The effect of a tiered provider staffing model on patient outcomes during the coronavirus disease 2019 pandemic: A single-center observational study.

BACKGROUND: During the coronavirus disease 2019 (COVID-19) pandemic, our hospital experienced a large influx of critically ill patients with acute respiratory failure. In order to increase intensive care unit (ICU) surge capacity, we adopted a "tiered model" for ICU provider staffing where multiple ICUs were staffed by noncritical care providers under the direction of an intensivist. We hypothesized that ICUs staffed with a tiered model would result in similar patient outcomes as ICU staffed with a traditional intensivist model.
METHODS: We performed a single-center, observational study in seven ICUs at a tertiary care center. We included consecutive adults admitted to the ICU with acute respiratory distress syndrome (ARDS) due to COVID-19 infection. We collected baseline demographics, treatments, and outcomes of interest in traditionally staffed ICUs versus ICUs staffed with a tiered model. The primary outcome was inpatient mortality. All outcomes were censored at day 28.
RESULTS: We included a total of 138 patients in our study: 66 patients were admitted to traditionally staffed ICUs and 52 were admitted to tiered staffing ICUs. Baseline characteristics were similar between groups. ARDS treatments were similar in traditionally staffed ICUs versus tiered staffing model ICUs, including daily mean tidal volume (6.2 mL/kg vs. 6.2 mL/kg, P = 0.95), median daily fluid balance (159 mL vs. 92 mL, P = 0.54), and use of prone ventilation (58% vs. 65%, P = 0.45). There was no difference in inpatient mortality between groups (50% vs. 42%, P = 0.46). We also found no difference in ventilator-free, ICU-free, vasopressor-free, and dialysis-free days between groups.
CONCLUSIONS: Our results suggest that patient outcomes are similar in ICUs with traditional staffing models when compared to ICUs with a tiered staffing mode during a pandemic. Copyright:

INTRODUCTION

In the setting of the coronavirus disease 2019 (COVID-19) pandemic, many hospitals experienced an influx of critically ill patients with acute respiratory failure. In order to increase surge capacity, new intensive care unit (ICU) space was often created by utilizing inpatient wards and other nontraditional areas as temporary ICUs. In order for intensivists to provide their expertise for a large number of critically ill patients during a pandemic, a “tiered system” for ICU staffing has been proposed.[123]Under this tiered system, a physician with formal training in critical care medicine collaborates with providers in multiple ICUs with nontraditional staffing (e.g. noncritical care providers). Little is known about the outcomes of patients cared for during a pandemic under a tiered system versus a traditional ICU staffing model. At our hospital, we experienced a surge in critically ill patients with acute respiratory distress syndrome (ARDS) during the peak of the COVID-19 pandemic and adopted a tiered staffing model. We hypothesized that the tiered model for ICU staffing would allow for a similar quality of ARDS care in ICUs with tiered staffing as compared to ICUs staffed with a traditional model, which would lead to similar patient outcomes across ICUs with different staffing models.

METHODS

Lahey Hospital and Medical Center is a tertiary care, Level I trauma center located in Massachusetts. The hospital has 52 critical care beds organized as multiple subspecialty ICUs (two 12-bed surgical ICUs, a 12-bed medical ICU, an 8-bed medical ICU, and an 8-bed cardiac ICU). The ICUs are staffed by board-certified intensivists 24 h per day, 7 days per week [Table 1]. From March through May of 2020, Massachusetts was an epicenter of the COVID-19 pandemic, ranking third among states with the most cases in the U. S. at the time.[4]A rapid surge in patients with acute respiratory failure required that multiple non-ICU areas in our hospital be converted into ICUs for the care of patients with COVID-19. At the peak of the surge, ICU bed capacity was more than doubled from baseline capacity. Critically ill patients with COVID-19 were cared for in seven different ICUs. Four of the spaces were traditional ICU spaces, and the other three were hospital floors that were converted into temporary ICUs. We temporarily redeployed noncritical care trained providers and trainees to the ICU to work under a tiered staffing model. Each ICU was staffed during the day by either a board-certified intensivist (with internal medicine, neurocritical care, surgery, or anesthesia training) or by a nonintensive care physician (with training in cardiology or anesthesiology) under the regional direction of a board-certified intensivist using a tiered system of staffing [Table 1]. Each ICU also had a mix of different support staff, using a staffing model that blended critical care advanced practice providers and trainees alongside providers without critical care training. “Regional intensivists” assisted nontraditional ICU staff with complex patient care management decisions, including optimal positive end-expiratory pressure titration, use of prone ventilation, use of inhaled pulmonary vasodilator medications, and readiness for extubation. Nontraditional ICU providers were also given educational materials and hospital guidelines describing best practice treatment for the management of ARDS, such as lung-protective ventilation, prone ventilation, and conservative fluid management.[5678]

