A Retrospective Review of the Characteristics and Outcomes of Patients through an Integrated Palliative Care Model during the First Wave of the SARS-COV-2 Pandemic

Abstract

Background:

The COVID-19 pandemic created surges of rapidly deteriorating patients straining health care necessitating the evaluation of novel models of palliative care (PC) integration to reduce patient suffering and hospital strain.

Objective:

To evaluate an integrated PC model's effect on code status change.

Design:

This is an observational retrospective study.

Setting:

Urban quaternary referral center in the southeastern United States from April 6th to August 20th, 2020.

Patients:

All patients admitted to our medical intensive care unit and stepdown unit were diagnosed with COVID-19.

Measurements:

Code status change, multivariate regression on patient characteristics.

Results:

In total, 79.7% (98/123) patients were full code at admission. After PC consultation, 33.3% (41/123) patients remained full code, 13.0% (16/123) were do not resuscitate (DNR), and 53.6% (66/123) changed to DNR/do not intubate (DNI). An ordinal logistic model determined that consultation location (odds ratio [OR] 3.35, p = 0.017) and patient age (OR 1.09, p < 0.001) were predictive of code status change to DNR/DNI.

Conclusion:

Within an integrated PC model, PC consultation was associated with code status change. The effect of an integrated PC model warrants further study in comparison with a traditional PC model in a similar patient cohort.

Background

The SARS-COV-2 that caused the COVID-19 pandemic has created surges in demand for health care services, including palliative and end-of-life care.^1–3 Because of its morbidity, mortality, and potential for rapid clinical decompensation during the first wave of the pandemic, COVID-19–related hospitalizations necessitated early goals-of-care (GOC) discussions with infected patients.^4–9 Prior studies advocated for appropriate palliative care (PC) utilization, particularly in high-risk patients.⁵ The majority of literature on the impact of PC focused on the need for PC to partner in the care of COVID-19 patients, however, few articles provided guidance on how PC can optimally partner with COVID-19 teams.^5,10

Our institution developed an integrated model that encouraged primary teams to proactively consult PC for patients at risk of COVID-19 complications. In this integrated model, all COVID-19 admissions during the first wave of the pandemic triggered a joint decision-making process with the primary team determining collaboratively with a PC specialist whether a formal PC consultation was indicated. Our goal was to retrospectively review the demographics, patient characteristics, and outcomes of COVID-19 patients admitted to our institution in our step-down unit and medical intensive care unit (MICU) who received a PC consultation through this integrated PC consultation model.

Methods

This retrospective study included all consecutive adult admissions with confirmed COVID-19, defined as a positive reverse transcription polymerase chain reaction (RT-PCR) SARS-COV-2 nucleic acid test, who received a PC consultation and were hospitalized between April 6th and August 26th, 2020, at an urban, quaternary care academic medical center located in the southeastern United States. Our institutional review board approved the study and waived the need for informed consent. No funding was procured for the creation of this study. Deidentified demographic data were collected from the electronic medical record (EMR) and stored in a secure password-protected REDCAP database.^11,12

All PC consultations generated during the inclusion period resulted from a collaborative phone or web-based conference between the PC team and either the primary MICU-based critical care team or stepdown unit-based hospital medicine team. PC consultation was encouraged for all patients at higher risk of COVID-19 complications including but not limited to age ≥70 years, frailty, presence of organ transplant, chronic lung disease, active malignancy, end-stage heart failure, liver disease, renal failure, dementia, or residence at a long-term care facility.

The primary outcome was to determine which patient characteristics if any were associated with a change in code status after PC consultation. In addition, patient characteristics were collected including demographics, comorbidities, time to PC consultation, hospital length of stay (LOS), discharge disposition, and functional status at time of admission, measured by Eastern Cooperative Oncology Group (ECOG) score.

Code status was defined as “full code” [pursue all life-sustaining treatments (LSTs) including intubation and cardiopulmonary resuscitation (CPR)], “do not resuscitate (DNR) only” [pursue all LSTs excluding CPR], “do not resuscitate/do not intubate (DNR/DNI)” [forgo CPR and intubation], or “unknown.” Patients with an “unknown” code status were treated as “full code” unless a new code status order was entered into the EMR.

All statistical analyses were performed on Stata version 16.2. Descriptive analyses were conducted. The Kruskal–Wallis/Wilcoxon rank sum test and Pearson chi-squared/Fisher exact test were performed to compare continuous or categorical variables by groups as appropriate. A multivariate ordinal logistic regression model was also fitted to determine which variables were associated with code status change. The covariates in our regression modeling were age, location of consultation, ECOG, presence of chronic lung disease, presence of “other” diseases, presence of chronic neurological disease, and number of comorbid conditions. “Other” disease included patients with immunocompromising conditions, organ transplants, and immunosuppressive medications. An example of our data extraction survey can be found in the Supplementary Data.

