key: cord-0901327-phsxlt4g
authors: Rodriguez-Nava, Guillermo; Yanez-Bello, Maria Adriana; Trelles-Garcia, Daniela Patricia; Chung, Chul Won; Chaudry, Sana; Khan, Aimen S.; Friedman, Harvey J.; Hines, David W.
title: Clinical characteristics and risk factors for mortality of hospitalized patients with COVID-19 in a community hospital: A retrospective cohort study
date: 2020-11-05
journal: Mayo Clin Proc Innov Qual Outcomes
DOI: 10.1016/j.mayocpiqo.2020.10.007
sha: 98bc84a4f37add93b7f7b9eb34459bbb615268ff
doc_id: 901327
cord_uid: phsxlt4g

Objective To describe the clinical characteristics, outcomes, and risk factors for death in patients with COVID-19 in a community hospital setting. Patients and methods Single-center retrospective cohort study that included 313 adult patients with laboratory-confirmed COVID-19 admitted to a community hospital in Cook County, Illinois, from March 1, 2020, to May 25, 2020. Demographics, medical history, underlying comorbidities, symptoms, signs, laboratory findings, imaging studies, management, and progression to discharge or death data were collected and analyzed. Results Of 313 patients, the median age was 68 years (interquartile range 59.0 – 78.5; range 19 – 98 years), 182 (58.1%) were males, 119 (38%) were Caucasian, and 194 (62%) were admitted from a Long-Term Care Facility (LTCF). As of May 25, 2020, 212 (67.7%) survivors were identified, whereas 101 (32.3%) non-survivors were identified. Multivariable Cox regression analysis showed increasing hazards of inpatient mortality associated with older age (hazard ratio [HR] 1.02; 95% CI 1.01 - 1.04), LTCF residence (HR 3.23; 95% CI1.68 - 6.20), and quick Sequential Organ Failure Assessment (qSOFA) scores (HR 2.59; 95% CI1.78 - 3.76). Conclusions In this single-center retrospective cohort study of 313 adult patients hospitalized with COVID-19 illness in a community hospital in Cook County, Illinois, older patients, LTCF residents, and patients with high qSOFA scores were found to have worse clinical outcomes and increased risk of death.

On December 31, 2019, the World Health Organization (WHO) received a report of a cluster of cases of pneumonia of unknown etiology detected in Wuhan City, Hubei Province of China [1] . The pathogen was identified as a novel enveloped RNA betacoronavirus, different from both MERS-CoV and SARS-CoV, later designated as SARS-CoV-2 [2] . The disease that it causes was later named Coronavirus Disease 2019 (COVID-19) [3] . This new disease rapidly spread globally, prompting the WHO to declare it on March 11, 2020, as a pandemic. The United States (US) has been, to the date, the country with the highest incidence of the disease. As of June 2, 2020, there had been a total of 1,802,470 cases reported in the US, while in Cook County, Illinois, there had been 78,495 confirmed cases [4, 5] .

After the emergence of this novel pathogen, several cohort and case series studies have described COVID-19 in populations overseas and large US academic centers [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] . Ecological studies have found significant differences within the US territory regarding mortality patterns as a function of geographical location and population composition [18] . Also, detailed data on demographics, comorbidities, risk factors, and clinical outcomes from community hospitals are needed to fully characterize the spectrum of the disease and allow healthcare providers to prepare tailored interventions for their communities. For that reason, the objective of this study is to describe the clinical characteristics, outcomes, and risk factors for death in patients with COVID-19 hospitalized in a community hospital setting.

This retrospective cohort study included 313 hospitalized adult patients (18 years or older) with COVID-19 from Saint Francis Hospital, a 216-bed community hospital located in the North Shore Chicago area, Cook County, Illinois, who had been admitted from March 1, 2020, to May 25, 2020 . A confirmed case of COVID-19 was defined by a positive result on a reverse-transcriptase-polymerase-chainreaction (RT-PCR) assay of a specimen collected on a nasopharyngeal swab. Only patients with laboratory-confirmed illness were included.

Information collected included demographic data, medical history, underlying comorbidities, symptoms, signs, laboratory findings, imaging studies, treatment measures, survival to hospital discharge or data abstraction cutoff date (survivors), and in-hospital mortality or referral to hospice (non-survivors). Clinical outcomes were monitored up to May 25, 2020 ; this was the cutoff date for data abstraction.

The study was approved by the Institutional Review Board of AMITA Health System (2020-0128-02). The Ethics Commission waived the requirement for informed consent given this research involves no more than minimal risk to subjects.