Table 1

Intensive care unit staffing model used at baseline compared to the staffing model used during the peak of the COVID-19 pandemic

Unit name	Number of beds	Day staffing	Night staffing*
Baseline staffing
SICU	24	2 intensivists (anesthesia or surgery background), 3-4 surgical residents, 1 CT ICU APP	1-2 surgical residents
MICU	12	1 intensivist (internal medicine background), 2 critical care APPs	1 critical care APP
Mixed MICU/CCU	16	1 intensivist, 1 cardiologist, 4-6 internal medicine residents, 1 PCCM fellow	2 internal medicine residents
Pandemic staffing
Mixed MICU/CCU†	16	1 noncritical care trained cardiologist, 1 PCCM fellow, 4-6 internal medicine residents, 1 PCCM regional intensivist	2 noncritical care support staff
SICU B†	12	1 noncritical care trained anesthesiologist, 1 critical care APP, 1-2 noncritical care support staff, 1 PCCM regional intensivist	2 noncritical care support staff
SICU A‡	12	1 PCCM intensivist, 1 critical care APP, 1-2 noncritical care support staff	2 noncritical care support staff
MICU‡	12	1 neurocritical care intensivist, 1 critical care APP, 1-2 noncritical care support staff	1 neurocritical care APP and 1 noncritical care support staff
Inpatient ward #1 converted into ICU‡	6	1 anesthesia intensivist, 1 noncritical care support staff, 1 critical care APP	1 noncritical care support staff
Inpatient ward #2 converted into ICU‡	9	1 anesthesia intensivist, 1-2 noncritical care support staff, 1 critical care APP	1 noncritical care support staff
PACU converted into ICU‡	29	1 surgical intensivist, 2-4 noncritical care support staff	1-2 noncritical care support staff

*Intensivist night coverage at baseline is provided by 1 in-house PCCM intensivist covering all ICUs and 1 anesthesia intensivist providing additional home call for SICU patients. Intensivist coverage for pandemic staffing was provided by 1 in-house PCCM intensivist and 1 in-house anesthesia or neurocritical care intensivist for all ICUs, †Nontraditional “tiered” staffing model used in this ICU, ‡Traditional staffing model used in this ICU. SICU: Surgical intensive care unit, MICU: Medical intensive care unit, CCU: Coronary care unit, PACU: Postanesthesia care unit, PCCM: Pulmonary and critical care medicine, APP: Advanced practice provider, ICU: Intensive care unit, CT: Computed tomography

We included consecutive adult patients >18 years old with ARDS in the setting of confirmed COVID-19 admitted to an ICU between March and May of 2020.[9]Patients were excluded if they were pregnant or a prisoner. We recorded baseline demographics from the medical record and we recorded on a daily basis ARDS-specific management strategies. Outcomes of interest were compared in traditionally staffed ICUs versus ICUs with nontraditional staffing. All outcomes were censored at day 28. The primary outcome was inpatient mortality. Other secondary outcomes included ventilator-free days, extubation success (defined at >48 h free of mechanical ventilation) vasopressor-free days, and dialysis-free days. In order to account for patients who were transferred from one ICU to another during their hospital course, all patient outcomes were analyzed based on the ICU to which the patient was originally admitted. Patient demographics, clinical variables, and outcomes data were recorded and entered into an online database (MS Excel). The statistical analysis for this study was generated using Statistical Analysis Software (SAS®), Cary, North Carolina, United States, version 9.4 for Windows.

RESULTS

Over the course of the study period, 138 patients were admitted to the ICUs with suspected or confirmed COVID-19. A total of 118 patients were included in the study. Sixty-six patients were admitted to ICUs with traditional staffing models, and 52 were admitted to ICUs with nontraditional staffing. The baseline characteristics of patients admitted to ICUs with traditional staffing models were similar to patients admitted to ICUs with nontraditional ICU staffing with the exception of body mass index, which was higher in the traditional ICUs [Table 2]. The admission SOFA score, initial PaO2-to-FiO2 ratio, and use of high-flow nasal cannula and noninvasive ventilation prior to intubation were similar between groups. More patients in the ICUs with nontraditional staffing received convalescent plasma (14% vs. 3%, P = 0.05). There was no difference in the primary outcome of inpatient mortality in the ICUs with traditional staffing when compared to ICUs with nontraditional staffing (50% vs. 42%, P = 0.46). Likewise, there was no significant difference between the two groups with regard to secondary outcomes, including ventilator-free days, ICU-free days, vasopressor-free days, dialysis-free days, and rates of successful extubation [Table 3]. The ARDS-specific treatments were similar in ICUs with traditional staffing compared to ICUs with nontraditional staffing models, including the daily median tidal volume per kg of ideal body weight (6.2 mL/kg vs. 6.2 mL/kg, P = 0.95), median daily fluid balance (159 mL vs. 92 mL, P = 0.54), and the use of prone ventilation (58% vs. 65%, P = 0.45).

