Lack of racial and ethnic disparities in mortality in minority patients hospitalised with COVID-19 in a mid-Atlantic healthcare system

Abstract

Introduction

In the USA, minoritised communities (racial and ethnic) have suffered disproportionately from COVID-19 compared with non-Hispanic white communities. In a large cohort of patients hospitalised for COVID-19 in a healthcare system spanning five adult hospitals, we analysed outcomes of patients based on race and ethnicity.

Methods

This was a retrospective cohort analysis of patients 18 years or older admitted to five hospitals in the mid-Atlantic area between 4 March 2020 and 27 May 2022 with confirmed COVID-19. Participants were divided into four groups based on their race/ethnicity: non-Hispanic black, non-Hispanic white, Latinx and other. Propensity score weighted generalised linear models were used to assess the association between race/ethnicity and the primary outcome of in-hospital mortality.

Results

Of the 9651 participants in the cohort, more than half were aged 18–64 years old (56%) and 51% of the cohort were females. Non-Hispanic white patients had higher mortality (p<0.001) and longer hospital length-of-stay (p<0.001) than Latinx and non-Hispanic black patients.

Discussion

In this large multihospital cohort of patients admitted with COVID-19, non-Hispanic black and Hispanic patients did not have worse outcomes than white patients. Such findings likely reflect how the complex range of factors that resulted in a life-threatening and disproportionate impact of incidence on certain vulnerable populations by COVID-19 in the community was offset through admission at well-resourced hospitals and healthcare systems. However, there continues to remain a need for efforts to address the significant pre-existing race and ethnicity inequities highlighted by the COVID-19 pandemic to be better prepared for future public health emergencies.

Correspondence to Dr Sherita Golden; [email protected]

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Early in the COVID-19 pandemic, certain patient populations were disproportionately impacted by the virus, populations that aligned with specific races and ethnicities.

WHAT THIS STUDY ADDS

• In this retrospective, observational study of a large health system’s response to COVID-19 patients, we found that such social variables (race, ethnicity) were not associated with worse health outcomes, such as mortality, once those patient populations were hospitalised.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Further, vaccination status positively influenced health outcomes in minority hospitalised patients by race and ethnicity with COVID-19.

• Future studies should assess how public health efforts and medical responses by health systems could synergistically collaborate to assure dire health outcomes for certain vulnerable populations are minimised.

Background

In the USA, racially and ethnically minoritised communities have suffered disproportionately from COVID-19 compared with non-Latinx white communities.^1–3 During the early stages of the pandemic, specifically between May and August 2020, black persons accounted for 18.7% of overall COVID-19-related mortality, despite representing 12.5% of the US population.² Among Latinx persons infected with COVID-19, mortality increased from 14% between February and May 2020 to 26.4% in August 2020.^{2 3} During later parts of the COVID-19 pandemic—January to July 2022—of the 692 570 adult cases, 13.8% were black/African American and 11.8% were Latinx, showing still an ongoing disproportionate incidence of COVID-19 on these minoritised populations.⁴

Racial and ethnic disparities evident in the COVID-19 public health emergency in the USA have spotlighted individual and contextual factors. Challenges with housing density (eg, multigenerational homes) and non-work from home occupations have been documented for black and Latinx individuals who may result in higher exposure to and acquisition of the virus.^5–7 Residential racial segregation, cumulative stress and barriers to quality healthcare have resulted in a higher prevalence of cardiovascular disease, diabetes, obesity and renal disease in these populations, which may increase the risk of severe, life-threatening COVID-19.^8–14 The inflammatory response to COVID-19¹⁵ is greater in the context of these pre-existing conditions. Adverse socioeconomic factors (eg, neighbourhood composition^16–19) may further compound the effect of pre-existing conditions leading to COVID-19-related disparities in hospitalisation and mortality. In addition, contextual-level variables, perpetuated by structural racism, such as neighbourhood socioeconomic status and disadvantage, may amplify COVID-19-related disparities.¹ Finally, access to high-quality and adequately resourced healthcare for COVID-19 management may play a role in COVID-19-related disparities.

In a large cohort of patients hospitalised for COVID-19 in a healthcare system spanning five hospitals across the mid-Atlantic region, we analysed baseline sociodemographic, clinical characteristics and clinical COVID-19-related outcomes by race and ethnicity. Our primary outcome of interest was in-hospital mortality of patients with COVID-19 while secondary outcomes were hospital length of stay and critical care resource utilisation. Further, we explored such findings before and after 4 months into the COVID-19 pandemic, taking into account more standardisation of severe COVID-19 management as opposed to the early months of the pandemic.

Methods

Patients

Patients were from the Johns Hopkins Health System’s five mid-Atlantic adult hospitals which includes two academic hospitals in urban regions of Baltimore City, MD (Johns Hopkins Hospital and Johns Hopkins Bayview Medical Center), two community hospitals located in suburban regions (Howard County General Hospital, Columbia, Maryland; Suburban Hospital, Bethesda, Maryland, USA) and one hospital located in Washington D. C (Sibley Memorial Hospital, Washington, D.C., USA). Collectively, these five hospitals account for 354 intensive care unit beds and 2513 hospital beds.

Patients were aged 18 years or older and were admitted to the hospital between 4 March 2020 and 27 May 2022 with a diagnosis of COVID-19 based on confirmed SARS-CoV-2 infection according to any nucleic acid test with an Emergency Use Authorisation from the US Food and Drug Administration laboratory. Time zero for all variables was the time of admission order.

