Perceived Hospital Stress, Severe Acute Respiratory Syndrome Coronavirus 2 Activity, and Care Process Temporal Variance During the COVID-19 Pandemic*

Abstract

OBJECTIVES: 

The COVID-19 pandemic threatened standard hospital operations. We sought to understand how this stress was perceived and manifested within individual hospitals and in relation to local viral activity.

DESIGN: 

Prospective weekly hospital stress survey, November 2020–June 2022.

SETTING: 

Society of Critical Care Medicine’s Discovery Severe Acute Respiratory Infection-Preparedness multicenter cohort study.

SUBJECTS: 

Thirteen hospitals across seven U.S. health systems.

INTERVENTIONS: 

None.

MEASUREMENTS AND MAIN RESULTS: 

We analyzed 839 hospital-weeks of data over 85 pandemic weeks and five viral surges. Perceived overall hospital, ICU, and emergency department (ED) stress due to severe acute respiratory infection patients during the pandemic were reported by a mean of 43% (sd, 36%), 32% (30%), and 14% (22%) of hospitals per week, respectively, and perceived care deviations in a mean of 36% (33%). Overall hospital stress was highly correlated with ICU stress (ρ = 0.82; p < 0.0001) but only moderately correlated with ED stress (ρ = 0.52; p < 0.0001). A county increase in 10 severe acute respiratory syndrome coronavirus 2 cases per 100,000 residents was associated with an increase in the odds of overall hospital, ICU, and ED stress by 9% (95% CI, 5–12%), 7% (3–10%), and 4% (2–6%), respectively. During the Delta variant surge, overall hospital stress persisted for a median of 11.5 weeks (interquartile range, 9–14 wk) after local case peak. ICU stress had a similar pattern of resolution (median 11 wk [6–14 wk] after local case peak; p = 0.59) while the resolution of ED stress (median 6 wk [5–6 wk] after local case peak; p = 0.003) was earlier. There was a similar but attenuated pattern during the Omicron BA.1 subvariant surge.

CONCLUSIONS: 

During the COVID-19 pandemic, perceived care deviations were common and potentially avoidable patient harm was rare. Perceived hospital stress persisted for weeks after surges peaked.

KEY POINTS

Question: How was stress perceived within individual hospitals and in relation to local viral activity during the COVID-19 pandemic?

Findings: In this prospective weekly hospital stress survey in seven U.S. health systems, during the Delta variant surge, perceived overall hospital stress persisted for a median of 11.5 weeks after local case peak. ICU stress had a similar pattern of resolution (median 11 wk after local case peak) while the resolution of emergency department stress (median 6 wk after local case peak) was significantly earlier.

Meaning: During the COVID-19 pandemic, hospital stress measures persisted for weeks after surges peaked.

The COVID-19 pandemic has been a generational challenge for hospitals globally. Hospitals have reported stress across numerous domains including: beds and care locations (^1–3), staffing (^4–6), equipment (^7,8), clinical treatment protocols (^9,10), and patient and family communication (¹¹). These organizational stressors might threaten standard care operations (^2,3,12,13), but it is not well known how such stress manifested at individual hospitals during the pandemic. How hospital stress might have changed over time during the pandemic, where and how stress is manifested within hospital care and organizational structures, and how individual measures of stress might be related to each other and to local severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) activity levels and trends are not well described.

Hospital stress, also called hospital capacity strain, is the operations concept of approaching or exceeding limits placed on a hospital’s ability to provide high-quality care for all patients who may need it at a given time (^14,15). Pre–COVID-19 pandemic efforts at rolling qualitative hospital stress assessments have been shown to be feasible but challenging during both public health emergencies and during more routine or predictable surges in care demands such as seasonal influenza epidemics (^16,17). Past lessons include the detection of significant stress and associated care deviations that potentially threatened patient safety during a winter respiratory viral disease surge not part of any declared public health emergency (¹⁷). The specific types of care deviations most likely to occur during such periods include changes in staffing, inter-hospital transfers, and elective procedures. Continued investigation is important, through both unanticipated and anticipated surges, to understand and optimize hospital adaptation and resiliency (¹⁸).

As part of the Society of Critical Care Medicine’s Discovery (SCCM Discovery) Severe Acute Respiratory Infection-Preparedness (SARI-PREP) prospective, multicenter, cohort study (¹⁹), we used a longitudinal prospectively collected hospital stress survey deployed during the COVID-19 pandemic to understand patterns of perceived hospital stress over time, where stress was located within hospitals, within-hospital correlations between individual stress measures, and temporal relationships between stress measures and local SARS-CoV-2 activity. Such results may inform longitudinally sustainable preparedness approaches to dynamically optimize hospital adaptation and resiliency to ongoing and future threats.