J o u r n a l P r e -p r o o f Descriptive statistics were used to summarize the data; categorical variables were described as frequency rates and percentages, and continuous variables were described using median and interquartile range values. We used the Mann-Whitney U test, χ2 test, or Fisher's exact test to compare differences between survivors and non-survivors when appropriate. Cox proportional hazards (PH) regression model was conducted to examine the relationship between independent variables and mortality, in which hazard ratios (HRs) were used to quantify the associations. Schoenfeld residuals were used to test the PH assumption of Cox models statistically. A two-sided α of less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS Version 23.0.

(Armonk, NY: IBM Corp). Additional details regarding data collection, interrater agreement and reliability, missing values, non-proportionality test, and definitions are provided in the Supplementary Appendix (Appendix 1). In this patient population, non-survivors were significantly older and were more likely to be men, Caucasians, and LTCF residents. A considerably higher proportion of non-survivors had two or more comorbidities compared to survivors.

Survivors were more likely to report influenza-like illness symptoms, including chills, fatigue or malaise, myalgia or body aches, cough, headache, diarrhea, and nausea or vomiting. On the other hand, non-survivors were more likely to present with shortness of breath (Table 1) . On presentation, non-survivors were more likely found with altered mental status, with a lower nadir in oxygen saturation and blood pressure, and with higher respiratory rates. The temperature and heart rate did not differ significantly between groups. Regarding laboratory findings, numerous differences were observed among survivors and non-survivors, including higher white blood cell J o u r n a l P r e -p r o o f count, lower absolute lymphocyte count, higher liver function tests values (except for alkaline phosphatase), and more elevated inflammatory markers (i.e., lactate dehydrogenase, ferritin, procalcitonin, D-dimer, C-reactive protein) and end-organ damage markers (lactic acid, serum creatinine, creatinine kinase, and highsensitivity troponin). Compared to non-survivors, survivors were more likely to present no chest x-rays parenchymal findings on presentation. However, there was no difference between the presence of unilateral, bilateral, or diffuse opacities (Table 2) . Table 3 shows the severity of illness, rates of complications or end-organ damage, and frequency of interventions. Of the 313 patients, 247 (78.9%) met the National Institutes of Health criteria for severe illness. This proportion was higher among non-survivors compared to non-survivors. The median quick Sequential Organ Failure Assessment (qSOFA) score was significantly higher in non-survivors than in survivors. Respiratory failure was the most common complication (68.2%), followed by acute kidney injury (38.2%), and sepsis (35.1%). Non-survivors presented statistically higher rates of complications or end-organ damage as compared to survivors ( Table 3) .

The most common interventions included the use of antibiotics (other than azithromycin, (90.1%), azithromycin (47.6%), and intravenous steroids (42.5%).

Among the 313 patients, 93.3% received either prophylactic or therapeutic anticoagulation, including oral anticoagulants. On presentation, a significantly higher of survivors required no oxygen support as compared to non-survivors. A J o u r n a l P r e -p r o o f total of 98 (31.3%) patients required intensive care unit admission, and nonsurvivors required higher rates of critical care compared to survivors. Additionally, higher rates of non-survivors required intubation, vasopressors, neuromuscular blockers, and new renal replacement therapy (Table 3) .

Until May 25, 2020, 14 (4.5%) patients remained active in the hospital (11 in the ICU and 3 on the general medical floor), 189 (60.4%) patients had been discharged home, 8 (2.6%) had been discharged to long-term acute care hospitals and 1 (0.3%) patient had been transferred to a tertiary center for Extracorporeal Membrane Oxygenation (ECMO). A total of 212 of 313 (67.7%) patients were discharged home or other healthcare facilities. On the other side, 101 (32.3%) patients had died or had been discharged to hospice (88 and 13, respectively).

The median time from symptom onset to admission was 2 days (IQR 1.0 -7.0).

This time was significantly shorter in non-survivors as compared to survivors. The median length of hospital stay (time from admission to event) was 7 days (IQR 4.0 -11-0). The median time from admission to event in survivors (i.e., discharge or last follow-up) was 6 days (4.0 -10.0), whereas the median time to event in nonsurvivors (i.e., death or transition to hospice) was 7 days (3.0 -12.0) ( Table 4 ).