The data used were part of JH-CROWN: The COVID Precision Medicine Analytics Platform Registry,^{20 21} which includes sociodemographic variables, medical conditions, laboratory results, medications and interventions, and patient-related outcomes extracted from the JHMI electronic medical record (EMR), Epic. Some patients were included in previous descriptions of the cohort.^21–24

Variables of interest and outcomes

Individual variables collected, such as (self-reported) race and ethnicity, as well as age, sex, body mass index and language status were extracted from the EMR. Race and ethnicity were reported separately in the EMR with participants included in the race or ethnicity analyses based on available reported data for each of these variables of interest. Patients were categorised as ‘non-Hispanic white’, ‘non-Hispanic black’, ‘Latinx’ or ‘other’ according to their race and ethnicity. Vaccination status was also collected, indicating whether participants were fully vaccinated at admission.

ICD-10 codes were used to obtain a medical history including diabetes, chronic kidney disease, chronic obstructive pulmonary disease, hypertension, heart failure, HIV and asthma. Charlson Comorbidity Index, a measure of the impact of diverse comorbidities on a 10-year survival, was also captured on admission.^{25 26} Finally, we extracted from the EMR if patients were admitted from nursing homes versus their own homes.

Laboratory values collected on patients within 24 hours of admission included C reactive protein, ferritin, leucocyte count, lymphocyte count and D-dimer. Acute COVID-19-related morbidities captured for participants by ICD-10 codes during their hospitalisation were pneumonia, sepsis or septic shock, non-septic shock, acute respiratory distress syndrome (ARDS), venous thromboembolism, stroke, acute heart failure and acute kidney injury. Medical interventions captured for patients were those that were supportive (eg, supplemental oxygen, invasive positive pressure ventilation) as well as those that were COVID-19-specific (eg, remdesivir, dexamethasone).

The primary outcome of interest was in-hospital mortality, stratified by individual sociodemographics of race and ethnicity. Secondary outcomes included hospital length of stay (defined by day of admission to day of discharge, patients who were dead or discharged alive after 28 days were censored at 28 days) and critical care resource utilisation (eg, non-invasive and invasive positive pressure ventilation, renal replacement therapy), stratified by race and ethnicity.

Patient and public involvement

Given the moments of the pandemic where certain populations were being disproportionately impacted, this study’s central question arose: how did such populations fair once admitted to the hospital? Given this was a retrospective review of persons admitted to the hospital, a statistical design that was necessary to answer the aforementioned question involved reviewing charts of previously hospitalised patients. Therefore, in accordance with our institutional review board’s review of this retrospective, observational study, these charts were allowed to be reviewed without prior consenting as personal health information would not be compromised.

Statistical analyses

All continuous variables were presented as mean±SE. Categorical variables were summarised as counts and percentages. For comparisons of more than two groups, an analysis of variance was performed. We estimated standardised means and risks using stabilised inverse probability weights, which is a useful approach for model-based standardisation.²⁷ The weights adjusted for age and vaccination status when examining disparities in characteristics and presentation. The weights additionally adjusted for comorbidity when examining disparities in interventions and outcomes. Our goal was to measure disparity, which required adjusting for certain ‘allowable’ differences in outcomes while avoiding adjustment for ‘non-allowable’ differences that are on the causal pathway between structural racism and inequities in healthcare and health.²⁸

We ran multivariable logistic regression analyses to examine mortality among ‘non-Hispanic white’, ‘non-Hispanic black’, ‘Latinx’ and ‘other’ patients, adjusting for age, Charlson Comorbidity Index and vaccination status.²⁹ We used a Cox proportional hazards model to examine length of hospital stay among these same four groups. In addition, for both mortality and length of stay analyses, we used propensity scores to weight measurements for analysis to address potentially extreme differences.^29–31

Missing values were imputed with multiple imputation where incomplete with 10 separate imputations using the R package mice.³² All statistical analyses were completed by using R V.4.2.2.

Sensitivity analyses

The analyses were repeated using the same multivariate logistic regression and Cox proportional hazards model to examine mortality and length of hospital stay among these four groups for patients admitted prior to and after 1 July 2020. Such a date was selected as in-patient management for severe COVID-19 began to be more formalised, specifically for interventions such as remdesivir and dexamethasone, with the latter being introduced into the National Institutes of Health COVID-19 Treatment Guideliens on 24 June 24 2020.³³ For patients prior to 1 July 2020, we only adjusted for age and Charlson comorbidity index since no patient was fully vaccinated before then. For patients after 1 July 2020, we adjusted for age, Charlson Comorbidity Index and vaccination status.

Results

Patient characteristics

Of the 9822 patients admitted with COVID-19 from March 2020 to May 2022, 171 were excluded because they did not specify race and/or ethnicity on admission. The remaining 9651 were included in this analysis. All 9651 had died or been discharged at the time of this analysis. Complete sociodemographic variables are presented in table 1. More than half were aged 18–64 years old (56%). Complete patient clinical characteristics are presented in table 2. Patients’ mean Charlson Comorbidity Index was 2.83±0.03, with the most common comorbidities being hypertension (60%), diabetes (36%) and chronic kidney disease (20%). Pneumonia was the most common clinical acute presentation (63%), followed by sepsis or septic shock (32%) and then acute kidney injury (27%). Some participants had more than one acute presentation (33%). A complete summary of clinical interventions is presented in table 3. The most common medical interventions were oxygen supplementation from nasal cannula (72%), followed by non-invasive positive pressure ventilation and high-flow nasal cannula (18%) and invasive positive pressure ventilation (12%).

Table 1

Sociodemographic variables of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

Patient characteristics	All patients (N=9651)	Non-Hispanic black (N=3693)	Non-Hispanic white (N=3812)	Latinx (N=1323)	Other (N=823)	P value
Age (years)
18–44	2297 (24)	984 (27)	558 (15)	568 (43)	187 (23)	<0.001
45–64	3092 (32)	1372 (37)	978 (26)	473 (36)	269 (33)	<0.001
65–74	1792 (19)	694 (19)	797 (21)	147 (11)	154 (19)	<0.001
>75	2467 (26)	643 (17)	1478 (39)	133 (10)	213 (26)	<0.001
Adjusted for age and vaccination status
Female (%)	4942 (51)	2049 (55)	1877 (49)	601 (45)	414 (50)	<0.001
Body mass index	31.64±0.83	33.02±1.2	30.86±1.36	32.21±1.98	28.18±2.42	0.245

Bolded p-values were found to statistically significant.

Table 2

Clinically presentation of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

Patient characteristics	All patients (N=9651)	Non-Hispanic black (N=3693)	Non-Hispanic white (N=3812)	Latinx (N=1323)	Other (N=823)	P value
		Expected	Expected	Expected	Expected
Adjusted for age and vaccination status
Charlson Comorbidity Index	2.83±0.03	3.45±0.05	2.68±0.05	2.05±0.08	2.02±0.1	<0.001
Laboratory values
CRP (mg/L)	8.05±0.12	7.9±0.2	7.72±0.18	9.63±0.31	7.71±0.35	<0.001
Ferritin (ng/mL)	990.71±43.4	1099.3±53.14	930.16±71.67	874.49±74.64	970.58±100.84	0.022
Leucocyte count (×109/L)	8.11±0.09	7.66±0.14	8.6±0.13	8.24±0.22	7.62±0.28	0.001
Lymphocyte count (×109/L)	1.26±0.06	1.28±0.1	1.37±0.1	1.05±0.16	1.04±0.2	0.203
D-dimer (ng/mL)	2.5±0.06	2.86±0.11	2.43±0.09	2.06±0.16	1.9±0.21	<0.001
Chronic conditions (%)
HIV	162 (2)	130 (4)	23 (1)	7 (1)	2 (0)	<0.001
Hypertension	5776 (60)	2575 (70)	2091 (55)	683 (52)	413 (50)	<0.001
Diabetes	3456 (36)	1595 (43)	1070 (28)	483 (37)	312 (38)	<0.001
Chronic kidney disease	1944 (20)	1015 (27)	621 (16)	197 (15)	108 (13)	<0.001
Heart failure	1835 (19)	861 (23)	717 (19)	156 (12)	96 (12)	<0.001
Asthma	1297 (13)	673 (18)	450 (12)	93 (7)	81 (10)	<0.001
COPD	936 (10)	393 (11)	454 (12)	47 (4)	36 (4)	<0.001
Acute presentations (%)
Pneumonia	6123 (63)	2303 (62)	2310 (61)	959 (72)	551 (67)	<0.001
Sepsis or septic shock	3057 (32)	1072 (29)	1239 (33)	489 (37)	258 (31)	<0.001
Shock	249 (3)	103 (3)	102 (3)	31 (2)	13 (2)	0.176
ARDS	965 (10)	327 (9)	381 (10)	172 (13)	86 (10)	<0.001
VTE	327 (3)	163 (4)	113 (3)	40 (3)	11 (1)	<0.001
Acute heart failure	1188 (12)	555 (15)	476 (12)	84 (6)	66 (8)	<0.001
Acute kidney injury	2620 (27)	1259 (34)	981 (26)	237 (18)	141 (17)	<0.001
Stroke	147 (2)	73 (2)	54 (1)	14 (1)	6 (1)	0.004

Bolded p-values were found to statistically significant.

ARDS, acute respiratory distress syndrome; COPD, chronic obstructive pulmonary disease; CRP, C reactive protein; VTE, venous thromboembolism.

Table 3

Medical interventions of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

	All patients (N=9651)	Non-Hispanic black (N=3693)	Non-Hispanic white (N=3812)	Latinx (N=1323)	Other (N=823)	P value
		Expected	Expected	Expected	Expected
Interventions*
Respiratory therapy (%)
Supplemental oxygen	6965 (72)	2542 (69)	2854 (75)	964 (73)	604 (73)	<0.001
Non-invasive PPV	1719 (18)	610 (17)	728 (19)	226 (17)	155 (19)	0.024
Invasive PPV	1147 (12)	424 (11)	444 (12)	184 (14)	96 (12)	0.05
Vasopressor (%)	238 (2)	89 (2)	94 (2)	34 (3)	21 (3)	0.978
ECMO (%)	48 (0)	13 (0)	24 (1)	9 (1)	2 (0)	0.15
Dexamethasone (%)	4401 (46)	1575 (43)	1873 (49)	547 (41)	403 (49)	<0.001
Remdesivir (%)	3487 (36)	1235 (33)	1455 (38)	467 (35)	330 (40)	<0.001
Renal replacement therapy (%) (%)	144 (1)	55 (1)	53 (1)	22 (2)	14 (2)	0.717

Bolded p-values were found to statistically significant.

*Adjusted for age, Charlson Comorbidity Index and vaccination status.

ECMO, extracorporeal membrane oxygenation; PPV, Positive Pressure Ventilation.

Patient characteristics by race/ethnicity

43% of Latinx participants were aged 18–44 years, compared with 27% of non-Hispanic black, 15% of non-Hispanic white and 23% of other participants in this age category (p<0.001) (table 1). In general, non-Hispanic black participants had greater c-morbidities compared with participants in the remaining three groups, with greater mean Charlson Comorbidity Index (3.45±0.05 vs 2.68±0.05, 2.05±0.08 and 2.02±0.1 in non-Hispanic white, Latinx and other, respectively, p<0.001). Latinx participants had higher mean CRP (9.63±0.26 mg/L vs 7.9±0.2 mg/L, 7.72±0.18 mg/L and 7.71±0.35 mg/L in non-Hispanic black, non-Hispanic white and other, respectively, p<0.001). Non-Hispanic white participants had higher leucocyte count (8.6±0.13×10⁹/L vs 7.66±0.14×10⁹/L, 8.24±0.22×10⁹/L and 7.62±0.28×10⁹/L in non-Hispanic black, Latinx and other, respectively, p=0.001) (table 2). Compared with participants in non-Hispanic black, non-Hispanic white and other groups, Latinx participants were more likely to present with pneumonia (72% vs 62%, 61% and 67%, respectively, p<0.001), ARDS (13% vs 9%, 10% and 10%, respectively, p<0.001) and sepsis or septic shock (37% vs 29%, 33% and 31%, respectively, p=0.020) (table 2). As for clinical interventions, participants in other group were more likely to receive remdesivir (40% vs 33%, 38% and 35%, p<0.001) and dexamethasone (50% vs 42%, 49% and 41%, p<0.001) as compared with non-Hispanic black, non-Hispanic white and Latinx participants (table 3).

Propensity score weighted generalised linear model

In the propensity score weighted generalised linear model method, adjusted for age, Charlson Comorbidity Index and vaccination status, there was a statistically significant difference in mortality among non-Hispanic black (5.8%), non-Hispanic white (7.7%), Latinx (6.2%) and other participants (7.2%) (p<0.001) (table 4). In the propensity score weighted Cox proportional hazards method, in an effort to compare all four groups (adjusting for age, Charlson Comorbidity Index and vaccination status), there was a statistically significant difference in hospital length of stay among the four groups (p<0.001; data are not shown).

Table 4

Mortality counts (percentage) of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

	All patients	Non-Hispanic black	Non-Hispanic white	Latinx	Other	P value
		Expected	Expected	Expected	Expected
All patients
Mortality (%)*	646 (7)	212 (6)	294 (8)	82 (6)	59 (7)	<0.001
Participants prior to 1 July 2020
Mortality (%)*	149 (8)	52 (8)	45 (9)	35 (6)	16 (8)	0.342
Participants after 1 July 2020
Mortality (%)*	473 (6)	149 (5)	238 (7)	47 (6)	41 (6)	<0.001

Bolded p-values were found to be statistically significant.

*Adjusted for age, Charlson Comorbidity Index and vaccination status.

We conducted pairwise comparisons of mortality and length of hospital stay among the four groups. For mortality, statistically significant differences were found for pair-wise comparisons of non-Hispanic black versus non-Hispanic white (p<0.001) and non-Hispanic white versus Latinx (p=0.025) (table 5). For length of hospital stay, there was only a statistically significant difference between non-Hispanic black and non-Hispanic white patients (adjusted HR: 0.916, p<0.001) (table 6). There were no statistically significant differences in length of hospital stay between other pairs.

Table 5

Pairwise comparisons (p values) for mortality of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

All patients (N=9651)
	Non-Hispanic black (N=3693)	Latinx (N=1323)	Other (N=823)	Non-Hispanic white (N=3812)
Non-Hispanic black		0.435	0.055	<0.001
Latinx			0.277	0.025
Other				0.527
Non-Hispanic white

Participants prior to 1 July 2020 (N=1938)
	Non-Hispanic black (N=685)	Latinx (N=545)	Other (N=190)	Non-Hispanic white (N=518)
Non-Hispanic black		0.264	0.704	0.454
Latinx			0.244	0.081
Other				0.878
Non-Hispanic white

Participants after 1 July 2020 (N=7713)
	Non-Hispanic black (N=3008)	Latinx (N=778)	Other (N=633)	Non-Hispanic white (N=3294)
Non-Hispanic black		0.133	0.060	<0.001
Latinx			0.696	0.132
Other				0.360
Non-Hispanic white

Bolded p-values were found to be statistically significant.

Table 6

Pairwise comparison (adjusted HRs; p values) for length of hospital stay of patients with COVID-19 by race and ethnicity admitted to the Johns Hopkins Health System between March 2020 and May 2022

All patients (N=9651)
	Non-Hispanic black	Latinx	Other	Non-Hispanic white
Non-Hispanic black		0.974 (0.448)	0.948 (0.208)	0.916 (<0.001)
Latinx			0.973 (0.573)	0.941 (0.078)
Other				0.967 (0.433)
Non-Hispanic white

Participants prior to 1 July 2020 (N=1938)
	Non-Hispanic black	Latinx	Other	Non-Hispanic white
Non-Hispanic black		1.045 (0.501)	0.928 (0.430)	0.990 (0.876)
Latinx			0.957 (0.222)	0.888 (0.425)
Other				1.067 (0.502)
Non-Hispanic white

Participants after 1 July 2020 (N=7713)
	Non-Hispanic black	Latinx	Other	Non-Hispanic white
Non-Hispanic black		0.970 (0.483)	0.966 (0.474)	0.892 (<0.001)
Latinx			0.996 (0.947)	0.920 (0.053)
Other				0.923 (0.095)
Non-Hispanic white

Bolded p-values were found to be statistically significant.

In sensitivity analyses, there were statistically significant differences in mortality and length of stay among these four groups after 1 July 2020 (p<0.001) while there was no statistically significant difference in participants prior to 1 July 2020 (p=0.342) (tables 4–6).

Discussion

In this large multihospital cohort of patients admitted with COVID-19 from March 2020 to May 2022, non-Hispanic black patients and Latinx patients did not have more in-hospital mortalities and hospital lengths-of-stay as compared with white patients. Compared with white patients, black patients had a higher prevalence of pre-existing conditions, such as obesity, diabetes and hypertension. Latinx patients were more likely to be younger when compared with black patients and white patients, and Latinx patients had higher rates of ARDS and sepsis or septic shock during their acute presentation to the hospital. When these factors were taken into consideration, non-Hispanic black and Hispanic patients did not have worse mortality or LOS than white patients. Further, when the data were dichotomised to admissions before 1 July 2020 and after 1 July 2020, it was non-Hispanic whites who had greater in-hospital mortality and hospital length of stay as compared with non-Hispanic black patients for admissions after 1 July 2020.

Studies of COVID-19-related outcomes document significant, disproportinate incidence, morbidity and mortality across US health systems by race and ethnicity.^{34 35} For example, the majority of adult COVID-19 deaths in Denver, Colorado were among persons of Latinx ethnicity (51%), even though adult Latinx persons comprised 24.9% of Denver’s population, between March and October 2020.³⁶ Such findings reaffirm national reporting that Latinx patients had a greater likelihood of COVID-19-related death than non-Latinx patients.³⁷ Similarly, we found that Latinx patients comprised a significant portion of our adult patient hospitalisations, although Latinx adults comprise 10.6% of the population in Maryland and 11.3% of the population in Washington, D.C.^{38 39} Latinx patients, when compared with other populations designated by race and ethnicity in our cohort, presented with more dire pathological consequences of COVID-19—ARDS and sepsis or septic shock—even though the Latinx population was younger and had a lower Charlson Comorbidity Index. Presenting with such advanced consequences of severe COVID-19 can likely be explained by the lack of healthcare access common in the Latinx populations where early symptoms could have been managed before progression to life-threatening syndromes.^{40 41} In fact, a significant portion of the Latinx patients in the Baltimore City community often were referred to the hospital by local community leaders, not clinicians.⁴² Such lack of healthcare access can be due to uninsured status as well as awareness of primary care modalities^{41 42}; therefore, such a modifiable factor should be explored to assure that marginalised Latinx populations have proper access to primary care before and during public health crises.

Disparities in COVID-19 have also been documented by race, specifically among black patients. Price-Haywood et al identified disparities in hospitalisations in the early weeks of the global public health crisis (1 March 2020 to 11 April 2020), where the majority of the COVID-19 hospitalised patients in a Louisiana-based health system (Ochsner) were black (76.9%), despite black patients comprising 31% of patients routinely cared for by Ochsner Health System.³⁵ Similar to the black patients in that study, black patients admitted to the Johns Hopkins Health System had higher rates of pre-existing conditions, such as diabetes and hypertension and had a higher Charlson Comorbidity Index when compared with White patients. Similar to Price-Haywood et al, we found that black race was not associated with in-hospital COVID-19-related mortality as compared with white race in the early months of the COVID-19 pandemic. However, after 1 July 2020, we found a significant difference in in-hospital mortality and length of stay between black and white patients, with non-Hispanic white patients having higher mortality and greater length of stay than non-Hispanic black patients.

One factor that warrants further discussion in the discrepancies seen in race and ethnicity is age. Non-Hispanic whites had 59% of their population aged 65 and older, as compared with 21% of the Latinx population and 36% of the non-Hispanic black population (table 1). Even though we adjusted for age in our model, it is possible that other factors that were associated with age influenced outcomes. Given that dexamethasone was a significant change in severe COVID-19 management after July 2020, exploring the use of the steroid may provide insight into mortality differences in our cohort. In the dexamethasone arm of the RECOVERY trial, half of the participants were aged 70 and older in both the intervention and placebo group.⁴³ In the CoDEX trial, dexamethasone was found to be beneficial in patients with ARDS from COVID-19; however, the mean age of the populations studied was 60.1±15.8 (intervention group) and 62.7±13.1 (placebo).⁴⁴ The reason to emphasise these two trials is that dexamethasone became a standard therapy used in the care of severe COVID-19. However, it is unclear if that survival benefit was similar across all age groups as the authors did not explore survival benefit by age. Interestingly, Søvik et al found that patients who developed superinfections while receiving mechanical ventilation due to COVID-19 infection had received dexamethasone more often as compared with those who did not develop a superinfection.⁴⁵ While we did not explore superinfections in our cohort, attempting to understand why one racial group had worse outcomes, especially after a specific time stamp, warrants ongoing investigation, especially if it turns out that a potent intervention, such as corticosteroids, may warrant dose changes or duration adjustments for elderly populations.⁴⁶

The ability of a healthcare system to mitigate the impact of pre-existing health disparities on COVID-19 outcomes through its clinical care delivery is important given the disease’s disproportionate impact on marginalised communities. It is worth paralleling our findings, specifically of our cohort prior to 1 July 2020, to those for sepsis and septic shock, a pathological syndrome that resembles a portion of the COVID-19 pathological syndrome, specifically in moderate and severe presentations.⁴⁷ Jones et al showed that after adjusting for hospital characteristics (eg, teaching status, rural vs urban location, hospital size and ownership), sepsis mortality rates in 2013 were similar between white, black and Hispanic patients.⁴⁸ Prior to adjusting for hospital characteristics, sepsis-related mortality rates were higher in all minoritised groups.⁴⁸ The authors concluded that hospital characteristics contributed to higher rates of in-hospital sepsis mortality for black and Hispanic patients. In addition to hospital healthcare status and resources, neighbourhood-level factors contribute to sepsis-related disparities in mortality and hospital readmissions.^{49 50} One consideration could be optimising hospital systems to also address several social determinants of health to achieve community-level justice in regard to health. When differences do arise, it is worth re-exploring the impact of community-level, individual-level, hospital-level factors or a combination of the three on outcomes. As in our case, we did not explore contextual-level factors, and we controlled for hospital-level factors given we analysed only one health system.