METHODS

Study and Study Sites

This research occurred as part of the SCCM’s Discovery SARI-PREP Study (¹⁹), a prospective, multi-center, cohort study of SARI patients (Appendix A, https://links.lww.com/CCM/H293) and treating hospitals beginning during the first year of the COVID-19 pandemic (ClinicalTrials.gov Identifier: NCT04786301) (^18–21). SARI-PREP study sites participating in the hospital stress survey included 13 hospitals across six U.S. states from seven health systems (eTable 1, https://links.lww.com/CCM/H293). Institutional Review Boards (IRBs) at all sites approved the study via the central coordinating site (University of Nebraska Medical Center single IRB No. 544-20-FB, “Severe Acute Respiratory Infection-Preparedness (SARI-Prep),” approved September 9, 2020). The procedures followed were in accordance with the ethical standards of the IRBs and with the Helsinki Declaration of 1975.

Hospital Stress Survey

We conducted a prospective hospital stress survey weekly beginning in November 2020 with rolling study site entry (eTable 1, https://links.lww.com/CCM/H293) and ending June 2022, completed by designated respondents among SARI-PREP site investigators. The survey was based on a previously developed, validated, and deployed health system stress survey created by the SCCM Discovery Program for Resilience and Emergency Preparedness for use during seasonal influenza epidemics (¹⁷). The present survey was adapted for the COVID-19 pandemic and measured perceived overall hospital, ICU, and emergency department (ED) stress and related perceived deviations from standard operating procedures, due to SARI patients during the COVID-19 pandemic. Importantly, in contrast to recent studies measuring hospital stress quantitatively using capacity strain metrics such as occupancy and patient acuity (^12,22), this survey study qualitatively measured the stress perceived at individual hospitals over time. See Appendix B (https://links.lww.com/CCM/H293) for stress survey administration details and Appendix C (https://links.lww.com/CCM/H293) for the complete hospital stress survey instrument. Primary respondents, predominantly critical care clinical investigators, were encouraged and allowed to use any local resources, such as contact with additional stakeholders and capacity dashboards, for their stress assessments, but the assessments are meant to be “real-use” perceptions and are not anchored on specific quantitative metrics. While the initial SARI case definition included both influenza and COVID-19, due to very low numbers of SARI patients without SARS-CoV-2 infection during the pandemic (^23,24), the hospital stress survey results can be interpreted as almost exclusively driven by patients with COVID-19.

The analytic units of measure were hospital-week (for within- and among-hospital analyses) and pandemic week (for across-pandemic analyses). We categorized periods of national SARS-CoV-2 surges defined as including days with a 7-day rolling average of greater than 60,000 incident SARS-CoV-2 cases per day nationally (eTable 2, https://links.lww.com/CCM/H293) (^25,26). Weeks without any such days were considered between-surge periods.

SARS-CoV-2 Case Data

Weekly SARS-CoV-2 infection case counts for corresponding study hospital counties and states were extracted from public health department repositories, and if not available, alternative public sources (eTable 3, https://links.lww.com/CCM/H293) (^27,28). SARS-CoV-2 infection case counts were standardized per 100,000 residents based on 2021 U.S. Census Bureau statistics (^29,30).

Stress Measures, Correlations, and Associations

Pairwise comparisons of binary or dichotomized hospital stress measures were calculated by Spearman correlation coefficients (ρ). Associations between county case counts and hospital stress measures were evaluated with multivariable logistic regression models—separately for overall hospital, ICU, and ED stress—on the level of the hospital-week adjusted for hospital and pandemic surge period, to account for both SARS-CoV-2 variant characteristics and longitudinal time. Appendix D (https://links.lww.com/CCM/H293) details the approach to missing data and related sensitivity analyses. Because in the present study perceived patient harm was used as a reflection of stress, we did not investigate the association between either SARS-CoV-2 case counts or hospital stress and perceived patient harm.

Among-Hospital Variation in Stress

Using multivariable logistic regression, we predicted the proportion of weeks during the Omicron BA.1 subvariant surge, prior to which all but one hospital had entered the cohort, that each hospital reported overall hospital, ICU, and ED stress, adjusting for county SARS-CoV-2 case counts standardized per 100,000 residents.

Temporal Relationship Between SARS-CoV-2 Cases and Hospital Stress

To assess the temporal relationship between local SARS-CoV-2 case counts and hospital stress, we examined the periods surrounding the Delta variant and Omicron BA.1 subvariant surges, prior to which a majority of hospitals had entered the cohort. By surge period, we calculated the median (and interquartile range [IQR]): duration between hospital, ICU, and ED stress start and county SARS-CoV-2 case peak; duration between county SARS-CoV-2 case peak to hospital, ICU, and ED stress end (i.e., sustained abatement > 1 wk in duration); total duration of hospital, ICU, and ED stress; and county SARS-CoV-2 cases at the start and end times of hospital, ICU, and ED stress. By surge period and for each stress measure, we compared the week of stress start and end relative to local county case peak, the stress duration, and the county SARS-CoV-2 case levels at the time of stress start and end, using two-sample Wilcoxon rank-sum (Mann-Whitney U) tests.