Table 2

Characteristics of patients treated in intensive care unit with traditional staffing models versus nontraditional staffing models

Variable	Total ICU patients (n=118), n (%)	Traditional ICU patients (n=66), n (%)	Nontraditional ICU patients (n=52), n (%)	P
Age (mean years±SDa)	67.3±12.1	67.8±11.2	66.5±13.1	0.55b
Race
Caucasian	86 (72.9)	51 (77.3)	35 (67.3)	0.40c
African American	12 (10.2)	7 (10.6)	5 (9.6)
Hispanic	16 (13.6)	6 (9.1)	10 (19.2)
Asian	3 (2.5)	1 (1.5)	2 (3.9)
Native American	1 (0.8)	1 (1.5)	0
Gender, male	74 (62.7)	40 (60.6)	34 (65.4)	0.70d
BMI (mean±SDa)	31.7±9.3	33.1±10.8	29.9±6.7	0.05b
Charlson Comorbidity Index (mean±SDa)	4.2±2.2	4.3±2.2	4.1±2.2	0.54b
Comorbidities
CKD	26 (22)	13 (19.7)	13 (25)	0.51d
Hypertension	81 (68.6)	46 (69.7)	35 (67.3)	0.84d
Diabetes mellitus	51 (43.2)	29 (43.9)	22 (42.3)	1.00d
CAD	19 (16.1)	12 (18.2)	7 (13.5)	0.62d
CHF	10 (8.5)	6 (9.1)	4 (7.7)	1.00d
Cirrhosis	1 (0.9)	0	1 (1.9)	0.45d
Cancer	14 (11.9)	10 (15.2)	4 (7.7)	0.26d
COPD	16 (13.6)	9 (13.6)	7 (13.5)	1.00d
Asthma	6 (5.1)	5 (7.6)	1 (1.9)	0.23d
Dementia	16 (13.6)	9 (13.6)	7 (13.5)	1.00d
SOFA (mean±SDa)	7.5±2.9	7.4±2.9	7.7±3	0.61b
Admission source
Home	33 (28)	22 (33.3)	11 (21.2)	0.26c
Nursing home	26 (22)	15 (22.7)	11 (21.2)
Outside hospital transfer	59 (50)	29 (43.9)	30 (57.6)
Initial PaO2/FiO2 (mean±SDa)	152.2±65.3	153.2±67.5	150.9±63.0	0.85b
High flow nasal cannula prior to intubation	31 (26.3)	15 (22.7)	16 (30.8)	0.40d
Noninvasive ventilation prior to intubation	2 (1.7)	1 (1.5)	1 (1.9)	1.00d
Time spent on>6 L/min of oxygen prior to intubation (h), median (IQRe)	7 (2-25)	7.5 (2-28)	7 (2-23)	0.90f
Laboratory studies, median (IQRe)
Ferritin (ng/ml)	1.811 (1.019-3.127)	1.798 (987-3.900)	1.811 (1.090-2.903)	0.91f
D-dimer (ng/ml)	2.000 (786-2.000)	1.914 (705-2.000)	2.000 (906-2.000)	0.23f
LDH (U/L)	566 (425-766)	566 (428-794)	560 (421-687)	0.60f
IL-6 (pg/ml)	26 (7-82)	45 (10-99)	19 (7-153)	0.41f
COVID-19 specific treatments
Convalescent plasma	9 (7.6)	2 (3)	7 (13.5)	0.04d
Corticosteroids	79 (67)	40 (60.6)	39 (75)	0.12d
Tocilizumab	25 (21)	12 (18.2)	13 (25)	0.38

aSD, bStudent’s t-test, cChi-square test, dFisher’s exact test, eIQR, fSigned rank Wilcoxon test. IQR: Interquartile range, SD: Standard deviation, ICU: Intensive care unit, BMI: Body mass index, CKD: Chronic kidney disease, CAD: Coronary artery disease, CHF: Congestive heart failure, COPD: Chronic obstructive pulmonary disease, LDH: Lactate dehydrogenase, IL-6: Interleukin 6, SOFA: Sequential organ failure assessment, COVID-19: Coronavirus disease 2019

Table 3

Acute respiratory distress syndrome management strategies and patient outcomes in intensive care unit with traditional staffing models versus nontraditional staffing models