During the pandemic, US-based hospitals made significant changes in a short period of time to adapt to the crisis. For instance, since the majority of patients with COVID-19 are cared for in general medical wards, many hospitals converted such medical wards into respiratory isolation units, allocating staff and resources to care for the patients.⁵¹ For patients critically ill with COVID-19 who developed hypoxaemia due to ARDS, prone position has been identified as a key intervention shown to improve COVID-19-related hypoxaemia.⁵² Prone position requires assistance of a number of individuals, and prior reports identified the lack of adequate staffing as a barrier to a prone position for ARDS.⁵³ One adaptation hospitals have executed to overcome this barrier was to have dedicated trained individuals to assist with prone positioning.⁵³ In our own hospital system, we implemented dedicated respiratory isolation units for persons with COVID-19, a team to implement prone positioning, and even dedicated vascular access teams for patients with COVID-19. In addition, we launched a language support consultation team, Juntos (‘Together’), which deployed qualified bilingual and culturally competent clinicians and social workers to work collaboratively with the primary team to optimise clinical care communication, engage family members as appropriate, and address relevant issues that might have affected recovery and safe discharge.⁵⁴ Such immediate, multimodal adaptations by major hospitals and their respective systems may help explain why several studies have not found significant in-hospital mortality differences among races and ethnicities despite certain groups presenting with greater acute illness severity. Our data suggest that highly resourced hospitals that can quickly implement these types of adaptations in the setting of COVID-19 may be a key to mitigating disparities in COVID-19 mortality and hospital length of stay. This makes it imperative to ensure that lower resourced, public hospitals, which care for a large proportion of black and Hispanic patients with pre-existing conditions from marginalised communities, need to be adequately resourced to support care of these communities not only during the pandemic but also in the post-COVID-19 era so that they are better prepared for future public health emergencies.

Our study should be interpreted in the context of the following limitations. First, our study was an observational study that evaluated COVID-19-related outcomes stratified by individual factors of race and ethnicity, with adjustments for other individual sociodemographics and comorbidity. The findings around racial and ethnic differences in in-hospital COVID-19 presentations and outcomes may be due to extensive factors outside of the scope of a healthcare system’s control, yet are important to understand and warrant further exploration. It is unclear if in-hospital clinical outcomes would differ in other healthcare and hospital systems if resources or intensive care expertise varied. However, while hospitals may level the disparities in clinical outcomes on presentation, health system resources alone cannot mitigate the pre-existing health-related inequities, warranting an emphasis on policy and advocacy to address safe and affordable housing, employment, education and other social determinants of health. Second, we did not take into account the day of COVID-19 diagnosis and day of hospital admission relative to the symptom onset. A delay in diagnosis or presentation to the hospital may result in a higher level of acuity on admission as seen in the Latinx population. However, even if such a delay existed, its impact may not be on mortality or in-hospital length of stay but rather on variables such as morbidity outcomes (eg, discharge to ongoing rehabilitation, chronic ventilator units or readmissions), which were not evaluated. These outcomes warrant further investigation, especially as they will capture the full extent of COVID-19’s impact on the US healthcare system. Third, it should be recognised that some of our statistically significant findings in this study (eg, as it applies to composite scores such as the Charlson Comorbidity Index or laboratory data) while statistically significant, may not be clinically meaningful or significant. Fourth, we excluded patients who did not identify as white or black. The exclusion was in an effort to retain consistency with current evidence that was being emphasised during the pandemic; specifically, black and white as well as Latinx and non-Latinx comparisons. Further emphasis on other racial categories and COVID-19-related outcomes should be a priority for future assessments. Finally, while this study evaluated a significant portion of mid-Atlantic hospitalisations to a large academic and community healthcare system, our findings may not be generalisable to other settings, especially in paediatric populations (since this was a review of only adult patients). Insight into outcomes from other regions of the USA, especially rural hospitals, must be a priority in order to strategise equitable interventions to enhance resources for the ongoing COVID-19 and potential future pandemics.

In conclusion, in this large multihospital cohort of patients admitted with COVID-19, after accounting for comorbidities and acute presentation, non-Hispanic black and Hispanic patients did not have worse in-hospital mortality and length of stay outcomes than white patients. These findings likely reflect how well-resourced hospitals and healthcare systems can assist in levelling the inequities in life-threatening presentations and survival that plague certain marginalised communities. However, we must continue to address the structural and societal factors that resulted in minoritised populations being significantly impacted by COVID and other public health emergencies. While healthcare systems remain prepared within the confines of the walls of a hospital for ongoing public health crises, consideration for addressing contextual-level factors in the communities and neighbourhoods that give rise to these identified pre-existing health disparities is still warranted.

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The Institutional Review Board at the Johns Hopkins School of Medicine approved this study (IRB00250975) for each hospital and all actions undertaken by the authors were in accordance with the Declaration of Helsinki.

Footnote

PG and BG contributed equally.

Contributors PG and BG both shared equally in the writing and creating of the initial manuscript, overseen by SG. All others assisted in identifying key themes to explore for the manuscript. DY, YX, JP and JWJ assisted with the statistical analyses. PG and BG both are acting as guarantors. All authors assisted in review of the manuscript, editing and revisions.

Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Ogedegbe G, Ravenell J, Adhikari S, et al. Assessment of racial/ethnic disparities in hospitalization and mortality in patients with COVID-19 in New York City. JAMA Netw Open 2020; 3: e2026881. doi:10.1001/jamanetworkopen.2020.26881

² Gold JAW, Rossen LM, Ahmad FB, et al. Race, ethnicity, and age trends in persons who died from COVID-19 - United States, may-August 2020. MMWR Morb Mortal Wkly Rep 2020; 69: 1517–21. doi:10.15585/mmwr.mm6942e1

³ Wortham JM, Lee JT, Althomsons S, et al. Characteristics of persons who died with COVID-19 - United States, February 12-May 18, 2020. MMWR Morb Mortal Wkly Rep 2020; 69: 923–9. doi:10.15585/mmwr.mm6928e1

⁴ Boehmer TK, Koumans EH, Skillen EL, et al. Racial and ethnic disparities in outpatient treatment of COVID-19 - United States, January-July 2022. MMWR Morb Mortal Wkly Rep 2022; 71: 1359–65. doi:10.15585/mmwr.mm7143a2

⁵ Haynes N, Cooper LA, Albert MA, et al. At the heart of the matter: unmasking and addressing the toll of COVID-19 on diverse populations. Circulation 2020; 142: 105–7. doi:10.1161/CIRCULATIONAHA.120.048126

⁶ Zaeh SE, Page KR, Berger ZD, et al. Juntos en La Pandemia de COVID-19 (together in the COVID-19 pandemic): health-care professionals and the latinx community. Lancet Respir Med 2021; 9: 13–5. doi:10.1016/S2213-2600(20)30519-1

