palliative care consultation compared to non-early DNR patients.
Pre–COVID-19 studies found that the rate of early DNR orders for hospitalized adults ranged between 10 and 20% for those admitted to a medical service.7,9,13,17,18 We found an early DNR rate of only 5.7%. Several hypotheses lend themselves to these lower rates. A sudden increase in the “denominator” of OAs hospitalized during the first surge of SARS-CoV-2 infection left little or no time to have meaningful goals of care or code-status conversations prior to admission. Inundated with acute patients, hospitals may not have had time to devote to high-quality discussions of life-sustaining treatment preferences. Furthermore, visitation restrictions likely precluded high-quality discussions with patients and their surrogates, biasing decision-makers toward a default of full code. 19 Overestimating survivability due to a lack of contemporaneous, high-quality data on inpatient post-arrest mortality in COVID-19 may have delayed code discussions between providers, patients and caregivers. Last, fears that early DNR orders will lead to less access to scarce resources (like ventilators) or less overall care may have also contributed. 20 Further study is warranted to investigate these hypotheses, and to reassess baseline resuscitation preferences through the evolution of the pandemic.
Pre–COVID-19 DNR studies in the context of specific medical conditions (e.g., septic shock, pneumonia, ischemic stroke), out-of-hospital cardiac arrest, critical illness, and general medical populations, found early DNR patients tended to be older, white, women, of higher economic status, with higher chronic comorbidity burden, and more likely to have baseline dementia.8,11-13,21 In addition to confirming the above findings in COVID-19, our study adds to this literature by indicating that coming from a facility was associated with an early DNR. This may reflect greater baseline cognitive and functional impairments and a greater likelihood of having discussed advanced directives with such patients. With respect to race, a pre–COVID-19 study by Richardson and colleagues found Black and Asian patients less likely to have early DNR orders after admission for out-of-hospital cardiac arrest. 11 While our study did find a significant association between race and early DNR in general, further study is needed to assess associations between specific races and early DNR. Regarding severity of illness, while higher MEWS was associated with early DNR, initial oxygen requirements was not. Derangements in overall vital signs as suggested by MEWS, as opposed to oxygen requirement alone may have been more of an impetus to discuss code status in COVID-19.
When controlling for baseline differences, we found that early DNR was associated with higher hospital mortality when compared to non-early DNR, a finding anticipated by previous DNR research and more recently in the context of COVID-19 by Alhatem and colleagues. 22 Pre–COVID-19 studies have also found that even when controlling for demographics, baseline comorbidities, admission diagnosis, and presenting severity of illness, early DNR status is a strong independent predictor of higher mortality.7,11,13 One reason for this association may be that early DNR patients receive fewer potentially life-saving interventions (“failure to rescue” phenomenon). 11 While DNR status precludes resuscitation for cardiac arrest, it does not preclude any other interventions, warranting an examination of the influence of DNR on provider decision making. Clinical vignette-based survey studies found that physicians and nurses were significantly less likely to offer diagnostic and therapeutic interventions and less likely to escalate concerns or increase monitoring in critically ill patients based solely on the presence of a DNR order.23-25 The tumult of a pandemic surge may potentially exacerbate such dynamics, but it is also possible that providers are uncovering preferences for fewer interventions that are in line with overall patient goals; these discussions and associated documentation practices by providers warrant further study. Although we found no difference between early and non-early DNR groups with respect to need for critical care, our critical care variable included the use of vasopressors, which non-intubated patients may have received outside of the ICU. Furthermore, our data did not account for critical care patients who subsequently underwent withdrawal of care and were transferred out of the ICU.
Our secondary analysis revealed that relative to early DNR, patients with late DNR were more likely to come from home (as opposed to a facility) and less likely to have documented dementia. This group was more likely to experience hospital morbidity without a mortality benefit. These findings highlight the propensity of patients who may have higher pre-hospital function to have DNR ordered later in their hospital course, reflecting a deferral of life-sustaining treatment preference discussions by providers, a reluctance to discuss or accept DNR on the part of decision-makers until later in the hospital course, or some combination of both. The rationale of patients and decision-makers in choosing when to discuss life-sustaining treatment preferences is undoubtedly complex and further studies are needed to investigate these dynamics.
Limitations
Our study is not without limitations, including its retrospective design and use of administrative EHR data, which does not capture all confounding variables that potentially impact mortality. The use of administrative EHR data implies the use of surrogate markers, such as MEWS, which has not yet been evaluated in COVID-19. Given the retrospective nature of the study, we could not delineate which early DNR decisions arose pre-hospitalization or within 24 hours of admission, nor could we verify whether code status was reviewed for all patients on admission. In addition, the study was not able to assess the quality or number of code-status discussions that were had, only whether an end-result of DNR was reached. Our data set also precludes any analysis of patients who had early DNR orders rescinded and of patients with DNR status on admission who were missed because their DNR orders were not written until after 24 hours of hospitalization. Last, we could not distinguish between DNRs that were ordered late due to delays on the part of providers vs decision-makers.
Conclusion
To our knowledge, this is the first study to evaluate the role of DNR in OAs hospitalized with COVID-19 during the initial pandemic peak. Early DNR was associated with higher mortality but lower morbidity. Establishing DNR late during hospitalization did not affect mortality in OAs but led to more aggressive interventions that were associated with higher morbidity. Determining code status for OAs is wrought with tension between an ethical imperative to establish a timely DNR in appropriate patient candidates and fears that this will adversely impact care. This tension has been brought to the forefront of our collective consciousness due to the COVID-19 pandemic. Future studies should evaluate the role of early DNR in achieving goal-concordant care in our current and future pandemic crises.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Data Availability Statement
The datasets used and/or analyzed are available from the corresponding author on reasonable request.
Disclosure Statements
The corresponding authors had full access to all data in the study and had final responsibility for the decision to submit for publication. The following material has not been published or submitted for publication elsewhere.
