Background
Coronavirus Disease 2019 (COVID-19), caused by infection with Severe Acute Respiratory Syndrome Coronavirus-2, has been declared by the World Health Organization to be a pandemic, with over seven million confirmed cases in the United States [
1,
2]. New York State, including the New York City, became the epicenter of the epidemic in the United States, accounting for more than 23% of the total U.S. cases by the end of May, 2020 [
2]. Such burden of disease is of particular concern since it disproportionately affects communities with considerable health disparities in New York City, where African-Americans and Latinos constitute as much as 53% of the population [
3]. Our medical center is located in such a community in Brooklyn, New York.
The spectrum of COVID-19 presentation ranges from mild influenza-like illness to life-threatening severe respiratory disease requiring ventilatory support [
3]. Comorbid conditions such as hypertension, diabetes mellitus, pulmonary and heart diseases, and demographic factors have been reported to influence outcomes [
4‐
6]. However, the relative influence of each of these comorbidities in different patient populations and age strata has not been assessed, leading to variability in management and outcomes. Key decisions in patient management such as the choice of antibiotic, blood pressure goals, and perhaps most importantly, airway management strategies, have remained variable across or within hospitals.
National health statistics have documented extensive health disparities for Black COVID-19 patients. They suffer a three-fold greater infection rate, and a six-fold greater mortality rate than their white counterparts [
7]. However, limited clinical and laboratory data of prognostic significance from Black COVID-19 patients are available [
8]. A range of cultural, linguistic, and healthcare access barriers have prevented clinical investigation. Our hospital, located in New York City, serves a predominantly Black population, and being declared a COVID-only facility, we were able to maintain a standard quality-of-care across all COVID-19 patients.
Here we explore the clinical aspects of COVID-19 and its outcomes in Black patients. This study evaluated clinical signs and symptoms, laboratory indicators, and management strategies to develop a data-driven COVID-19 patient-care approach. Our findings provide an evidence-based resource for physicians to assess patient progress in the early days of hospitalization to direct patient management decisions.
Methods
This study analyzed the electronic medical records of COVID-19 patients hospitalized at the State University of New York (SUNY), Downstate Medical Center, Brooklyn, New York. The hospital was designated a COVID-only facility by the State of New York as of March 4th, 2020, and provided ample equipment and supplies. The hospital is located in a majority Black neighborhood with high rates of poverty [
9]. This study was approved by the SUNY Downstate Institutional Review Board [1587476–1].
COVID-19 diagnosis was based on clinical presentation and a positive real-time reverse transcriptase polymerase chain reaction (rtPCR) from a nasopharyngeal swab (Xpert Xpress SARS-CoV-2, Cepheid, Sunnyvale, CA). Of the 1103 patients who tested positive over a 52-day period (March 2nd – April 23rd), when the hospital was under peak caseload; 529, who met the following criteria were admitted and included in this study. Patients were admitted if deemed to be in respiratory distress (respiratory rate > 22 breaths/min and in need of supplemental oxygen to maintain oxygen saturation > 92%), were encephalopathic, or were judged sufficiently ill to require hospitalization. Patients were followed up for up to 7 months, thus we have been able to document an outcome (death or discharge) on all patients. COVID-19 positive pregnant patients who came for obstetrics related visit, and otherwise asymptomatic, were excluded.
Demographic factors, comorbidities, presenting clinical symptoms, and outcomes (discharge/death) were recorded for 529 patients. Complete medical history was available for 484 of these patients, however, 45 patients were too sick to respond or were in altered mental status at presentation and were excluded from analyses of co-morbidities. Laboratory data were recorded for 286 patients on admission or within 24 h of hospitalization, and at a second time point between 48 and 96 h post-admission. Pre-admission medications were recorded based on admission medication reconciliation by admitting physicians. Based on self-reported race/ethnicity, patients were grouped into Black and Others (White Hispanic/non-Hispanic and Asian). HIV-positive patients [with CD4 counts < 50% of the lower limit of the reference range (404–1612/μL)] and transplant recipients were categorized as “immunocompromised”. Chronic kidney disease (CKD) was defined as kidney damage and reduced glomerular filtration rate (GFR < 60 ml/min/1.73 m
2) of more than 3 months [
10]. We separated patients with kidney disease into 3 groups: 1) CKD without dialysis, defined as patients who were admitted with baseline CKD and did not receive dialysis during hospitalization; 2) CKD with dialysis, defined as patients with baseline CKD who started dialysis as inpatients because of worsened acute kidney injury; 3) ESRD, defined as patients who were on dialysis prior to admission and continued dialysis as per their routine schedule during hospitalization.
Patients were treated with hydroxychloroquine (200 mg twice a day, for 5 days) and azithromycin (250 mg once a day, for 5 days). All patients received standard venous thromboembolism prophylaxis with low-molecular weight heparin or direct oral anticoagulants based on their creatinine clearance rate. Patients with elevated D-dimer received a full dose anticoagulation regimen. Hypoxia, a sign of Acute Respiratory Distress Syndrome (ARDS), was monitored by a continuous pulse oximeter and with arterial blood gas measurements, and supplemental oxygen was provided as needed via noninvasive ventilation. Patients with worsening respiratory distress despite supportive care, as determined by declining pulse oximeter saturation, increasing respiratory rate, or worsening partial pressure of arterial oxygen/percentage of inspired oxygen ratio) were intubated and placed on mechanical ventilation. Patients who developed acute kidney injury (AKI) with oliguria (< 30 ml/hr. for > 12 h) unresponsive to diuretics or hemodynamic optimization, or decreased creatinine clearance (CrCl < 20 ml/min) received hemodialysis [
11].
Computational analysis was conducted using R (ver. 3.6.3) [
12]. Continuous variables are presented as median and interquartile range (IQR). Categorical variables such as gender or race are presented as number and percent of patients with 95% confidence intervals (CI). Percentages are expressed based on the available data for the subgroup relative to the total available data for that variable.
Parametric variables were evaluated through a Shapiro-Wilk test of normality with a significance cutoff of P < 0.01. Non-parametric variables were compared using Mann-Whitney rank sum test, with 95% CIs reported. Categorical variables were evaluated using the Fisher exact test, and odds ratios (OR) alongside 95% CIs are presented. All tests were two-tailed and statistical significance was defined as P < 0.05. No multiple testing correction was applied. A multivariate logistic regression analysis was performed on comorbidities and demographic factors for in-hospital mortality, and ORs with 95% CIs are presented. Cox proportional hazards analysis for time to death was conducted on comorbidities, demographic factors, and pre-admission medications [(angiotensin-converting enzyme (ACE) inhibitors and/or angiotensin II receptor blockers (ARBs)] and hazard ratios with 95% Cis are presented.
Discussion
This study documents the demographic, clinical features, and outcomes for patients admitted with COVID-19 at an urban hospital located in an underserved majority-Black neighborhood. We also identify indicators available to physicians at two early time points of evaluation to predict outcomes and develop management plans for appropriate levels of care.
The Black patient population in our study faces unique obstacles such as linguistic and cultural barriers to care and understudied comorbidities [
13,
14]. Despite reports that African-Americans face significantly greater mortality from COVID-19, recent studies have examined the clinical outcomes in largely East-Asian or Caucasian cohorts [
13]. Here, we present an analysis of 529 patients admitted with COVID-19, over a 52-day period at the height of the pandemic in New York City, and have either been discharged or died.
Older age at admission correlated with higher mortality rate, with the 60+ year age group most at risk, and was an independent risk factor for mortality. Males suffered significantly higher mortality than females, despite identical representation at admission. Recent reports of high plasma concentrations of ACE-2, a receptor for coronavirus, in men may account for higher mortality [
15]. Our inpatient population had a mean BMI in the “obese” range, higher than the national average; this finding mirrors higher BMI amongst the Black population nationwide [
16] However, BMI was not a predictor of survival; higher BMIs were more commonly seen amongst younger patients. Smoking was less prevalent in our patient population than the national average; 4% were current smokers and 15% had quit [
17]. We found smoking to be unrelated to poor outcome.
The majority (88%) of our patients were Black. Race was not an independent prognostic factor for survival; higher mortality in our patient population can be attributed to a greater number and prevalence of comorbidities common amongst this group. Comorbidities were present in 98% of our patients, and the presence of any comorbidity was a strong predictor of mortality, as noted in other recent studies [
18‐
20]. HT and DM were the two most prevalent preexisting conditions; prevalence of HT (79%) and DM (56%) was considerably higher than previously reported (up to 63 and 36%, respectively) [
21‐
23]. In the multivariate analysis, coronary artery disease was strongly associated with adverse outcome (OR,2.38 CI, 1.11–5.50,
P 0.03), followed by DM (OR, 1.22, CI, 0.81–1.84,
P = 0. 35). A 2.5-fold increase in the risk of mortality from COVID-19 in hypertensive patients has been reported, however, this was not discernable in our patients [
22]. Although past history of cancer, HT, autoimmune diseases, and immunosuppression were not independent predictors of mortality, the combined effect of these comorbidities on multiple organ systems and resultant dysregulation of the immune system likely increases susceptibility to COVID-19 [
23,
24].
A notable finding in multivariate analysis was that patients with CKD who were dialyzed early in the course of treatment had better outcomes than those who did not (2%, OR, 0.27, CI, 0.04–1.11,
P = 0.10). Although not statistically significant, we speculate that a larger number of patients with CKD on dialysis (currently
n = 11) would allow for a definitive conclusion. These findings are notable considering patients with CKD had more comorbidities as compared to all other patients in the study. Early dialysis stands out as a potentially beneficial treatment option for patients with CKD. It is likely that dialysis removes inflammatory mediators, cytokines, and other effector molecules responsible for the end-organ damage. CKD and ESRD were more prevalent in our patient population (26%) than reported in other studies (between 3 to 8.5%), most likely due to complications from HT and DM [
25].
We found laboratory data at admission vital for triaging patients to receive intensive care. CRP, LDH, and procalcitonin were significantly increased at both admission and at 48–96 h in non-survivors. Indicators of AKI, elevated levels of BUN, creatinine, glucose, and reduced levels of bicarbonate or albumin were significant predictors of adverse outcome at both initial and secondary time points. These findings correlate with reported tubular, endothelial, and glomerular capillary loop injury, likely the result of direct injury or systemic hypoxia [
26]. Hypoproteinemia and hypoalbuminemia in non-survivors may result from renal insufficiency and suboptimal nutritional status in critically-ill patients, or could reflect stressed state [
25]. As reported elsewhere, we found hyperglycemia to be a predictor of adverse outcome in COVID-19 patients, regardless of their history of diabetes [
27]. Multivariate analysis of laboratory data was not performed due to sample size limitations.
Peripheral blood analysis showed that a high median NLR at admission and at 48–96 h was an independent predictor of adverse outcome in COVID-19 patients, as had been reported in other studies [
28]. The presence of COVID-19 associated coagulopathy (CAC), a condition characterized by elevation in fibrinogen and D-dimer levels, high PT, relatively normal aPTT, and mild thrombocytopenia without evidence of microangiopathy, was confirmed in our study [
29]. The mechanisms underlying CAC remain poorly understood, but it can possibly result from activation of extrinsic coagulation pathway, leading to excess consumption of Factor-VII following endothelial cell infection by the virus [
30,
31] Elevated D-dimer levels at the second evaluation time point were associated with higher mortality, likely reflecting coagulation activation from sepsis, “cytokine-storm”, or impending organ failure.
By the end of our study, 48% of the inpatients had died, including 86% who received invasive mechanical ventilation. Reported mortality rates from other retrospective cohort studies ranged from 21% (New York metropolitan area) to 26% (Lombardy region, Italy) and 33% (UK) [
4,
6,
32]. Relative to other studies, the mortality rate among our patients was elevated, which we believe is due to the largely poor and disadvantaged neighborhood where our hospital is located. Race was not found to be an independent predictor of mortality. Patients from similar underprivileged communities tend to present at an advanced stage of the disease leading to increased morbidity and mortality [
33]. Rate ratios of hospital admission and mortality in US patients show a 4.7 and 2.1 times higher prevalence among Blacks as compared to Whites [
34].
Our patients from a minority and underserved population had an unusually high burden of co-morbidities some of which proved to be independent predictors of the observed in-hospital high mortality; 1/3 of the patients died within the first 3 days of admission. We found some of the early laboratory data, together with demographics and co-morbidities, pivotal in predicting the clinical course of COVID-19. Early institution of dialysis in patients with chronic renal insufficiency reduced mortality significantly.
Our study has limitations. It examined a predominantly Black patient cohort, which makes comparisons to other races and ethnicities difficult to quantify. This study was carried out on patients admitted at the height of the pandemic in New York City, admissions were restricted to the most seriously ill and hospital resources were under strain, which may have contributed to an increase in overall mortality rates. Initiation of dialysis during admission occurred at the discretion of treating physicians, and there may be unmeasured differences between patients started on dialysis and those not-started on dialysis that are not accounted for in this analysis. As knowledge and understanding of COVID-19 was developing during March and April, complete laboratory studies were not systematically ordered for all patients. The routine use of steroids and Remdesivir were not established yet during the time of this study and so these findings, particularly the mortality rate, should be taken in that context. BMI was not included in the multivariate regression model as BMI was available in only a subset of patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.