⁷ McClure ES, Vasudevan P, Bailey Z, et al. Racial capitalism within public health-how occupational settings drive COVID-19 disparities. Am J Epidemiol 2020; 189: 1244–53. doi:10.1093/aje/kwaa126

⁸ Chidambaram V, Tun NL, Haque WZ, et al. Factors associated with disease severity and mortality among patients with COVID-19: a systematic review and meta-analysis. PLoS One 2020; 15: e0241541. doi:10.1371/journal.pone.0241541

⁹ Richardson S, Hirsch JS, Narasimhan M, et al. Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York City area. JAMA 2020; 323: 2052–9. doi:10.1001/jama.2020.6775

¹⁰ Yancy CW. COVID-19 and African Americans. JAMA 2020; 323: 1891–2. doi:10.1001/jama.2020.6548

¹¹ Emami A, Javanmardi F, Pirbonyeh N, et al. Prevalence of underlying diseases in hospitalized patients with COVID-19: a systematic review and meta-analysis. Arch Acad Emerg Med 2020; 8: e35.

¹² Javanmardi F, Keshavarzi A, Akbari A, et al. Prevalence of underlying diseases in died cases of COVID-19: a systematic review and meta-analysis. PLoS One 2020; 15: e0241265. doi:10.1371/journal.pone.0241265

¹³ Bailey ZD, Feldman JM, Bassett MT. How structural racism works - racist policies as a root cause of U.S. N Engl J Med 2021; 384: 768–73. doi:10.1056/NEJMms2025396

¹⁴ Golden SH, Joseph JJ, Hill-Briggs F. Casting a health equity lens on Endocrinology and diabetes. J Clin Endocrinol Metab 2021; 106: e1909–16. doi:10.1210/clinem/dgaa938

¹⁵ Klocperk A, Bloomfield M, Parackova Z, et al. Complex immunometabolic profiling reveals the activation of cellular immunity and biliary lesions in patients with severe COVID-19. J Clin Med 2020; 9: 3000. doi:10.3390/jcm9093000

¹⁶ Beelen R, Hoek G, Houthuijs D, et al. The joint association of air pollution and noise from road traffic with cardiovascular mortality in a cohort study. Occup Environ Med 2009; 66: 243–50. doi:10.1136/oem.2008.042358

¹⁷ Shalev I, Moffitt TE, Sugden K, et al. Exposure to violence during childhood is associated with Telomere erosion from 5 to 10 years of age: a longitudinal study. Mol Psychiatry 2013; 18: 576–81. doi:10.1038/mp.2012.32

¹⁸ Kliewer W. Violence exposure and cortisol responses in urban youth. Int J Behav Med 2006; 13: 109–20. doi:10.1207/s15327558ijbm1302_2

¹⁹ Gallo LC, Fortmann AL, de Los Monteros KE, et al. Individual and neighborhood socioeconomic status and inflammation in Mexican American women: what is the role of obesity Psychosom Med 2012; 74: 535–42. doi:10.1097/PSY.0b013e31824f5f6d

²⁰ PMAP: the Johns Hopkins precision medicine analytics platform. 2020. Available: https://pm.jh.edu [Accessed 17 May 2020 ].

²¹ Garibaldi BT, Fiksel J, Muschelli J, et al. Patient trajectories among persons hospitalized for COVID-19: a cohort study. Ann Intern Med 2021; 174: 33–41. doi:10.7326/M20-3905

²² Ignatius EH, Wang K, Karaba A, et al. Tocilizumab for the treatment of COVID-19 among hospitalized patients: a matched retrospective cohort analysis. Open Forum Infect Dis 2021; 8: ofaa598. doi:10.1093/ofid/ofaa598

²³ Andersen KM, Mehta HB, Palamuttam N, et al. Association between chronic use of immunosuppresive drugs and clinical outcomes from coronavirus disease 2019 (COVID-19) hospitalization: a retrospective cohort study in a large US health system. Clin Infect Dis 2021; 73: e4124–30. doi:10.1093/cid/ciaa1488

²⁴ Wongvibulsin S, Garibaldi BT, Antar AAR, et al. Development of severe COVID-19 adaptive risk predictor (SCARP), a calculator to predict severe disease or death in hospitalized patients with COVID-19. Ann Intern Med 2021; 174: 777–85. doi:10.7326/M20-6754

²⁵ Hall WH, Ramachandran R, Narayan S, et al. An electronic application for rapidly calculating Charlson Comorbidity score. BMC Cancer 2004; 4: 94. doi:10.1186/1471-2407-4-94

²⁶ Quan H, Li B, Couris CM, et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol 2011; 173: 676–82. doi:10.1093/aje/kwq433

²⁷ Sato T, Matsuyama Y. Marginal structural models as a tool for standardization. Epidemiology 2003; 14: 680–6. doi:10.1097/01.EDE.0000081989.82616.7d

²⁸ Jackson JW. Meaningful causal Decompositions in health equity research: definition, identification, and estimation through a weighting framework. Epidemiology 2021; 32: 282–90. doi:10.1097/EDE.0000000000001319

²⁹ Krieger N. On the causal interpretation of race. Epidemiology 2014; 25: 937. doi:10.1097/EDE.0000000000000185

³⁰ VanderWeele TJ. Mediation analysis: a practitioner’s guide. Annu Rev Public Health 2016; 37: 17–32. doi:10.1146/annurev-publhealth-032315-021402

³¹ Karlson KB, Holm A, Breen R. Comparing regression coefficients between same-sample nested models using logit and probit: a new method. Sociological Methodology 2012; 42: 286–313. doi:10.1177/0081175012444861

³² Lee KJ, Carlin JB. Multiple imputation for missing data: fully conditional specification versus multivariate normal imputation. Am J Epidemiol 2010; 171: 624–32. doi:10.1093/aje/kwp425

³³ National Institutes of health COVID-19 treatment guidelines archives. Available: https://www.covid19treatmentguidelines.nih.gov/about-the-guidelines/guidelines-archive [Accessed 27 Feb 2023 ].