Variable	Total patients (n=118), n (%)	Traditional ICU patients (n=66), n (%)	Nontraditional ICU patients (n=52)	P
Daily tidal volume (mL/kg IBW), median (IQRa)	6.2 (5.9-6.6)	6.2 (5.9-6.7)	6.2 (5.9-6.6)	0.95b
Daily fluid balance (mL), median (IQRa)	119 (−341-666)	159 (−307-681)	92 (−365-652)	0.54b
Cumulative fluid balance (mL) for ICU stay, median (IQRa)	492 (−1274-2538)	635 (−1130-2695)	235 (−1786-2293)	0.31b
Prone ventilation	72 (61)	38 (57.6)	34 (65.4)	0.45d
Inhaled pulmonary vasodilator therapy	15 (12.7)	11 (16.7)	4 (7.7)	0.17d
Inpatient mortality	55 (46.6)	33 (50)	22 (42.3)	0.46e
Ventilator days, mean±SDa	15±9.4	14.3±9.2	15.8±9.7	0.40b
Ventilator-free days, median (IQRd)	0 (0-11)	0 (0-10)	0 (0-15)	0.45d
ICU LOS, mean±SDa	15.7±9.3	14.8±9.3	16.9±9.3	0.24b
ICU-free days, median (IQRc)	0 (0-10)	0 (0-9)	0 (0-11)	0.35d
Vasopressor days, median (IQRc)	8 (3-16)	8 (3-15)	10 (3-17)	0.48d
Vasopressor-free days, median (IQRc)	7.5 (0-20)	4.5 (0-19)	11 (0-22)	0.24d
Dialysis days, median (IQRc)	0 (0-2)	0 (0-2)	0 (0-4.5)	0.86d
Dialysis-free days, median (IQRc)	20 (3-28)	17.5 (3-28)	28 (2.5-28)	0.49d
Successfully extubated	45 (38.1)	24 (36.4)	21 (40.4)	0.71e
Disposition
Discharged home	10 (8.5)	5 (7.6)	5 (9.6)	0.85f
Skilled nursing facility	14 (11.9)	7 (10.6)	7 (13.5)
Died	55 (46.6)	33 (50)	22 (42.3)
Still in hospital	39 (33.0)	21 (31.8)	18 (34.6)

aSD, bStudent’s t-test, cIQR, dSigned rank Wilcoxon test. eFisher’s exact test, fChi-square test. LOS: Length of stay, IQR: Interquartile range, SD: Standard deviation, ICU: Intensive care unit

DISCUSSION

During the COVID-19 pandemic, many hospitals in the United States experienced a large influx of critically ill patients with hypoxic respiratory failure.[10]In order to staff temporary ICUs and care for a large population of patients with acute respiratory failure in the setting of a pandemic, a “tiered system” for ICU staffing has been proposed.[123]This tiered model approach to both provider and nursing staffing was employed by multiple hospital systems during the COVID-19 pandemic.[1112]However, little is known about patient outcomes and the quality of care delivered using this type of provider staffing model in a pandemic. Our results suggest that the quality of ARDS care delivered in ICUs staffed by intensivists was similar to that of ICUs staffed with nontraditional ICUs. Indeed, a lung-protective ventilation strategy was achieved with a median tidal volume of approximately 6 mL/kg per ideal body weight in ICUs staffed with both traditional and nontraditional staffing models. Likewise, the use of prone ventilation and a conservative fluid strategy were similar in ICUs with traditional staffing compared to ICUs with a tiered staffing model. When examining patient outcomes, both mortality and ventilator-free days were similar under both staffing models. Our overall inpatient mortality rate of 47% is similar to the 40%–46% mortality rate reported in a large sample of patients with moderate-to-severe ARDS.[13]Thus, our data suggest that the quality of care delivered and the associated patient outcomes during a pandemic are similar under a tiered staffing model when compared to traditional ICU staffing models. Although the baseline characteristics and ARDS management strategies were similar in both groups, we did note a statistically significant difference in the use of convalescent plasma. This difference in treatment may be related to temporal changes in the care of patients with COVID-19 and the availability of different therapies over the course of the pandemic.

Our study is subject to a number of limitations. The observational study design makes the results subject to bias and confounding. However, we did try to account for these limitations by evaluating baseline characteristics and markers of severity of illness between ICUs of different staffing models. Furthermore, it would be difficult to randomize patients to specific ICUs given that bed availability is limited in the setting of a pandemic and it most likely would not be feasible to randomize patients to one ICU or another under such bed constraints. Other factors unrelated to provider staffing may have also influenced patient outcomes, including other staffing models used during the pandemic (nursing, respiratory therapy, and pharmacy) and challenges with equipment and other resource availability in nontraditional ICU spaces.

CONCLUSIONS

We found that the quality of care delivered and the observed patient outcomes were similar in ICUs with traditional staffing models when compared to ICUs with a tiered staffing model during a pandemic. Further study is needed to confirm these findings.

Research and quality and ethics statement

This study was approved by the Institutional Review Board at Lahey Hospital and Medical Center (Approval #1581237; Approval Date: March 19, 2020). The authors followed the applicable EQUATOR Network (http://www.equator-network.org) guidelines, specifically the STROBE Guidelines, during the conduct of this research project.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.