Results

A total of 123 patients were identified during the study period. As described in Table 1, the median age was 71.2 (range 33–97). Sixty-four percent (79/123) of patients were white, 22% (27/123) were black, and 8% (10/123) were Hispanic. Eighty-eight percent (108/123) spoke English as their primary language and 10.6% (13/123) spoke Spanish. Table 2 shows seventy-nine percent (98/123) of admitted patients were identified as full code on admission. In total, 2.4% (4/123) patients were DNR and 17% (22/123) were DNR/DNI. After PC consultation, 57 patients changed their code status from full code as shown in Table 3. Thirteen patients changed their code status from full code to DNR, and 44 patients changed their code status from full code to DNR/DNI.

Table 1.

Demographics of 123 Patients

Demographics	No. (%)
Age (years), median (range)	71.2 (33–97)
<50	10 (8.13)
51–60	18 (14.63)
61–70	31 (25.20)
71–80	27 (21.95)
> 80	37 (30.08)
Gender
Male	69 (56)
Female	54 (44)
Race/ethnicity
White	79 (64.23)
Black	27 (21.95)
Latino	10 (8.13)
Unknown	3 (2.44)
Asian	1 (0.81)
Native American	1 (0.81)
Pacific Islander	1 (0.81)
Other	1 (0.81)
Marital status
Married	56 (45.33)
Single	34 (27.64)
Widowed	26 (21.14)
Divorced	7 (5.69)
Location before admission
Home	78 (70.27)
Long-term care facility	16 (14.41)
Skilled nursing facility	12 (10.81)
Inpatient rehabilitation	5 (4.5)
Insurance
Medicare	48 (39.02)
Commercial	28 (22.76)
Medicare intermediary	27 (21.95)
None	10 (8.13)
Medicaid	5 (4.07)
Veterans affairs	5 (4.07)
Language
English	108 (87.8)
Spanish	13 (10.57)
Arabic	1 (0.81)
Nepali	1 (0.81)

Table 2.

Clinical Characteristics of 123 Patients

Characteristics	No. (%)
Comorbidities, median (range)	4 (2–5)
Other^a	63 (51.22)
Hypertension	63 (51.22)
Obesity	50 (40.65)
Frailty	49 (39.84)
Diabetes mellitus	44 (35.77)
Chronic kidney disease/end-stage renal disease	27 (21.95)
Dementia	25 (20.33)
Chronic lung disease	23 (18.70)
Chronic heart failure	21 (17.07)
Cancer	21 (17.07)
Vascular disease	17 (13.82)
Delirium	15 (12.20)
Psychiatric disease	14 (11.38)
Chronic neurological disease	14 (11.38)
Cerebrovascular accident	13 (10.57)
End-stage liver disease	3 (2.44)
Peripheral artery disease	2 (1.63)
ECOG
1	0 (0)
2	8 (6.50)
3	48 (39.34)
4	67 (54.92)
O₂ support at time of consultation
Yes	86 (69.92)
No	37 (30.08)
Admission code status
Full	83 (78.48)
DNR/DNI	22 (17.89)
Unknown	15 (12.20)
DNR	4 (2.44)
Disposition at hospital discharge
Deceased	38 (30.9)
Skilled nursing facility	19 (15.5)
Home health	16 (13.0)
Rehabilitation	15 (12.2)
Home with no services	13 (10.6)
Hospice scatter bed	8 (6.5)
LTC facility	6 (4.88)
Hospice at LTC	4 (3.25)
Home hospice	2 (1.63)
Other	2 (1.63)

Other comorbidities included history of spinal cord injury, quadriplegia, tetraplegia, paraplegia, history of organ transplant (kidney, lung, heart, and liver), chronic tracheostomy/percutaneous endoscopic gastrostomy tube dependence, history of rheumatological conditions on immunosuppressives, HIV/AIDS, intellectual disability, and hypothyroidism.

DNI, do not intubate; DNR, do not resuscitate; ECOG, Eastern Cooperative Oncology Group; LTC, long-term care.

Table 3.

Outcomes

Code status	Patient no. (%)
Code status	At time of admission	After PC consultation	Code status at time of discharge/death
Full code	83 (67.47)	40 (32.52)	39 (31.71)
DNR	3 (2.44)	16 (13.01)	13 (10.57)
DNR/DNI	22 (17.89)	66 (53.66)	69 (56.10)
Unknown	15 (12.20)	1 (0.81)	2 (1.63)

PC, palliative care.