Independent predictors for mortality in the cohort are shown in Table 5 . In the bivariable analysis, older age, Caucasian ethnicity, LTCF residence, hypertension, neurocognitive impairment, altered mental status, low blood pressure, high J o u r n a l P r e -p r o o f respiratory rate, elevated qSOFA scores, high white blood cell count, and increased sodium, blood urea nitrogen, lactic acid, and procalcitonin were associated with increased risk of inpatient mortality in this cohort. Elevated lactate dehydrogenase (HR 1.001, p =.04), D-dimer (HR 1.000007, p <.001), creatinine kinase (HR 1.000032, p =.049), and high-sensitivity troponin (HR 1.001, p =.006) also showed a significant association with increased risk of death. For sensitivity analysis of unmeasured confounders, we calculated the E-value (with the lower confidence limit) described by VanderWeele TJ et al. [19, 20] for the predictors with stronger associations, LTCF residence, and elevated qSOFA score. The E-value is defined as the minimum strength of association on the risk ratio scale that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on the measured covariates, to explain away a specific exposure-outcome association fully. The higher the E-value, the stronger the confounder associations must be to explain away the effect. For LTCF residence, the E-value for the point estimate (HR 7.35) was 6.97 and 3.45 for the CI lower limit (2.79). Then, an unmeasured confounder associated with LTCF residence and inpatient mortality in patients with COVID-19 in this population by an HR of 3.45 fold each could explain away the lower confidence limit, but weaker confounding could not. For qSOFA, the E-value for the estimate (HR 2.90) was 3.56 and 2.57 for the CI lower limit (1.97).

Given the observed clinical characteristics of this patient population and based on previously reported cohorts, we selected older age, LTCF residence status, body mass index (BMI), number of comorbidities, hypertension, neurocognitive impairment, lactic acid, and qSOFA score to fit into the multivariable Cox PH J o u r n a l P r e -p r o o f regression model [21] . LTCF residence and qSOFA remained as reliable predictors for inpatient mortality in this patient population ( 

As of May 25, 2020, of the 313 adult patients with COVID-19 admitted in a community-hospital in Cook County, Illinois, 63.2% had been discharged from the hospital, whereas 28.1% had died and 4.2% had been transitioned to hospice.

Non-survivors were more likely to be older, Caucasians, and LTCF residents.

Additionally, non-survivors overall had more underlying conditions, in particular higher rates of hypertension, neurocognitive impairment, and obesity. These risk factors are similar to previously described US cohorts [10, 11, [13] [14] [15] [16] .

Respiratory failure, acute kidney injury, and sepsis were the most common complications observed. Non-survivors presented statistically higher rates of complications or end-organ damage as compared to survivors. Remarkably, our patient population showed higher rates of some non-pulmonary complications compared to other cohorts. Initial Chinese studies reported acute kidney injury (AKI) in 0.5% to 15% of the patients, while 38.2% of our patients presented AKI [6] [7] [8] [9] 23] . Acute cardiac injury and septic shock were seen in 7.2% to 17%, and in 1.1% to 20% of these patients [6] [7] [8] [9] 23] . In contrast, in our cohort, 21.4% of the patients developed acute cardiac injury, and 19.8% developed septic shock. The rates of respiratory failure were higher in our patients (68.4% vs. 54% in China) [22] . Still, the rate of acute respiratory distress syndrome was similar (3.4% to 31%

in Chinese reports vs. 19 .5% in our population) [6] [7] [8] [9] . Compared to other US J o u r n a l P r e -p r o o f cohorts, the rates of AKI were higher (38.2% vs. 22.2%) [14] , while the rates of septic shock were lower [19.8% vs. 27 .5% to 32.6%) [11, 15] .

The case fatality rate (32.3%) was also two to three times higher than for other US cohorts. Nevertheless, the overall rates of patients requiring critical care were similar (31.3%) to those observed in hospitalized patients from New York City (12.2% to 33.1%), California (8.7% to 30%), and Georgia (39%) [11, [13] [14] [15] [16] [17] .

We hypothesize that the differences in severity and rates of complications and inhospital mortality observed in our patients compared to other US and overseas cohorts are related to a worse baseline clinical status. More than 20 long-term care facilities surround our community hospital, vastly dominating the patient population we serve, and its residents generally have a higher disease burden. Up to 95% of the patients admitted with COVID-19 had at least one underlying condition, 85% had two or more comorbidities, and the median number among the entire cohort was three medical conditions. Even though there was no specific COVID-19 admission policy, given the surge of cases and the limited resources, mostly only patients requiring respiratory support, were admitted. Additionally, many patients presented with advance directives or goals of care were soon established upon admission, with 46% (144/313) of the patients having do not resuscitate (DNR) orders at the cutoff date of data abstraction and LTFC residents being more likely to have this orders (58.8% vs. 25.2%, p<.001).