RESULTS

Hospital Stress, Care Deviations, and Perceived Potential Patient Harm

Thirteen hospitals across seven health systems in six U.S. states contributed 839 hospital-weeks of data over 85 pandemic weeks and five viral surges from November 2020 to June 2022. After cohort entry, overall response rate was 91%.

Tables 1 and 2 (by pandemic week), eTables 4–8 (https://links.lww.com/CCM/H293) (by hospital-week), and Appendix E (https://links.lww.com/CCM/H293) report perceived hospital stress, care deviations, and resource scarcity across pandemic surge periods. Across all pandemic weeks, perceived overall hospital, ICU, and ED stress due to SARI patients during the COVID-19 pandemic was reported by a mean of 43% (sd, 36%), 32% (30%), and 14% (22%) of hospitals per week, respectively. Care deviations of some kind (“operating differently”) were perceived in a mean of 36% (sd, 33%) of hospitals per week with the most common perceived care deviations being: increasing hospital staffing (mean of 19% [sd, 22%] of hospitals per week), denying inter-hospital transfers (15% [19%]), and canceling elective surgeries (14% [22%]). The presence of perceived hospital stress, care deviations, and resource scarcity were variably reported differently between pandemic surges and compared with between-surge periods (Appendix E, https://links.lww.com/CCM/H293).

TABLE 1. - Perceived Hospital Stress and Care Deviations Across Pandemic Weeks

Hospital Stress Metric or Care Deviation	Total Study Period	Between-Surge Periods	Ancestral Wuhan Straina	Alpha Variant	Delta Variant	Omicron BA.1 Subvariant	Omicron BA.2.12.1 Subvariantb
	November 17, 2020–June 30, 2022		November 17, 2020–March 5, 2021	March 26, 2021–April 23, 2021	July 27, 2021–November 6, 2021	November 7, 2021–March 1, 2022	May 2, 2022–June 30, 2022
	Presence of Stress or Care Deviations, Mean (sd) % of Hospitals per Pandemic Weekc
Weeks, n (%)
Pandemic study (calendar) weeks	85 (100.0)	22 (25.9)	16 (18.8)	5 (5.9)	15 (17.7)	17 (20.0)	10 (11.7)
Hospital-weeks, n (%)	839 (100.0)	214 (25.5)	86 (10.3)	40 (4.8)	156 (18.6)	213 (25.4)	130 (15.5)
Hospital stress
Overall hospital stress	42.7 (36.1)	13.9 (16.5)	88.2 (14.4)	40.7 (2.9)	56.4 (22.8)	50.5 (35.3)	0.0 (0.0)
ICU stress	32.4 (30.1)	7.9 (11.4)	66.5 (19.9)	27.1 (2.0)	43.1 (19.0)	43.1 (31.8)	0.0 (0.0)
Emergency department stress	13.7 (22.3)	0.6 (3.0)	16.6 (13.8)	2.9 (6.4)	16.4 (13.6)	36.9 (35.7)	0.0 (0.0)
Care deviations
Operating differently overall	35.7 (33.2)	11.3 (13.6)	78.2 (21.4)	34.3 (11.5)	41.6 (19.5)	43.4 (35.2)	0.0 (0.0)
Increasing hospital staffing	19.1 (22.4)	9.1 (12.0)	58.9 (13.4)	24.3 (5.9)	9.1 (9.9)	13.5 (9.7)	0.0 (0.0)
Denying inter-hospital transfers	15.4 (19.3)	2.4 (5.3)	14.7 (18.0)	5.4 (7.4)	32.3 (13.1)	29.9 (23.6)	0.0 (0.0)
Unable to transfer a patient elsewhere	7.7 (8.7)	4.7 (6.2)	3.7 (8.5)	0.0 (0.0)	7.9 (7.7)	16.5 (9.4)	9.4 (4.1)
Unable to accept a requested transfer	26.7 (24.4)	9.2 (9.2)	16.9 (27.2)	16.1 (11.3)	45.3 (8.7)	54.0 (19.9)	11.4 (6.1)
Canceling elective surgeries	14.0 (22.3)	0.0 (0.0)	44.0 (27.7)	0.0 (0.0)	17.0 (16.8)	13.6 (17.0)	0.0 (0.0)
Increased emergency department boarding	8.7 (16.6)	0.6 (3.0)	2.8 (6.1)	0.0 (0.0)	14.2 (13.6)	27.7 (25.1)	0.0 (0.0)
Using surge ICU spaces	4.2 (7.7)	1.9 (4.8)	9.7 (10.6)	5.7 (7.8)	0.0 (0.0)	7.6 (8.8)	0.0 (0.0)
Changing staff-to-patient ratios	4.0 (8.2)	0.6 (3.0)	8.2 (11.7)	0.0 (0.0)	1.9 (5.5)	10.0 (9.7)	0.0 (0.0)
Limiting certain interventions	1.6 (3.8)	0.3 (1.6)	1.8 (4.9)	0.0 (0.0)	0.0 (0.0)	5.8 (4.7)	0.0 (0.0)

^aAncestral Wuhan strain wave includes the third U.S. national wave in Winter 2020–2021 but not two earlier ancestral surges prior to the study period.

^bThe Omicron BA.2.12.1 subvariant wave includes the early rise of Omicron BA.5 and BA.4 subvariants.

^cOrdinal stress measures were dichotomized; see eTables 4–8 (https://links.lww.com/CCM/H293) for ordinal results by hospital-week.

Analyses performed by study week. Primary results and percentages are reported as complete case analyses. See sensitivity analysis results for alternative handling of rare missing data. Between-surge periods are weeks with no days with a 7-d rolling average of greater than 60,000 incident severe acute respiratory syndrome coronavirus 2 cases nationally.

TABLE 2. - Perceived Care Resource Availability Across Pandemic Weeks

Resource	Total Study Period	Between-Surge Periods	Ancestral Wuhan Straina	Alpha Variant	Delta Variant	Omicron BA.1 Subvariant	Omicron BA.2.12.1 Subvariantb
	November 17, 2020–June 30, 2022		November 17, 2020–March 5, 2021	March 26, 2021–April 23, 2021	July 27, 2021–November 6, 2021	November 7, 2021–March 1, 2022	May 2, 2022–June 30, 2022
	Presence of Resource Scarcity, Mean (sd) % of Hospitals per Pandemic Weekc
Staffing availability issues
Attending physicians	19.4 (19.8)	11.8 (12.0)	51.6 (16.4)	24.3 (5.9)	10.5 (11.0)	16.3 (8.6)	0.9 (2.9)
Residents or fellows	20.4 (25.5)	10.3 (12.3)	61.6 (28.5)	24.8 (9.8)	10.6 (8.4)	14.1 (11.2)	0.0 (0.0)
Bedside registered nurses	19.8 (20.1)	7.1 (7.7)	46.6 (22.7)	16.4 (6.8)	19.6 (14.3)	18.4 (17.7)	9.2 (8.0)
Physician assistants or nurse practitioners	18.7 (19.9)	9.7 (9.4)	49.4 (22.9)	18.0 (1.8)	15.5 (10.4)	14.8 (9.6)	0.9 (2.9)
Respiratory therapists	19.9 (17.7)	7.2 (5.9)	34.9 (17.8)	13.4 (1.0)	33.3 (20.6)	21.1 (11.7)	4.5 (6.0)
Environmental services	7.3 (25.7)	2.7 (4.2)	5.1 (9.3)	0.0 (0.0)	9.1 (6.7)	13.7 (8.9)	11.0 (2.7)
Medication availability issues
Antibiotics	6.1 (7.9)	2.7 (4.2)	1.0 (4.2)	0.0 (0.0)	1.7 (3.5)	16.2 (3.1)	14.1 (9.1)
Crystalloid fluids	4.5 (6.2)	2.7 (4.1)	0.0 (0.0)	0.0 (0.0)	1.7 (3.5)	9.7 (5.3)	13.4 (7.0)
Bronchodilators	1.5 (4.0)	0.3 (1.6)	0.0 (0.0)	0.0 (0.0)	1.2 (3.2)	0.5 (2.2)	9.0 (6.7)
Vasopressors	3.1 (4.8)	2.7 (4.1)	0.0 (0.0)	0.0 (0.0)	4.3 (5.5)	2.0 (3.8)	10.8 (2.7)
Neuromuscular blockade agents	3.1 (5.4)	3.5 (5.9)	1.0 (4.2)	0.0 (0.0)	0.0 (0.0)	3.0 (4.1)	11.8 (4.0)
Organ support availability issues
High-flow oxygen	0.2 (1.6)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	1.4 (3.7)	0.0 (0.0)	0.0 (0.0)
Renal replacement therapy	4.6 (7.8)	5.0 (7.0)	9.0 (12.3)	5.4 (7.4)	0.0 (0.0)	5.2 (6.6)	2.1 (4.5)
Extracorporeal membrane oxygenationd	10.0 (13.1)	4.0 (6.5)	20.2 (21.3)	0.0 (0.0)	9.6 (9.7)	14.5 (9.3)	4.8 (6.5)
Personal protective equipment availability issues
Respiratory protection (N95 respirators, powered air purifying respirators)	15.6 (26.7)	7.9 (12.7)	66.9 (15.0)	5.0 (6.8)	0.8 (3.2)	2.4 (3.9)	0.0 (0.0)
Surgical masks	1.4 (4.9)	0.0 (0.0)	4.4 (9.5)	0.0 (0.0)	0.7 (2.9)	1.1 (3.0)	2.1 (4.5)
Eye protection (face shields, goggles)	1.0 (4.1)	0.0 (0.0)	3.9 (8.4)	0.0 (0.0)	1.3 (3.6)	0.0 (0.0)	0.0 (0.0)
Examination gloves	3.1 (8.0)	1.6 (3.4)	4.2 (16.7)	0.0 (0.0)	0.0 (0.0)	5.5 (4.2)	6.6 (4.7)
Environmental hygiene supplies	4.4 (8.8)	1.2 (3.0)	6.3 (18.1)	0.0 (0.0)	6.1 (5.6)	6.7 (3.9)	4.6 (5.0)

^aAncestral Wuhan strain wave includes the third U.S. national wave in Winter 2020–2021 but not two earlier ancestral surges prior to the study period.

^bThe Omicron BA.2.12.1 subvariant wave includes the early rise of Omicron BA.5 and BA.4 subvariants.

^cOrdinal stress measures were dichotomized; see eTables 4–8 (https://links.lww.com/CCM/H293) for ordinal results by hospital-week.

^dExtracorporeal membrane oxygenation availability does not distinguish between availability that normally exists and is currently unavailable or no availability at baseline that is now desired.

Analyses performed by study week. Primary results and percentages are reported as complete case analyses. See sensitivity analysis results for alternative handling of rare missing data. Between-surge periods are weeks with no days with a 7-d rolling average of greater than 60,000 incident severe acute respiratory syndrome coronavirus 2 cases nationally.

Study sites reported perceived unavailability of some hospital resources resulting in potentially avoidable patient harm during 39 (5%) hospital-weeks; alterations in care processes and/or staffing, which were fully compensated for during 260 (34%) hospital-weeks (299 [40%] hospital-weeks had some degree of reported stress); and no stress during 459 (61%) hospital-weeks (eTable 4, https://links.lww.com/CCM/H293). Among perceived staffing deviations (Table 2; and eTable 5 and Appendix E, https://links.lww.com/CCM/H293), a mean of approximately one in five hospitals per week required at least some reassigning of bedside clinical staff, but these deviations were predominantly perceived to be “adequate” utilizing the reassignment of ICU-experienced staff as part of the pandemic response; less than 5% of hospital-weeks were perceived to be “inadequate” within any staffing discipline.

Perceived shortages of medications and routine organ support equipment (i.e., high-flow oxygen and renal replacement therapy) were rare (Table 2; and eTable 6, https://links.lww.com/CCM/H293). Among personal protective equipment (Table 2; and eTable 8 and Appendix E, https://links.lww.com/CCM/H293), respiratory protection equipment (i.e., N95 respirators and powered air purifying respirators) shortages were most commonly perceived (mean of 16% [sd, 27%] of hospitals per week) and reported as most impactful (affecting clinical protocols in 7% of hospital-weeks).

Hospital Stress Measure Trends, Correlation, and Among-Hospital Variation

Figure 1 and eFigure 1 (https://links.lww.com/CCM/H293) report overall hospital, ICU, and ED stress, and county SARS-CoV-2 cases by pandemic week and across pandemic surge periods, and eFigures 2–9 (https://links.lww.com/CCM/H293) report the same stratified by study hospital counties. By visual inspection, there are differences when comparing surge and between-surge periods and patterns of overall hospital, ICU, and ED stress.

Figure 1.: Hospital stress and county severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) cases by pandemic week. The percentage of contributing study hospitals who reported overall hospital stress (red solid), ICU stress (green dotted), and emergency department (ED) stress (blue dashed) (left axis), and county SARS-CoV-2 cases per 100,000 residents for all study hospital counties (black solid; right axis) are plotted per pandemic week from November 2020 to June 2022. The SARS-CoV-2 variant/subvariant–dominated surges are noted in shaded colors. Stress percentages are based on complete case analyses. By visual inspection, overall hospital stress and ICU stress appear more closely related than either are to ED stress and at the end of surge periods, ED stress appears to abate earlier while hospital stress and ICU stress persist.

eTable 9 (https://links.lww.com/CCM/H293) reports within-hospital correlation analysis results between hospital stress measures by hospital-week. Overall, hospital stress was highly correlated with ICU stress (ρ = 0.82; p < 0.0001) and operating differently (ρ = 0.83; p < 0.0001) but was only moderately correlated with ED stress (ρ = 0.52; p < 0.0001). ICU stress and ED stress were themselves only moderately correlated (ρ = 0.53; p < 0.0001). Correlation results were similar across varied imputation approaches.

Figure 2 displays among-hospital variation in hospital stress during Omicron BA.1 subvariant surge. There was significant among-hospital variation in the frequency of perceived overall hospital stress (range 11–94% of weeks), ICU stress (range 8–77% of weeks), and ED stress (range 3–71% of weeks), adjusted for local county SARS-CoV-2 case counts per 100,000 residents, during this pandemic surge period.

Figure 2.: Among-hospital variation in hospital stress during the Omicron BA.1 subvariant surge. The adjusted percentage of weeks during the Omicron BA.1 subvariant surge that overall hospital, ICU, and emergency department (ED) stress were reported, adjusted for local county severe acute respiratory syndrome coronavirus 2 case counts per 100,000 residents, is plotted by study hospital (ranked by overall hospital stress proportion). Error bars represent 95% CIs. There is significant among-hospital variation in the frequency of adjusted overall hospital stress (red, range 11–94% of weeks), ICU stress (green, range 8–77% of weeks), and ED stress (red, range 3–71% of weeks) during this pandemic surge period.

Association of SARS-CoV-2 Cases and Hospital Stress

In the adjusted logistic regression models (eTable 10, https://links.lww.com/CCM/H293), county SARS-CoV-2 cases were associated with the presence of overall hospital stress (odds ratio [OR], 1.087 per change in 10 SARS-CoV-2 cases per 100,000 residents; 95% CI, 1.051–1.125; p = 0.001), ICU stress (OR, 1.065; 95% CI, 1.034–1.096; p < 0.001), and ED stress (OR, 1.038; 95% CI, 1.015–1.061; p = 0.001). A county increase in 10 SARS-CoV-2 cases per 100,000 residents would therefore be expected to increase the odds of perceived overall hospital, ICU, and ED stress in that county by 9% (95% CI, 5–12%), 7% (95% CI, 3–10%), and 4% (95% CI, 2–6%), respectively. Model results were again similar across varied imputation approaches. In stratified post hoc analyses, these results persisted in the Omicron BA.1 subvariant surge and point estimates were amplified in the Delta variant surge but with loss of statistical significance (eTable 10, https://links.lww.com/CCM/H293).

Temporal Relationships Between SARS-CoV-2 Cases and Hospital Stress

Figure 3, Table 3, and eTables 11–13 (https://links.lww.com/CCM/H293) report the temporal relationship between county SARS-CoV-2 cases and hospital stress during the Delta variant and the Omicron BA.1 subvariant surges. During the Delta variant surge (Fig. 3A), the start of ICU stress (median 1 wk after [IQR 1 wk before to 3 wk after] local SARS-CoV-2 case peak; z = 2.04; p = 0.04) and ED stress (median 2 wk [IQR 1–3 wk] after local case peak; z = 2.04; p = 0.04) were both statistically significantly later than the start of overall hospital stress (median 1 wk [IQR 1–2 wk] prior to local case peak). At the end of the Delta variant surge, overall hospital stress persisted for a median of 11.5 weeks (IQR 9–14 wk) after local case peak. In comparison, ICU stress had a similar pattern of resolution (median 11 wk [6–14 wk] after local case peak; z = 0.54; p = 0.59), while the end of ED stress (median 6 wk [IQR 5–6 wk] after local case peak; z = 2.97; p = 0.003) was statistically significantly earlier.

TABLE 3. - Temporal Relationships Between Severe Acute Respiratory Syndrome Coronavirus 2 Cases and Hospital Stress

Stress Measure	Duration From Stress Start to County Case Peak	Duration From County Case Peak to Stress End	Stress Duration	County Cases at Time of Stress Start	County Cases at Time of Stress End
	Wk, Median (IQR)a			Daily Severe Acute Respiratory Syndrome Coronavirus 2 Case Count per 100,000, Median (IQR)
Delta variant surgeb
Overall hospital stress	–1 (–2 to –1)	11.5 (9–14)	11 (11–14.5)	27 (25–32)	13 (10–46)
ICU stress	1 (–1 to 3)d	11 (6–14)	10 (3–14)	27 (24–35)	15 (14–48)
ED stress	2 (1–3)d	6 (5–6)d	3 (3–4)d	24 (24–120)	15 (15–116)
Omicron BA.1 subvariant surgec
Overall hospital stress	–2 (–2 to –2)	6 (5–6.5)	8 (6–9)	98 (81–139)	17 (13–72)
ICU stress	–1 (–2 to –1)	5.5 (4–6)	7 (6–8)	175 (81–296)	38 (17–72)
ED stress	–2 (–2 to –1)	5 (4–6)d	7 (5–8)d	139 (81–175)	83 (17–116)

ED = emergency department, IQR = interquartile range.

^aNegative weeks can be interpreted as stress start occurring before county case peak.

^bFour hospitals excluded due to cohort entry during Delta variant surge or later. Three additional hospitals excluded from ED stress results due to no reported ED stress during the Delta variant surge. One additional hospital excluded from overall hospital, ICU, and ED stress end results due to stress present at end of Delta variant surge unabated to start of subsequent Omicron BA.1 subvariant surge.

^cOne hospital excluded due to cohort entry during Omicron BA.1 subvariant surge. One additional hospital excluded from overall hospital, ICU, and ED stress start results due to stress present at start of Omicron BA.1 subvariant surge and unabated since the prior Delta variant surge.

^dStatistically significantly different compared with overall hospital stress. See eTables 11–13 (https://links.lww.com/CCM/H293) for results of Wilcoxon rank-sum tests.

Figure 3.: Temporal relationships between severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) cases and hospital stress. The black curve reports county SARS-CoV-2 cases per 100,000 residents among study site hospital counties. The start (circles) and end (square) of overall hospital stress (red), ICU stress (green), and emergency department (ED) stress (blue) are plotted such that the x-axis position represents the median weeks relative to the surge case peak and the y-axis position represents the median county SARS-CoV-2 cases per 100,000 residents at each stress start or end time point; error bars represent interquartile ranges (IQRs) for each axis. A, During the Delta variant surge, overall hospital, ICU, and ED stress began within a median of 1–2 wk of case peak, and stress persisted in the overall hospital, ICU, and ED for a median of 11.5 wk (IQR, 9–14 wk), 11 wk (IQR, 6–14 wk), and 6 wk (IQR, 5–6 wk) after county case peak, respectively. B, During the Omicron BA.1 subvariant surge, overall hospital, ICU, and ED stress started a median of 1–2 wk before county case peak, and stress persisted in the overall hospital, ICU, and ED for median of 6 wk (IQR, 5–6.5 wk), 5.5 wk (IQR, 4–6 wk), and 5 wk (IQR, 4–6 wk) after county case peak, respectively. ^Upper bound extends beyond figure range, see Table 3 for complete IQRs.

During the Omicron BA.1 subvariant surge (Fig. 3B), the start of ICU stress (median 1 wk [IQR 1–2 wk] before local case peak; z = 1.64; p = 0.10) and ED stress (median 2 wk [IQR 1–2 wk] before local case peak; z = 1.60; p = 0.11) were not significantly different than the start of overall hospital stress start (median 2 wk [IQR 2–2 wk] before local case peak). At the end of the Omicron BA.1 subvariant surge, overall hospital stress persisted for a median of 6 weeks (5–6.5 wk) after local case peak. In comparison, ICU stress had a similar pattern of resolution (median 5.5 wk [IQR 4–6 wk] after local case peak; z = 1.50; p = 0.13) while the end of ED stress (median 5 wk [IQR 4–6 wk] after local case peak; z = 2.09; p = 0.04) was statistically earlier.

The median county SARS-CoV-2 case levels at which overall hospital, ICU, and ED stress started and ended were not statistically different within each surge (eTable 12, https://links.lww.com/CCM/H293). There were differences between the median county SARS-CoV-2 cases at the time of stress start between the Delta variant and Omicron BA.1 subvariant surges, such that there was not a consistent threshold effect.

DISCUSSION

In this prospective study from the SARI-PREP study group, we report granular, rolling data on qualitative perceived hospital stress due to SARI patients during the COVID-19 pandemic at 13 hospitals from seven health systems in six U.S. states across 85 pandemic weeks including 839 hospital-weeks of data and five pandemic surges. The primary findings of this study include that: 1) perceived hospital stress and care deviations during the COVID-19 pandemic were common and varied by surge period and by hospital, and perceived potentially avoidable patient harm was present but rare; 2) county SARS-CoV-2 case activity was associated with overall hospital, ICU, and ED stress; 3) overall hospital stress due to SARI patients during the COVID-19 pandemic was highly correlated with ICU stress but only moderately correlated with ED stress; and 4) perceived hospital stress measures persisted for weeks after surges peaked, with ED stress resolving earliest.

The survey responses provide a detailed look at perceived stress due to SARI patients during the COVID-19 pandemic over time and in a way that allows for comparisons of hospital and local stress measures and between pandemic phases. This work adds a qualitative component to a growing body of literature measuring quantitative hospital capacity strain and establishing links to adverse patient outcomes during the pandemic (^2,3,12,13). The difference between quantitative stress, which measures individual identified strain metrics, and qualitative “perceived” stress generally and in the present study are important. Qualitative stress perceived by clinicians and operations leadership may be reported at times when quantitative strain measures are low or absent. Potential explanations for this phenomenon include: qualitative stress as a measure of lingering clinician and leadership stress, such as via burnout, but which may still impact care delivery and outcomes; qualitative stress as an early warning of impending or building stress not yet reflected in quantitative metrics; or qualitative stress as a more holistic stress assessment not reliant on a finite list of quantitative stress metrics that may be missing key, unmeasured or unknown variables. Alternatively, qualitative perceived stress assessments may not capture real stress that would have been reflected in quantitative metrics, if known to the assessors, and in that sense reflect a “real-use” stress assessment. Because in the present study perceived patient harm was used as a reflection of stress and was present but rare, we did not investigate the association between either SARS-CoV-2 case counts or hospital stress and perceived patient harm, which has been reported in an expanding literature base (^2,3,12,13). Future work is required to link and investigate the potentially nuanced differences between these quantitative and qualitative stress measures and connections between qualitative stress and patient-level outcomes.

The findings that all hospital stress measures persisted well after local SARS-CoV-2 cases peaked during surges might be important for dynamic staffing and other operational decisions geared toward meeting the increased demands during and importantly after respiratory virus patient surges. The longer persistence of inpatient stress and earlier resolution of ED stress, if confirmed, might reflect both the long clinical course of severe COVID-19 that keeps patients hospitalized well after incident case counts decrease, as well as persistent inpatient organizational disruptions that do not return to baseline quickly after a surge.

Perceived care deviations across a variety of domains—in particular, increasing and reassigning hospital staffing, denying inter-hospital transfers, and canceling elective surgeries—were common (Appendix E, https://links.lww.com/CCM/H293). While care deviations were significantly more common during surge periods, they did not entirely normalize when cases decreased: a mean of over 11% of hospitals per week were still operating differently during in-pandemic between-surge periods. Nearly one in five hospitals per week perceived at least some reassigning of bedside clinical staff and this also dissipated but did not disappear during between-surge periods. Importantly, care deviations as the survey defined them may exist along a spectrum from expected modifications (e.g., increasing ICU-trained staffing) to more abnormal deviations (e.g., use of non-ICU-trained personnel for critical care delivery).

Our study has some important limitations. First, the study design and survey instrument assessed qualitative perceived stress specifically due to SARI patients and did not measure quantitative hospital strain metrics, such as occupancy, or holistic stress due to all patients (SARI and non-SARI). This is particularly important in the interpretation of ED stress patterns, which do not reflect the often-high ED stress that is experienced from all patients, with and without COVID-19 (³¹). In addition, dichotomous stress questions limit further within-variable nuance. Second, predominantly critical care survey respondents with potentially differing degrees of information about other hospital locations and the inclusion of respondents as article authors might lead to social desirability and confirmation bias in responses. While respondents were instructed to find appropriate local sources of information from each level of care within each responding hospital, there is likely bias specifically with respect to the ED stress, which may be under-reported; these results should be interpreted with caution and reassessed in future work. Third, study sites are a majority large academic hospitals; the results are not generalizable to all U.S. hospitals (selection bias). Due to rolling cohort entry, earlier surges reflect data from a smaller number of hospitals and therefore results have less precision. Fourth, the rise of at-home SARS-CoV-2 testing and decreased overall testing and reporting of positive test results implies that more recent case counts are likely underestimates of true case activity (^32,33). Because our analyses focused on time relative to case peak, bias is likely limited. Additionally, because the study period started during the pandemic, we do not have pre-pandemic baseline measurements and cannot comment on how far study hospitals were from routine operations. Fifth, the ascertainment of potentially avoidable patient harm might be limited by lack of respondent knowledge of subtle but impactful harm signals felt better by bedside clinicians or patients and families or only detected in larger quantitative outcomes studies, in addition to underreporting due to fears of negative impact on individual hospitals and health systems. This study measured perceived but not true realized patient harm. Finally, this survey study provides hypothesis-generating data but cannot provide mechanistic understanding about stress relationships, which may require more in-depth ethnographic work or studies designed and powered for causal inference.

As shown in pre-pandemic surveillance work, hospital stress is not limited to public health emergencies and surveillance and early detection of hospital stress or a novel threat is a key under-realized piece of regional, national, and global preparedness for inevitable future events (^16,17). Funded, longitudinally sustainable surveillance networks providing rolling within- and among-hospital stress data would inform dynamic staffing decisions, coordination of regionalized patient transfers, and equipment supply chains.

CONCLUSIONS

Perceived hospital stress and care deviations during the COVID-19 pandemic were common and perceived potentially avoidable patient harm was present but rare. Perceived hospital stress measures persisted for weeks after surges peaked.

ACKNOWLEDGMENTS

We wish to thank the Society of Critical Care Medicine’s Discovery research infrastructure that housed the project, the Centers for Disease Control and Prevention Foundation for funding the research, and the research coordinators and other research staff at all the study and coordinating sites who contributed to the project.