Table 4 shows in total, 30.9% (38/123) of patients died while inpatient, with 71% (27/38) of the deceased patients presenting initially as full code. Of these 27 patients who were initially full code, after PC consultation, only 3 patients remained full code at the time of death. Of the 38 patients who died while hospitalized, 39% (15/38) died in the stepdown unit, and 61% (23/38) died in the MICU. Of the 15 patients located on the stepdown unit, 14 changed their code status from full code to DNR/DNI after PC consultation and 1 changed their code to DNR. Of the 23 patients who died in the MICU, 12 arrived intubated on a ventilator, and after PC consultation, only 3 remained full code.

Table 4.

Outcomes of Deceased Patients

MICU code status	Patient no. (%)
MICU code status	At time of admission	After PC consultation	Code status at time of discharge/death
Full code	17 (73.9)	8 (34.8)	3 (13.0)
DNR	0 (0)	2 (8.7)	5 (21.7)
DNR/DNI	3 (13.0)	13 (56.5)	15 (65.2)
Unknown	3 (13.0)	0 (0)	0 (0)

Stepdown code status	Patient no. (%)
Stepdown code status	At time of admission	After PC consultation	Code status at time of discharge/death
Full code	10 (66.7)	2 (13.3)	0 (0)
DNR	1 (6.67)	0 (0)	1 (6.67)
DNR/DNI	3 (20)	13 (86.7)	14 (93.2)
Unknown	1 (6.67)	0 (0)	0 (0)

MICU, medical intensive care unit.

Results of multivariable ordinal logistic regressions are shown in Table 5. Patients who were admitted to the MICU at the time of consultation had an increased odds of a code status change (OR 3.35, p < 0.05). Age (OR 1.09 for every year >33 years, p < 0.05) and patients with chronic lung disease (OR 3.74, p < 0.05) were also associated with an increased odds of changing code status. Age (OR 1.09 for every year >33 years, p < 0.05) was also associated with an increased odds of changing code status with chronic lung disease trending towards statistical significance (OR 3.74, p = 0.053). Chronic neurological disease (OR 0.214, p < 0.05) was the only characteristic associated with a decreased odds of code status change. ECOG score at admission directly correlated with median time to PC consultation with higher ECOG scores being associated with a later PC consultation (ECOG 2: 2 days, ECOG 3: 2.5 days, ECOG 4: 4 days, p < 0.05).

Table 5.

Results of Multivariable Ordinal Logistic Regression Model

Variables	Odds ratio	p
MICU patient location^a	3.35	0.017
Age^b	1.09	<0.001
Chronic lung disease^c	3.74	0.053
“Other diseases”^d	2.71	0.031
ECOG^e
3	0.422	0.368
4	0.452	0.414
No. of comorbid conditions^f	0.934	0.601
Chronic neurological disease^g	0.214	0.029

MICU patient location: stepdown floor is the reference group (binary).

Age: 33 years is the reference group (continuous).

Chronic lung disease: no chronic lung disease is the reference group (binary).

“Other disease”: no “other disease” is the reference group (binary).

ECOG: 2 is the reference group (categorical with three values).

No. of comorbid conditions: 0 is the reference group (continuous variable).

Chronic neurological disease: no chronic neurological disease is the reference group (binary).

We noted a statistically significant association between patient characteristics and time to PC consultation in Figure 1. Patients who were full code with a longer time to PC consultation and an inverse association between age and time to PC consultation. Increasing age was associated with earlier PC consultation time. Patients in the MICU had a longer time to PC consultation (t = −3.87, p < 0.05) than stepdown unit patients. Differences in gender, marital status, religion, race, residential facility (RF), or insurance were not associated with code status outcomes.

FIG. 1.

Secondary analysis. Box plots with p values are displayed showing the association between (A) Initial code status and PC consult time, (B) Length of stay and code status after palliative care consult, (C) ECOG and PC consult time, (D) Age and palliative care consult time, and (E) Referring service and palliative care consult time.

Discussion

In our retrospective review, PC consultation was associated with a 51.8% (43/83) transition from full code to DNR or DNR/DNI status. PC consultation reduced resource utilization as evidenced in patient mortality with only 3 of the 38 deceased patients remaining full code. Most significantly, 14 of the 15 patients on the stepdown floor changed their code status to DNR/DNI and 1 of the 15 patients changed code status to DNR before the time of death resulting in improved goal-concordant care and resource utilization. Patients who received PC consultations while in the MICU had the highest odds to change their code status to DNR/DNI.

Older patients tended to get PC consultations earlier in their admission than their younger counterparts, with 29.7% of patients (33/111) presenting from nursing RFs, and 90.7% (30/33) of RF inhabitants >61 years of age. Median time to consultation was shortest for those patients in the 71–80 years of age group. This is not surprising, as age was a trigger for referral to PC given the evidence age plays in poorer outcomes, and older patients are found disproportionately in RFs.^13,14

Multivariable regression modeling found an association between code status change from full code to DNR or DNR/DNI with patients who had chronic lung disease, increased age, MICU consultation location, and “other” diseases. When reviewed, 31.7% (20/63) of patients with “other” diseases were on immunosuppressive medications for solid organ transplants or for rheumatological conditions, with the next largest percentage 15.9% (10/63) associated with patients experiencing spinal cord or brain injuries, and chronic debility with 80% (8/10) presenting from nursing facilities.