Bivariable Cox regression showed an increased risk of inpatient mortality associated with older age, Caucasian ethnicity, LTCF residence, a higher number of comorbidities, hypertension, and neurocognitive impairment. Initial abnormal vital signs and laboratories, including altered mental status, hypotension, J o u r n a l P r e -p r o o f tachypnea, qSOFA scores, white blood cell count, sodium, blood urea nitrogen, lactic acid, procalcitonin, lactate dehydrogenase, D-dimer, creatinine kinase, and high-sensitivity troponin were also associated with increased risk of mortality. Other studies have found similar risk factors [22] [23] [24] [25] [26] [27] .

In the multivariable Cox regression, we identified LTCF residence and elevated qSOFA scores as strong independent predictors for mortality. LTCF residents had a probability of 76.3% of dying sooner than patients admitted from home (when Probability = HR / [1+ HR]) [28] , in the same manner, patients with high qSOFA scores had a probability of 72.1% of sooner death compared with those with lower scores. Of mention, we noted an inverse association between current smoker status and risk of death; nevertheless, the vast majority of the hospitalized patients in this cohort already presented with severe disease; hence smoking as a risk factor for worse clinical outcomes cannot be disregarded. In the same way, other apparent risk factors associated with death observed in previous studies, such as obesity, diabetes, chronic obstructive pulmonary disease, and cardiovascular disease, that were not found in this cohort should not be overlooked. The fact that the baseline functional status of our local population seems to be worse compared to other communities also has to be taken into account. This study has several limitations. First, due to the retrospective study design, not all laboratory tests were done in all patients. Some markers previously described, as interleukin-6, were excluded from the analysis given high rates of missing values. Second, this study was conducted at a single-center hospital, likely introducing selection bias and limiting the extrapolation of the findings. Third, a large proportion of patients were admitted with severe disease, limiting the capacity J o u r n a l P r e -p r o o f of our models to assess the relationship between independent variables and mortality compared with patients with mild disease. The possibility to extrapolate our results with populations with better baseline functional status is limited. Fourth, readmissions were not considered for analysis, therefore discharged patients, along with active patients, were right-censored at the cutoff date for data abstraction, which may have introduced bias in the survival analysis.

In conclusion, in this single-center retrospective cohort study of 313 adult patients hospitalized with COVID-19 in a community hospital in Cook County, Illinois, older patients, LTCF residents, and high qSOFA scores were found to have worse clinical outcomes and increased risk of death.

We would like to dedicate this work to all the registered nurses in our institution which exposed themselves more than anyone else to provide compassionate and comprehensive care, despite the incredible workload. They did a fantastic job, and we will forever be grateful. Second, to the ancillary staff, including respiratory therapists, occupational and physical therapists, pharmacists, and non-clinical personnel that allowed us, the physicians, to concentrate and take care exclusively of medical issues during this health care emergency. Third, to the first responders of the City of Evanston, who were always on edge assisting the community. 

World Health Organization. Pneumonia of unknown cause -China

A Novel Coronavirus from Patients with Pneumonia in China

Naming the coronavirus disease (COVID-19) and the virus that causes it

COVID-19) -Cases in

Illinois Department of Public Health

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan

Clinical Characteristics of Coronavirus Disease 2019 in China

Hospitalization Rates and Characteristics of Patients Hospitalized with Laboratory-Confirmed Coronavirus Disease 2019 -COVID-NET, 14 States

Clinical Characteristics of Covid-19 in New York City

Clinical characteristics of coronavirus disease (COVID-19) early findings from a teaching hospital in Pavia

Characteristics of Hospitalized Adults With COVID-19 in an Integrated Health Care System in California

Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the New York City Area

Characteristics and Clinical Outcomes of Adult Patients Hospitalized with COVID-19 -Georgia

Characteristics and outcomes of coronavirus disease patients under nonsurge conditions

incidence, clinical outcomes, and transmission dynamics of hospitalized 2019 coronavirus disease among 9,596,321 individuals residing in California and Washington, United States: a prospective cohort study

Death rate variation in US subpopulations

Sensitivity analysis in observational research: introducing the E-value

Website and R package for computing E-values

Variable selection -A review and recommendations for the practicing statistician

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Severe Outcomes Among Patients with Coronavirus Disease

Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study

Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study

Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease

Clinical characteristics associated with COVID-19 severity in California

Hazard ratio in clinical trials

Outcomes Hospital length of stay 7 (4 -11) 6 (4 -10) 7 (3 -12) .46