³⁴ Gold JAW, Wong KK, Szablewski CM, et al. Characteristics and clinical outcomes of adult patients hospitalized with COVID-19. MMWR Morb Mortal Wkly Rep 2020; 69: 545–50. doi:10.15585/mmwr.mm6918e1

³⁵ Price-Haywood EG, Burton J, Fort D, et al. Hospitalization and mortality among black patients and white patients with COVID-19. N Engl J Med 2020; 382: 2534–43. doi:10.1056/NEJMsa2011686

³⁶ Podewils LJ, Burket TL, Mettenbrink C, et al. Disproportionate incidence of COVID-19 infection, hospitalizations, and deaths among persons identifying as Hispanic or Latino - Denver, Colorado March-October 2020. MMWR Morb Mortal Wkly Rep 2020; 69: 1812–6. doi:10.15585/mmwr.mm6948a3

³⁷ Macias Gil R, Marcelin JR, Zuniga-Blanco B, et al. COVID-19 pandemic: disparate health impact on the Hispanic/Latinx population in the United States. J Infect Dis 2020; 222: 1592–5: jiaa474. doi:10.1093/infdis/jiaa474

³⁸ US Census. Quick facts - Maryland. Available: https://www.census.gov/quickfacts/MD [Accessed 14 Jan 2024 ].

³⁹ US Census. Quick facts - Disctric of Columbia. Available: https://www.census.gov/quickfacts/DC [Accessed 14 Jan 2021 ].

⁴⁰ Flores LE, Verduzco-Gutierrez M, Molinares D, et al. Disparities in health care for Hispanic patients in physical medicine and rehabilitation in the United States: a narrative review. Am J Phys Med Rehabil 2020; 99: 338–47. doi:10.1097/PHM.0000000000001342

⁴¹ Shiro AG, Reeves RV. Latinos often lack access to Healthcare and have poor health outcomes. Here’s how we can change that [ Brookings ]. 2020. Available: https://www.brookings.edu/blog/how-we-rise/2020/09/25/latinos-often-lack-access-to-healthcare-and-have-poor-health-outcomes-heres-how-we-can-change-that/ [Accessed 05 Mar 2022 ].

⁴² Zaeh SE, Page KR, Berger ZD, et al. Juntos en La Pandemia de COVID-19 (together in the COVID-19 pandemic): health-care professionals and the Latinx community. Lancet Respir Med 2021; 9: 13–5. doi:10.1016/S2213-2600(20)30519-1

⁴³ The RECOVERY Collaborative Group. Dexamethasone in hospitalized patients with COVID-19. N Engl J Med 2021; 384: 693–704. doi:10.1056/NEJMoa2021436

⁴⁴ Tomazini BM, Maia IS, Cavalcanti AB, et al. Effect of dexamethasone on days alive and ventilator-free in patients with moderate or severe acute respiratory distress syndrome and COVID-19: the Codex randomized clinical trial. JAMA 2020; 324: 1307–16. doi:10.1001/jama.2020.17021

⁴⁵ Søvik S, Barrat-Due A, Kåsine T, et al. Corticosteroids and Superinfections in COVID-19 patients on invasive mechanical ventilation. J Infect 2022; 85: 57–63. doi:10.1016/j.jinf.2022.05.015

⁴⁶ Jung C, Wernly B, Fjølner J, et al. Steroid use in elderly critically ill COVID-19 patients. Eur Respir J 2021; 58: 2100979. doi:10.1183/13993003.00979-2021

⁴⁷ Beltrán-García J, Osca-Verdegal R, Pallardó FV, et al. Sepsis and Coronavirus disease 2019: common features and anti-inflammatory therapeutic approaches. Crit Care Med 2020; 48: 1841–4. doi:10.1097/CCM.0000000000004625

⁴⁸ Jones JM, Fingar KR, Miller MA, et al. Racial disparities in sepsis-related in-hospital mortality: using a broad case capture method and multivariate controls for clinical and hospital variables, 2004-2013. Crit Care Med 2017; 45: e1209–17. doi:10.1097/CCM.0000000000002699

⁴⁹ Galiatsatos P, Brigham EP, Pietri J, et al. The effect of community socioeconomic status on sepsis-attributable mortality. J Crit Care 2018; 46: 129–33. doi:10.1016/j.jcrc.2018.01.008

⁵⁰ Galiatsatos P, Follin A, Alghanim F, et al. The association between neighborhood socioeconomic disadvantage and Readmissions for patients hospitalized with sepsis. Crit Care Med 2020; 48: 808–14. doi:10.1097/CCM.0000000000004307

⁵¹ Auerbach A, O’Leary KJ, Greysen SR, et al. Hospital ward adaptation during the COVID-19 pandemic: a national survey of academic medical centers. J Hosp Med 2020; 15: 483–8. doi:10.12788/jhm.3476

⁵² Caputo ND, Strayer RJ, Levitan R. Early self-proning in awake, non-intubated patients in the emergency department: a single ED’s experience during the COVID-19 pandemic. Acad Emerg Med 2020; 27: 375–8. doi:10.1111/acem.13994

⁵³ Klaiman T, Silvestri JA, Srinivasan T, et al. Improving prone positioning for severe ARDS during the COVID-19 pandemic: an implementation mapping approach. Annals ATS 2021; 18: 300–7. doi:10.1513/AnnalsATS.202005-571OC

⁵⁴ Golden SH, Galiatsatos P, Wilson C, et al. Approaching the COVID-19 pandemic response with a health equity lens: a framework for academic health systems. Acad Med 2021; 96: 1546–52. doi:10.1097/ACM.0000000000003999