Interestingly, a higher ECOG score and initial admission to the MICU delayed PC consultation. Chronic neurological illness was protective against code status change to DNR/DNI. We suspected patients with chronic illness tolerate a worse quality of life, and patients with impaired functional status required family members to be coordinated to discuss GOC.

From our literature review, this is the first study to use multivariate logistic regression to examine variables associated with a code status change. It is important to note that this study was a retrospective review of patients who required hospital admission for COVID-19 infections early in the pandemic from a single institution in the southeastern United States. It is, therefore, difficult to generalize our findings to other institutions across the country and across all severity of COVID-19 infections. In addition, advances in therapies, including vaccination, continue to occur and may impact those patients affected by this pandemic. Our study also lacked a control group to compare the effect of an integrated versus traditional consultation model.

Conclusion

The impact of an integrated approach warrants further study in comparison with a traditional PC model in similar patients to determine whether an integrated model is more effective at changing code status and decreasing LOS. As this pandemic continues for an unknown duration, our longitudinal findings may provide insight into how PC can be appropriately utilized to effectively meet the demands of future patients with COVID-19.

Footnotes

Funding Information

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

References

Chua

, Shi

, Levine

. Place of death and end-of-life care utilization among COVID-19 decedents in a Massachusetts Health Care System. J Palliat Med, 2021; 24(3):322–323; doi: 10.1089/jpm.2020.0674

Bloomer

, Bouchoucha

. Editorial: COVID-19 and what it means for end-of-life care in ICU: Balancing the priorities. Collegian, 2020; 27(3):248–249; doi: 10.1016/j.colegn.2020.05.007

Ting

, Edmonds

, Higginson

, et al. Palliative care for patients with severe covid-19. BMJ, 2020; 370(8255):m2710; doi: 10.1136/bmj.m2710

Moore

, Sampson

, Kupeli

, et al. Supporting families in end-of-life care and bereavement in the COVID-19 era. Int Psychogeriatr, 2020; 32(10):1245–1248; doi: 10.1017/S1041610220000745.

Pattison

. End-of-life decisions and care in the midst of a global coronavirus (COVID-19) pandemic. Intensive Crit Care Nurs, 2020; 58:102862; doi: 10.1016/j.iccn.2020.102862

Etkind

, Bone

, Lovell

, et al. The role and response of palliative care and hospice services in epidemics and pandemics: A rapid review to inform practice during the COVID-19 pandemic. J Pain Symptom Manage, 2020; 60(1):e31–e40; doi: 10.1016/j.jpainsymman.2020.03.029

Abbott

, Johnson

, Wynia

. Ensuring adequate palliative and hospice care during COVID-19 Surges. JAMA, 2020; 324(14):1393–1394; doi: 10.1001/jama.2020.16843

Lopez

, Decastro

, Van Ogtrop

, et al. “Palliative Pandemic Plan,” Triage and Symptoms Algorithm as a strategy to decrease providers' exposure, while trying to increase teams availability and guidance for goals of care (GOC) and symptoms control. Am J Hosp Palliat Care, 2020; 37(11):980–984; doi: 10.1177/1049909120942494

Pahuja

, Wojcikewych

. Systems barriers to assessment and treatment of COVID-19 positive patients at the end of life. J Palliat Med, 2021; 24(2):302–304; doi: 10.1089/jpm.2020.0190

10.

Rao

, Kelemen

. Lessons learned from caring for patients with COVID-19 at the end of life. J Palliat Med, 2020; 24(3):468–471; doi: 10.1089/jpm.2020.0251

11.

Harris

, Taylor

, Thielke

, et al. Conde, Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42(2):377–381; doi: 10.1016/j.jbi.2008.08.010

12.

Harris

, Taylor

, Minor

, et al.; REDCap Consortium. The REDCap consortium: Building an international community of software partners. J Biomed Inform, 2019; 95:103208; doi: 10.1016/j.jbi.2019.103208

13.

Williamson

, Walker

, Bhaskaran

, et al. Factors associated with COVID-19-related death using OpenSAFELY. Nature, 2020; 584(7821):430–436; doi: 10.1038/s41586-020-2521-4

14.

Ibrahim

. An equation to predict deaths of nursing home residents during a pandemic. Nat Aging, 2021; 1(7):571–573; doi: 10.1038/s43587-021-00083-x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB