Abstract
Background
At the beginning of the COVID-19 pandemic, whether performing CPR on patients with COVID-19 would be effective or increase COVID-19 transmission to health care workers was unclear.
Research Question
Did the prevalence of do-not-resuscitate (DNR) orders by COVID-19 status change over the first year of the pandemic as risks such as COVID-19 transmission to health care workers improved?
Study Design and Methods
This cross-sectional study assessed DNR orders for all adult patients admitted to ICUs at two academic medical centers in Chicago, IL, between April 2020 and April 2021. DNR orders by COVID-19 status were assessed using risk-adjusted mixed-effects logistic regression and propensity score matching by patient severity of illness.
Results
The study population of 3,070 critically ill patients were 46% Black, 53% male, with median age (interquartile range [IQR]) 63 (50-73) years. Eighteen percent were COVID-19 positive and 27% had a DNR order. Black and Latinx patients had higher absolute rates of DNR orders than White patients (30% vs 29% vs 23%; P = .006). After adjustment for patient characteristics, illness severity, and hospital location, DNR orders were more likely in patients with COVID-19 in the nonpropensity score-matched (n = 3,070; aOR, 2.01; 95% CI, 1.64-2.38) and propensity score-matched (n = 1,118; aOR, 1.91; 95% CI, 1.45-2.52) cohorts. The prevalence of DNR orders remained higher for patients with COVID-19 than patients without COVID-19 during all months of the study period (difference in prevalence over time, P = .751).
Interpretation
In this multihospital study, DNR orders remained persistently higher for patients with COVID-19 vs patients without COVID-19 with similar severity of illness during the first year of the pandemic. The specific reasons why DNR orders remained persistently elevated for patients with COVID-19 should be assessed in future studies, because these changes may continue to affect COVID-19 patient care and outcomes.
Key Words: COVID-19, do-not-resuscitate, health disparities
Take-home Points.
Study Question: Did the prevalence of do-not-resuscitate (DNR) orders by COVID-19 status change over the first year of the pandemic as risks such as COVID-19 transmission to health care workers improved?
Results: In our study population of 3,070 critically ill patients, DNR orders were more likely in patients with COVID-19 in the nonpropensity score-matched (n = 3,070; aOR, 2.01; 95% CI, 1.64-2.38) and propensity score-matched (n = 1,118; aOR, 1.91; 95% CI, 1.45-2.52) cohorts, and the prevalence of DNR orders remained higher for patients with COVID-19 than for patients without COVID-19 during all months of the 1-year study period (difference in prevalence over time, P = .751).
Interpretation: In this multihospital study, DNR orders remained persistently higher for patients with COVID-19 compared with patients without COVID-19 with similar severity of illness during the first year of the pandemic.
At the beginning of the COVID-19 pandemic, there was significant concern clinicians might acquire COVID-19 from patient care, especially for clinicians without adequate access to personal protective equipment or working in high-risk settings involving aerosolizing procedures such as CPR.1, 2, 3, 4, 5, 6, 7 For patients experiencing an in-hospital cardiac arrest, whether performing CPR for these patients would be effective or increase risk of COVID-19 transmission to health care workers was unclear.8,9 There was concern that patients with COVID-19 would have higher mortality rates after in-hospital cardiac arrest than other patients,10 potentially putting health care workers at risk from a procedure with a low likelihood of success. These concerns may have contributed to an elevation in prevalence of do-not-resuscitate (DNR) orders for patients with COVID-19, noted to exceed 60% in the first months of the pandemic.11
How the prevalence of DNR orders in patients with COVID-19 was affected throughout the first year of the pandemic is unknown. Although previous studies have assessed the prevalence of DNR orders in patients with COVID-19 in the first months of the pandemic,11,12 studies comparing prevalence of DNR orders in patients with and without COVID-19 across the first year of the pandemic have limitations, such as only assessing a single medical center.13 It is possible DNR prevalence for patients with COVID-19 may have changed over time, similar to patients with AIDS, who had higher prevalence of DNR orders when compared with patients with other diseases at the beginning of the AIDS epidemic, which did not persist over time.14,15 Factors that may influence DNR order placement, such as risk for COVID-19 transmission to health care workers during CPR and elevated risk of mortality for patients admitted with COVID-19, changed during the first year of the pandemic. For example, personal protective equipment and COVID-19 vaccines, which were scarce and absent at the beginning of the pandemic, became readily available to hospital staff by the end of the first year of the pandemic.16 In addition, mortality for hospitalized patients with COVID-19 in the ICU decreased during the study period, from 16% before May 2020 to 9% in July-September 2021, based on data from one Midwest hospital.17,18 Changes in these factors may have affected prevalence of DNR order placement for patients with COVID-19 differently at the beginning than at the end of the first year of the pandemic.
We assessed how prevalence of DNR order placement in patients with COVID-19 changed over the first year of the pandemic. We hypothesized prevalence of DNR orders for patients with COVID-19 were significantly higher during the beginning of the pandemic when compared to patients without COVID-19. We also hypothesized that as COVID-19 mortality improved and risk for COVID-19 transmission to health care workers decreased, prevalence of DNR orders placed for patients with COVID-19 would approach the prevalence in patients without COVID-19. This study is important because if patients with COVID-19 are found to have DNR prevalence different from patients without COVID-19 with similar severity of illness over time, this would support further study into exploring possible institutional, social, or personal reasons contributing to these findings, which may continue to affect COVID-19 patient care and outcomes.
Study Design and Methods
Study Participants
We completed a retrospective cross-sectional assessment evaluating all adult patients admitted to an ICU at two academic, urban medical centers, Rush University and the University of Chicago, between April 2020 and April 2021. The study dates were chosen to evaluate DNR orders across multiple surges of the first year of the pandemic, including periods with differing COVID-19 patient mortality rates, periods of adequate and scarce resources, and the time before and after COVID-19 vaccine availability.19,20 The two independent study hospitals both had more than 100 ICU beds and cared for 25% of all ICU-admitted patients with COVID-19 in the state of Illinois at the peak of the pandemic. During the beginning of the study period, increased admissions required expansion of ICU beds staffed by clinicians who do not normally care for adult ICU patients. All patients admitted to any ICU at each hospital were eligible for inclusion in the study analysis to select for patients with high severity of illness with high necessity for code status discussions. The Rush University and University of Chicago Institutional Review Boards granted exemption for this project (IRB approval number 20062102). This study followed Strengthening the Reporting of Observational Studies in Epidemiology guidelines.21
Patient Measures
We evaluated patient demographics, including age, sex, race/ethnicity, primary language, insurance status, and marital status. We assessed medical information about each patient, including DNR status during hospital admission, COVID-19 positive status, admission date, length of admission, intubation status, use of vasopressors or inotropes, and death during admission or discharge with hospice. We combined the outcomes of death during admission and discharge to home hospice or inpatient hospice in our analysis to create a category including deceased patients and patients with near imminent mortality, similar to previous studies.22, 23, 24 We chose not to include calculated measures of severity of illness such as the Sequential Organ Failure Assessment score in our study as we aimed to assess racial differences in DNR order use, and Sequential Organ Failure Assessment scores may overestimate severity of illness for Black patients.25 We instead used intubation status and vasopressor or inotrope status as measures of severity of illness.
Statistical Analysis
We completed analysis of categorical variables, using quantity and percent or median and interquartile range. We assessed absolute rates of DNR orders in the nonpropensity score-matched cohort by patient demographics, severity of illness, patient outcomes, and time of admission. We used Fisher exact and χ2 tests as appropriate, using two-sided tests to compare patient characteristics and outcomes. P ≤ .05 was determined to be statistically significant.
Statistical analysis was performed using R Core Team (2019; R: A language and environment for statistical computing. R Foundation for Statistical Computing)26,27 and GraphPad Prism (version 9.0 for Mac, GraphPad Software).
We assessed the outcome of differences in DNR prevalence by COVID-19 status throughout the first year of the COVID-19 pandemic using nonpropensity score-matched and propensity-matched cohorts of patients. We fit mixed-effects logistic regression models with DNR status as the outcome to estimate the effect COVID-19-positive status has on DNR orders using both the nonpropensity score-matched and propensity-matched cohorts. We assessed multiple models including a full cohort nonpropensity score-matched group with DNR risk adjusted for age, intubation and pressor or inotrope status, and hospital location. We also assessed a full cohort nonpropensity score-matched group with DNR risk adjusted for age, intubation and pressor/inotrope status, race, language, insurance status, and hospital location. We assessed the main propensity score-matched cohort with DNR risk adjusted for race, language, insurance status, and hospital location. We also assessed the additional propensity-matched cohort, including exact matching of admission month with DNR risk adjusted for race, language, insurance status, and hospital location. We used cluster SE by hospital location to determine 95% CIs.
The main propensity-matched cohort for COVID-19 status was developed using the nearest neighbor approach matching for age, intubation status, and vasopressor/inotrope status. We chose these variables as predictors for acute patient severity of illness. We used 1:1 matching to assess patients with and without COVID-19 with similar severity of illness because our objective was to assess DNR prevalence among these two groups. Standardized differences were less than 0.1 for confounding variables in the matched groups for this main model, which confirms the even distribution of propensity scores.
We also developed an additional propensity-matched cohort adding exact matching of admission month to nearest neighbor matching of age, vasopressor/inotrope status, and intubation status. Evaluating a propensity-matched cohort with exact matching by admission month in addition to the main propensity-matched model was important to compare patients at similar timepoints during the study period because factors that may affect DNR order placement changed during the first year of the pandemic, such as patient mortality and likelihood of COVID-19 transmission to health care workers.
We assessed for changes in differences between DNR prevalence by COVID-19 status over time, using analysis of variance to compare logistic regression models with and without the interaction of COVID-19 status and admission month. We completed pairwise comparisons of the estimated marginal means of this analysis to assess for specific differences in DNR prevalence by month of admission throughout the study period.
Results
Study Population
A total of 3,070 patients were admitted to an ICU at the two study hospitals between April 2020 to April 2021 and were included in study analysis. Patients had median age (interquartile range) of 63 (50-73) years. Most patients were Black (46%) or White (29%), English speaking (86%), and male (53%). Sixteen percent of patients were Hispanic or Latinx, and 9% were primary Spanish speaking. Eighteen percent were COVID-19 positive, 27% had a DNR order placed during hospital admission, and 37% died or were discharged with hospice after admission.
Univariate Analysis and Nonpropensity Score-Matched Full Cohort Prevalence of DNR Orders
Patient demographics associated with a higher absolute rate of DNR order include patients who were older than 50 years of age (19% age ≥ 50 vs 14% age < 49; P ≤ .0001), patients who were Spanish speaking (39% Spanish vs 25% English; P ≤ .0001), and patients lacking medical insurance or patients with Medicare (42% no insurance, 31% Medicare, 23% Medicaid, 21% private insurance; P ≤ .0001). Patients who were White had a lower absolute rate of DNR order placement (30% Black vs 29% Hispanic or Latinx vs White 23%; P = .006). Medical factors associated with higher absolute rate of DNR order placement include death during admission or discharge with hospice (64% vs 5%; P ≤ .0001), receipt of intubation (46% vs 17%; P ≤ .0001), use of vasopressors/inotropes (44% vs 10%; P ≤ .0001), and admission for longer than 20 days (35% ≥ 20 days vs 25% < 20 days; P ≤ .0001) (Table 1). Time of admission was associated with higher absolute rate of DNR order placement for all patients in the months of April through June 2020 (36%) and October through December 2020 (27%) when compared with July through September 2020 (22%), correlating with higher rates of COVID-19-positive patients in the ICU during these periods (e-Fig 1). Comparing odds of DNR order placement by COVID-19 status throughout the study period, patients with COVID-19 had higher odds of DNR order placement across all periods, including April through June 2020 (OR, 3.8; 95% CI, 2.7-5.2), July through September 2020 (OR, 1.9; 95% CI, 1.1-3.3), October through December 2020 (OR, 2.9; 95% CI, 2.0-4.2), and January through April 2021 (OR, 3.0; 95% CI, 2.0-4.6) (Fig 1). Odds of DNR order placement for all patients during the study period was lower at Hospital B when compared with Hospital A (21% vs 48%; OR, 0.3; 95% CI, 0.24-0.34; Table 2, e-Fig 2).
Table 1.
Patient Characteristics for All ICU Admissions by DNR Status
| Characteristic | No DNR (N = 2,250) | DNR (N = 820) |
|---|---|---|
| Age, y | ||
| < 49 | 659 (29) | 103 (13) |
| 50-69 | 922 (41) | 369 (45) |
| 70-89 | 628 (28) | 325 (40) |
| ≥ 90 | 41 (2) | 23 (3) |
| Sex | ||
| Female | 1,044 (46) | 393 (48) |
| Male | 1,206 (54) | 427 (52) |
| Race/ethnicity | ||
| Black or African American | 1,040 (46) | 384 (47) |
| Hispanic or Latinx | 377 (17) | 156 (19) |
| Other | 150 (7) | 75 (9) |
| White | 683 (30) | 205 (25) |
| Married | 880 (39) | 309 (38) |
| Primary language | ||
| English | 1,986 (88) | 656 (80) |
| Spanish | 184 (8) | 116 (14) |
| Other | 80 (4) | 48 (6) |
| Insurance status | ||
| Private | 561 (25) | 149 (18) |
| Medicaid | 567 (25) | 172 (21) |
| Medicare | 1,011 (45) | 458 (56) |
| Other | 81 (4) | 19 (2) |
| No insurance | 30 (1) | 22 (3) |
| Death or discharge with hospice | 407 (18) | 730 (89) |
| Intubated | 575 (26) | 486 (59) |
| Received vasopressor/inotrope | 818 (36) | 655 (80) |
| COVID-19 positive | 288 (13) | 271 (33) |
| Time of admission | ||
| April 2020-June 2020 | 483 (21) | 271 (33) |
| July 2020-September 2020 | 602 (27) | 171 (21) |
| October 2020-December 2020 | 525 (23) | 192 (23) |
| January 2021-April 2021 | 640 (28) | 186 (23) |
| Hospital length of stay | ||
| ≥ 20 d | 338 (15) | 183 (22) |
| Hospital | ||
| Hospital A | 357 (16) | 325 (40) |
| Hospital B | 1,893 (84) | 495 (60) |
Data are presented as No. (%). DNR = do-not-resuscitate.
Figure 1.
Do-not-resuscitate orders by COVID-19 status for patients in the ICU over time. DNR = do-not-resuscitate.
Table 2.
OR of DNR Order Placement for Full Nonpropensity Score-Matched Cohort by Patient Characteristics
| Characteristic | OR Comparing Patients With DNR Orders With Patients Without DNR Orders (95% CI) |
|---|---|
| Age, y | |
| < 49 | Reference |
| 50-69 | 2.6 (2.01-3.26) |
| 70-89 | 3.3 (2.59-4.24) |
| ≥ 90 | 3.6 (2.07-6.23) |
| Sex | |
| Female | Reference |
| Male | 0.9 (0.80-1.10) |
| Race/ethnicity | |
| Black or African American | Reference |
| Hispanic or Latinx | 1.1 (0.90-1.40) |
| Other | 1.4 (1.00-1.83) |
| White | 0.8 (0.67-0.99) |
| Married | 0.9 (0.80-1.11) |
| Primary language | |
| English | Reference |
| Spanish | 1.9 (1.49-2.45) |
| Other | 1.8 (1.26-2.62) |
| Insurance status | |
| Private | Reference |
| Medicaid | 1.1 (0.89-1.46) |
| Medicare | 1.7 (1.38-2.11) |
| Other | 0.9 (0.52-1.50) |
| No insurance | 2.8 (1.55-4.92) |
| Death or discharge with hospice | 36.8 (28.78-46.87) |
| Intubated | 4.2 (3.58-5.02) |
| Received vasopressor/inotrope | 6.9 (5.74-8.41) |
| COVID-19 positive | 3.4 (2.78-4.07) |
| Time of admission | |
| April 2020-June 2020 | 2.0 (1.58-2.48) |
| July 2020-September 2020 | Reference |
| October 2020-December 2020 | 1.3 (1.02-1.63) |
| January 2021-April 2021 | 1.0 (0.81-1.29) |
| Hospital length of stay | |
| ≥ 20 d | 1.6 (1.33-1.99) |
| Hospital | |
| Hospital A | Reference |
| Hospital B | 0.3 (0.24-0.34) |
Bold indicates statistical significance, P < .05. DNR = do-not-resuscitate.
Propensity-Matched Cohorts
We assessed propensity score-matched cohorts for patients with and without COVID-19 because of the differences in severity of illness between these two groups in the nonpropensity score-matched cohort. The propensity score-matched cohort (matched for age, intubation status, and vasopressor/inotrope status) had 0 standard mean difference between the COVID-19 and non-COVID-19 groups, indicating balanced matching because the standard mean difference was < 0.128 (e-Table 1, e-Fig 3). We completed an additional propensity score-matched analysis to account for possible changes in treatment of patients with COVID-19 over time, adding exact matching by admission month to the propensity score-matched model (e-Fig 4).
Risk-Adjusted Outcomes
In the full nonpropensity score-matched and propensity score-matched cohorts, patients with COVID-19 had higher risk-adjusted prevalence of DNR orders, death or discharge with hospice, and hospital length of stay greater than 20 days than patients without COVID-19 (Fig 2). In the full nonpropensity score-matched cohort, after adjusting for patient characteristics, hospital location, and severity of illness, risk-adjusted DNR orders were higher for patients positive for COVID-19 (n = 3,070; aOR, 2.01; 95% CI, 1.64-2.38). In the propensity score-matched cohort after adjusting for patient characteristics and hospital location, risk-adjusted DNR orders were higher for patients positive for COVID-19 (n = 1,118; aOR, 1.91; 95% CI, 1.45-2.52). An additional propensity score-matched cohort adding exact matching by admission month was also associated with higher DNR orders for patients with COVID-19 (n = 1,066; aOR, 2.03; 95% CI, 1.54-2.68) (Fig 3). We did not find significant interaction between COVID-19 status and language nor COVID-19 status and wave of the pandemic in our assessed outcomes.
Figure 2.
Adjusted ORs for patient outcomes by COVID-19 positive status for the nonpropensity score-matched and propensity score-matched cohorts. Adjusted ORs for the nonpropensity score-matched cohorts was calculated with mixed-effects logistic regression and cluster SE adjusted for age, intubation and pressor/inotrope status, race/ethnicity, language, insurance, and hospital location. Adjusted ORs for the propensity score-matched cohorts include age, intubation and pressor/inotrope status, race/ethnicity, language, insurance, and hospital location. DNR = do-not-resuscitate.
Figure 3.
Mixed-effects logistic regression analyses of DNR order placement for ICU patients by COVID-19 status. Model 1: Full nonpropensity score-matched group with DNR risk adjusted for age, intubation and pressor/inotrope status, and hospital location. Model 2: Full nonpropensity score-matched group with DNR risk adjusted for age, intubation and pressor/inotrope status, race, language, insurance status, and hospital location. Model 3: Propensity score-matched cohort with DNR risk adjusted for age, intubation and pressor/inotrope status, language, insurance status, and hospital location. Model 4: Repeated propensity score matching with exact matching by admission month and nearest neighbor matching for age, vasopressor/inotrope status, and intubation status and used this propensity score-matched cohort with DNR risk adjusted for age, intubation and pressor/inotrope status, race, language, insurance status, and hospital location. DNR = do-not-resuscitate.
Differences in Prevalence of DNR Orders by COVID-19 Status Over Time
When assessing differences in rates of DNR orders by COVID-19 status over time, these rates did not statistically change across each month of the 1-year study period (P = .751). Similar differences in DNR order prevalence by COVID-19 positive or negative status were observed in the beginning of the study period (April 2020, 60% vs 39%; difference, 21%) vs the end of the study period (April 2021; 32% vs 11%; difference, 21%).
Discussion
Of the patients admitted to the ICU at both study hospitals during the first year of the COVID-19 pandemic, 27% had a DNR order placed during admission. Hospital A had a significantly higher prevalence of patients with DNR orders during this period than Hospital B. Risk-adjusted DNR orders were higher for patients with COVID-19 when compared with patients without COVID-19 in the full nonpropensity score-matched and propensity score-matched cohorts throughout the first year of the pandemic. Although we also hypothesized DNR prevalence in patients with COVID-19 would approach a prevalence similar to that of patients without COVID-19 because risk of COVID-19 transmission and mortality decreased over time, our findings instead showed sustained higher prevalence of DNR orders for patients with COVID-19 throughout the first year of the pandemic.
This study adds to previous literature in that it compares DNR status in patients with and without COVID-19 at multiple centers across the first year of the pandemic using nonpropensity score-matched and propensity score-matched cohorts. Previous studies have evaluated DNR orders by COVID-19 patient status but are limited in that they evaluated patients at a single center13 or primarily assessed characteristics associated with DNR orders in the first months of the pandemic rather than assess DNR placement throughout the first year of the pandemic and differences in DNR placement over time.11,12 This study also adds to previous literature in that it uses propensity score-matched cohorts in addition to nonpropensity score-matched cohorts, minimizing bias from confounders that may contribute to differences in DNR prevalence between patients with and without COVID-19.
We found patients with COVID-19 had a higher prevalence of DNR orders when compared with patients without COVID-19 with similar severity of illness. One reason for this finding may be that clinicians made changes to how they discussed code status with patients with COVID-19 to nudge patients to accept DNR status because of concerns such as resource scarcity, reduced effectiveness of CPR in this population, and risk of COVID-19 transmission to health care workers during CPR. Clinician changes to code status conversations for patients with COVID-19 may have included increasing conversation frequency, emphasizing the low likelihood of patient recovery with CPR, and explicitly recommending patients to not pursue CPR, influencing the likelihood of patients to accept a DNR status.10 Hospital culture likely contributed to how clinicians approached discussing code status with patients; we found significantly higher odds of DNR orders for patients at Hospital A vs Hospital B, consistent with our past findings.29 One reason for this difference may be that Hospital A provided unified multidisciplinary recommendations for hospital staff describing how to approach code status for patients with COVID-19, which did not occur at Hospital B.
The sustained elevation in prevalence of DNR orders for patients with COVID-19 when compared with patients without COVID-19 with similar severity of illness is surprising given that factors we hypothesized may increase DNR prevalence in patients with COVID-19, such as risk of COVID-19 transmission and COVID-19 mortality, decreased during the study period.18,29,30 It is possible availability biases in clinicians encouraged continuation of their behavior to increase code status discussions and recommendations for DNR orders in patients with COVID-19, contributing to sustained higher prevalence of DNR orders in patients with COVID-19, even as risks of COVID-19 transmission and mortality decreased.31, 32, 33 Availability bias occurs when clinicians disproportionately emphasize information that readily comes to mind rather than rationally evaluating data.34 Clinician experiences seeing coworkers become infected with COVID-19 and witnessing frequent deaths in patients with COVID-19 at the beginning of the pandemic may have contributed to availability bias that the risk of COVID-19 transmission and mortality was higher than it actually was 1 year into the pandemic. As a counterpoint to this argument, it is possible clinicians were not convinced by data suggesting the risk of COVID-19 transmission and COVID-19 mortality was decreasing and thus did not change their practice in discussing code status with patients. Although the risk of COVID-19 transmission and severe illness decreased after COVID-19 vaccines were available to health care workers during the study period,29,30 15% of Chicago health care workers remained vaccine hesitant 1 year into the pandemic,35 and ambiguity surrounding the long-term effects of COVID-19 transmission remained.36 In addition, although mortality rates for patients with COVID-19 in the ICU decreased during the study period,18 other data published at the end of our study period continued to identify higher rates of death after in-hospital cardiac arrest for patients with COVID-19 vs those without COVID-19.10
Further research is needed to identify reasons why patients with COVID-19 had sustained higher prevalence of DNR orders than patients without COVID-19 with similar severity of illness across the first year of the pandemic. Although we propose our findings may be associated with institutional factors or cognitive biases contributing to sustained clinician behavior to increase discussion, documentation, and recommendation of DNR status for patients with COVID-19, we lack data to support this proposition. In addition, because we only assessed the first year of the pandemic, future research should assess whether the higher sustained prevalence of DNR orders for patients with vs without COVID-19 with similar severity of illness persists to this day.
Limitations
We assessed patient illness severity in the ICU by the delivery of invasive mechanical ventilatory support and receipt of vasopressors/inotropes. We used binary assessments of the presence or absence of these measures, inhibiting our ability to adjust for the considerable range of illness severity within patients who received mechanical ventilation and vasopressors/inotropes. In addition, our measurement of severity of illness is limited in that it did not account for chronic comorbidities that may affect patient and surrogate choice for DNR status. We did not control for patient or surrogate attitudes or beliefs that may affect their choice for DNR status, such as religious or spiritual beliefs.37 Although this study included two independent academic medical centers caring for racially and socioeconomically diverse populations, it is limited in that both hospitals had similar geographic locations in Chicago, IL. We did find significant variation in the overall prevalence of DNR orders across hospitals, whereas the prevalence of DNR orders for patients with COVID-19 was consistently higher than that for patients without COVID-19 in both hospitals, however, increasing the generalizability of our overall results.
Interpretation
In this multihospital study, DNR orders remained persistently higher for patients with COVID-19 compared with patients without COVID-19 with similar severity of illness during the first year of the pandemic. Identifying specific reasons why DNR orders remained persistently higher for patients with COVID-19 is important, because these reasons may be associated with institutional practices or clinician biases that may continue to affect COVID-19 patient care and outcomes.
Funding/Support
G. M. P. was supported by the National Institures of Health (NIH [Grant No. 5TL1TR001858]). W. F. P. was supported by NIH (Grant Nos. K08 HL150291 and R01LM014263).
Financial/Nonfinancial Disclosures
None declared.
Acknowledgments
Author contributions: G. M. P. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. G. M. P. and W. F. P. contributed substantially to the study design, data analysis and interpretation, and the writing of the manuscript.
Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript.
Other contributions: We are grateful to Robert Arnold, MD, and Yael Schenker, MD, for their editing assistance on this project.
Additional information: The e-Figures and e-Table are available online under "Supplementary Data."
Supplementary Data
References
- 1.González-Gil M.T., González-Blázquez C., Parro-Moreno A.I., et al. Nurses’ perceptions and demands regarding COVID-19 care delivery in critical care units and hospital emergency services. Intensive Crit Care Nurs. 2021;62 doi: 10.1016/j.iccn.2020.102966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wang X., Ferro E.G., Zhou G., Hashimoto D., Bhatt D.L. Association between universal masking in a health care system and SARS-CoV-2 positivity among health care workers. JAMA. 2020;324(7):703–704. doi: 10.1001/jama.2020.12897. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Venugopal U., Jilani N., Rabah S., et al. SARS-CoV-2 seroprevalence among health care workers in a New York City hospital: a cross-sectional analysis during the COVID-19 pandemic. Int J Infect Dis. 2021;102:63–69. doi: 10.1016/j.ijid.2020.10.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gómez-Ochoa S.A., Franco O.H., Rojas L.Z., et al. COVID-19 in health-care workers: a living systematic review and meta-analysis of prevalence, risk factors, clinical characteristics, and outcomes. Am J Epidemiol. 2021;190(1):161–175. doi: 10.1093/aje/kwaa191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Norful A.A., Rosenfeld A., Schroeder K., Travers J.L., Aliyu S. Primary drivers and psychological manifestations of stress in frontline healthcare workforce during the initial COVID-19 outbreak in the United States. Gen Hosp Psychiatry. 2021;69:20–26. doi: 10.1016/j.genhosppsych.2021.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wilkins J.T., Gray E.L., Wallia A., et al. Seroprevalence and correlates of SARS-CoV-2 antibodies in health care workers in Chicago. Open Forum Infect Dis. 2021;8(1):ofaa582. doi: 10.1093/ofid/ofaa582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Firew T., Sano E.D., Lee J.W., et al. Protecting the front line: a cross-sectional survey analysis of the occupational factors contributing to healthcare workers’ infection and psychological distress during the COVID-19 pandemic in the USA. BMJ Open. 2020;10(10) doi: 10.1136/bmjopen-2020-042752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Couper K., Taylor-Phillips S., Grove A., et al. COVID-19 in cardiac arrest and infection risk to rescuers: a systematic review. Resuscitation. 2020;151:59–66. doi: 10.1016/j.resuscitation.2020.04.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nolan J.P., Monsieurs K.G., Bossaert L., et al. European Resuscitation Council COVID-19 guidelines executive summary. Resuscitation. 2020;153:45–55. doi: 10.1016/j.resuscitation.2020.06.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ippolito M., Catalisano G., Marino C., et al. Mortality after in-hospital cardiac arrest in patients with COVID-19: a systematic review and meta-analysis. Resuscitation. 2021;164:122–129. doi: 10.1016/j.resuscitation.2021.04.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Alhatem A., Spruijt O., Heller D.S., Chokshi R.J., Schwartz R.A., Lambert W.C. “Do-not-resuscitate (DNR)” status determines mortality in patients with COVID-19. Clin Dermatol. 2021;39(3):510–516. doi: 10.1016/j.clindermatol.2020.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Moin E.E., Okin D., Jesudasen S.J., et al. Code status orders in patients admitted to the intensive care unit with COVID-19: a retrospective cohort study. Resusc Plus. 2022;10:100219. doi: 10.1016/j.resplu.2022.100219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Jacobson E., Troost J.P., Epler K., et al. Change in code status orders of hospitalized adults with COVID-19 throughout the pandemic: a retrospective cohort study. J Palliative Med. 2023;26(9):1188–1197. doi: 10.1089/jpm.2022.0578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wachter R.M., Luce J.M., Hearst N., Lo B. Decisions about resuscitation: inequities among patients with different diseases but similar prognoses. Ann Intern Med. 1989;111(6):525–532. doi: 10.7326/0003-4819-111-6-525. [DOI] [PubMed] [Google Scholar]
- 15.Turvey S.L., Gregory A., Bagshaw S., Sligl W.I. What determines do-not-resuscitate status in critically ill HIV patients? Open Forum Infect Dis. 2017;4(Suppl 1):S435. [Google Scholar]
- 16.Dooling K., Marin M., Wallace M., et al. The Advisory Committee on Immunization Practices’ updated interim recommendation for allocation of COVID-19 vaccine—United States, December 2020. MMWR Morb Mortal Wkly Rep. 2021;69(51–52):1657–1660. doi: 10.15585/mmwr.mm695152e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Centers for Disease Control and Prevention . Centers for Disease Control and Prevention website; 2022. COVID-19 Hospital Data—In-hospital mortality among confirmed COVID-19 encounters by week.https://www.cdc.gov/nchs/covid19/nhcs/hospital-mortality-by-week.htm [Google Scholar]
- 18.Fiore M.C., Smith S.S., Adsit R.T., et al. The first 20 months of the COVID-19 pandemic: mortality, intubation and ICU rates among 104,590 patients hospitalized at 21 United States health systems. PLOS One. 2022;17(9) doi: 10.1371/journal.pone.0274571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tabah A., Ramanan M., Laupland K.B., et al. Personal protective equipment and intensive care unit healthcare worker safety in the COVID-19 era (PPE-SAFE): an international survey. J Crit Care. 2020;59:70–75. doi: 10.1016/j.jcrc.2020.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rebmann T., Vassallo A., Holdsworth J.E. Availability of personal protective equipment and infection prevention supplies during the first month of the COVID-19 pandemic: a national study by the APIC COVID-19 task force. Am J Infect Control. 2021;49(4):434–437. doi: 10.1016/j.ajic.2020.08.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Field N., Cohen T., Struelens M.J., et al. Strengthening the reporting of molecular epidemiology for infectious diseases (STROME-ID): an extension of the STROBE statement. Lancet Infect Dis. 2014;14(4):341–352. doi: 10.1016/S1473-3099(13)70324-4. [DOI] [PubMed] [Google Scholar]
- 22.Petrilli C.M., Jones S.A., Yang J., et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. BMJ. 2020;369:m1966. doi: 10.1136/bmj.m1966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Rhee C., Jones T.M., Hamad Y., et al. Prevalence, underlying causes, and preventability of sepsis-associated mortality in US acute care hospitals. JAMA Netw Open. 2019;2(2) doi: 10.1001/jamanetworkopen.2018.7571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Youngwerth J.M., Bartley J.B., Yamashita T.E., Kutner J.S. Characteristics associated with higher cost per day or longer length of stay in hospitalized patients who died during the hospitalization or were discharged to hospice. J Hosp Med. 2011;6(6):338–343. doi: 10.1002/jhm.889. [DOI] [PubMed] [Google Scholar]
- 25.Miller W.D., Han X., Peek M.E., Charan Ashana D., Parker W.F. Accuracy of the sequential organ failure assessment score for in-hospital mortality by race and relevance to crisis standards of care. JAMA Netw Open. 2021;4(6) doi: 10.1001/jamanetworkopen.2021.13891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ho D.E., Imai K., King G., Stuart E.A. MatchIt: nonparametric preprocessing for parametric causal inference. J Statistical Software. 2011;42(8):1–28. [Google Scholar]
- 27.Hansen B.B., Klopfer S.O. Optimal full matching and related designs via network flows. JCGS. 2006;15:609–627. [Google Scholar]
- 28.Zhang Z., Kim H.J., Lonjon G., Zhu Y. Balance diagnostics after propensity score matching. Ann Transl Med. 2019;7(1):16. doi: 10.21037/atm.2018.12.10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Baden L.R., El Sahly H.M., Essink B., et al. Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine. N Engl J Med. 2021;384(5):403–416. doi: 10.1056/NEJMoa2035389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Polack F.P., Thomas S.J., Kitchin N., et al. Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine. N Engl J Med. 2020;383(27):2603–2615. doi: 10.1056/NEJMoa2034577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Nebel J.M. Status quo bias, rationality, and conservatism about value. Ethics. 2015;125(2):449–476. [Google Scholar]
- 32.Blumenthal-Barby J.S. Biases and heuristics in decision making and their impact on autonomy. Am J Bioeth. 2016;16(5):5–15. doi: 10.1080/15265161.2016.1159750. [DOI] [PubMed] [Google Scholar]
- 33.Blumenthal-Barby J.S. The MIT Press; Cambridge, MA: 2021. Good ethics and bad choices: the relevance of behavioral economics for medical ethics. [Google Scholar]
- 34.Li P., Yan Cheng Z., Lin Liu G. Availability bias causes misdiagnoses by physicians: direct evidence from a randomized controlled trial. Intern Med. 2020;59(24):3141–3146. doi: 10.2169/internalmedicine.4664-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Toth-Manikowski S.M., Swirsky E.S., Gandhi R., Piscitello G. COVID-19 vaccination hesitancy among health care workers, communication, and policy-making. Am J Infect Control. 2022;50(1):20–25. doi: 10.1016/j.ajic.2021.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Centers for Disease Control and Prevention . Centers for Disease Control and Prevention website; 2022. Post-COVID conditions.https://www.cdc.gov/coronavirus/2019-ncov/long-term-effects/index.html [Google Scholar]
- 37.Kelly E.P., Henderson B., Hyer M., Pawlik T.M. Intrapersonal factors impact advance care planning among cancer patients. Am J Hosp Palliat Care. 2021;38(8):907–913. doi: 10.1177/1049909120962457. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.