Pre-existing traits associated with Covid-19 illness severity.

IMPORTANCE: Certain individuals, when infected by SARS-CoV-2, tend to develop the more severe forms of Covid-19 illness for reasons that remain unclear.
OBJECTIVE: To determine the demographic and clinical characteristics associated with increased severity of Covid-19 infection.
DESIGN: Retrospective observational study. We curated data from the electronic health record, and used multivariable logistic regression to examine the association of pre-existing traits with a Covid-19 illness severity defined by level of required care: need for hospital admission, need for intensive care, and need for intubation.
SETTING: A large, multihospital healthcare system in Southern California. PARTICIPANTS: All patients with confirmed Covid-19 infection (N = 442).
RESULTS: Of all patients studied, 48% required hospitalization, 17% required intensive care, and 12% required intubation. In multivariable-adjusted analyses, patients requiring a higher levels of care were more likely to be older (OR 1.5 per 10 years, P<0.001), male (OR 2.0, P = 0.001), African American (OR 2.1, P = 0.011), obese (OR 2.0, P = 0.021), with diabetes mellitus (OR 1.8, P = 0.037), and with a higher comorbidity index (OR 1.8 per SD, P<0.001). Several clinical associations were more pronounced in younger compared to older patients (Pinteraction<0.05). Of all hospitalized patients, males required higher levels of care (OR 2.5, P = 0.003) irrespective of age, race, or morbidity profile. CONCLUSIONS AND RELEVANCE: In our healthcare system, greater Covid-19 illness severity is seen in patients who are older, male, African American, obese, with diabetes, and with greater overall comorbidity burden. Certain comorbidities paradoxically augment risk to a greater extent in younger patients. In hospitalized patients, male sex is the main determinant of needing more intensive care. Further investigation is needed to understand the mechanisms underlying these findings.

Introduction

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is now well recognized as the cause of the coronavirus disease 2019 (Covid-19) global pandemic [1-3]. The rate of rise in Covid-19 infection and its associated outcomes in the United States is now comparable to rates observed in other severely affected countries such as China, Italy, and Spain [4-10]. The spread of Covid-19 in the United States has been especially pronounced in the states of California, New York, Michigan, Louisiana, and Washington [11]. Consistently reported across all regions is the observation that, of all individuals who become infected with SARS-CoV-2, a majority tend to have mild or no symptoms; however, an important minority will develop predominantly respiratory disease that can lead to critical illness and death [12-15]. Multiple, reports suggest that certain demographic and clinical characteristics may predispose infected persons to more severe manifestations of Covid-19, such as older age, male sex, and pre-existing hypertension, pulmonary disease, or cardiovascular disease [4, 16–19]. Given that these traits tend to cluster among the same persons, the relative contribution of each trait to the risk for developing more severe presentations of Covid-19 illness remains unclear.

We conducted a comprehensive investigation of the pre-existing demographic and clinical correlates of Covid-19 illness severity observed among patients evaluated for Covid-19 within our multi-site healthcare system in Los Angeles, California. We deliberately focused our study on pre-existing characteristics for two main reasons: first, we recognize that patients with Covid-19 illness can present early or late in the disease course, causing many clinical features to vary at the time of initial clinical encounter; and, second, we anticipate that ongoing public health efforts can be informed and augmented by understanding which predisposing factors may render certain segments of the population at higher risk for the most morbid sequelae of SARS-CoV-2 infection.

Methods

Study sample

The Cedars-Sinai Health System is located in Los Angeles, California with a diverse catchment area of 1.8 million individuals, 33% of whom are over the age of 45 years and 80% identify as a racial or ethnic minority. The Cedars-Sinai Health System includes Cedars-Sinai Medical Center (CSMC), Marina Del Rey Hospital (MDRH), and affiliated clinics. For the current study, we included all patients who were found to have a laboratory confirmed diagnosis of SARS-CoV-2 infection while being evaluated or treated for signs or symptoms concerning for Covid-19 at CSMC or MDRH, beginning after the first confirmed case of community transmission was reported in the U.S. on February 26, 2020. Subsequently, the first laboratory confirmed Covid-19 case in our health system was on March 8, 2020. All laboratory testing for SARS-CoV-2 has been performed using reverse transcriptase polymerase chain reaction of extracted RNA from nasopharyngeal swabs. All patient testing was performed by the Los Angeles Department of Public Health until March 21, 2020, at which time the CSMC Department of Pathology and Laboratory Medicine began using the A*STAR FORTITUDE KIT 2.0 COVID-19 Real-Time RT-PCR Test (Accelerate Technologies Pte Ltd, Singapore). For the minority of patients in our study who had SARS-CoV-2 testing performed at an outside facility (3.6%), documentation of a positive test was carefully reviewed by our medical staff and considered comparable for accuracy.

Data collection

For all patients considered to have Covid-19, based on direct or documented laboratory test result and suggestive signs and/or symptoms, we obtained information from the electronic health record (EHR) and verified data for the following demographic and clinical characteristics: age at the time of diagnosis; sex; race; ethnicity; smoking status defined as current versus prior, never, or unknown; comorbidities, including obesity, as clinically assessed and documented by a provider with ICD-10 coding; and, chronic use of angiotensin converting enzyme (ACE) inhibitor or angiotensin II receptor blocker (ARB) medications. Chronic use of ACE or ARB medications was verified by confirming presence of documented ongoing medication use in an outpatient provider’s clinic note along with presence of an active outpatient prescription for the medication, both dated from prior to Covid-19 testing. We conducted iterative quality control and quality assurance analyses on all information extracted from the EHR; all data variables included in the main analyses were verified for completeness and accuracy through manual chart review, to avoid variable missingness or potential impact of inappropriate outliers in statistical modeling. Because presenting clinical measures such as vital signs and laboratory values can be highly variable, based on timing of the original clinical presentation, we elected to focus on pre-existing traits that may predispose to Covid-19 illness severity in a manner less dependent on the timing of patients presenting to medical care. To capture variation in relative comorbid status, in a way that is not captured by distinct medical history variables alone, we calculated the Elixhauser Comorbidity Index (ECI) with van Walraven weighting for all patients based on all available clinical data [20-23]. The ECI uses 31 categories to quantify a patient’s burden of comorbid conditions and has been shown to outperform other indices in predicting adverse outcomes (S1 Table) [22-28]. For patients admitted to the hospital, length of stay, admission to an intensive care unit (ICU) and death were ascertained from time stamps recorded for admission, unit transfers, and discharge. Interventions such as intubation and prone positioning were identified through time stamped orders in the EHR and verified by manual chart review. Dates and times of onset for reported or observed relevant signs and/or symptoms were also determined via manual chart review. All care was provided at the discretion of the treating physicians. Our outcomes for this study included: severe illness (defined as requiring any kind of hospital admission), critical illness (defined as the need for intensive care during hospitalization), and respiratory failure (defined as the need for intubation and mechanical ventilation). The CSMC institutional review board approved all protocols for the current study and waived the requirement for informed consent.

Statistical analyses

For the total sample of Covid-19 patients, we used parametric tests to compare normally distributed continuous variables and non-normally distributed or categorical variables, respectively. We also used histograms to display age and sex distribution for the total cohort, the patients admitted but not requiring intensive care, and patients requiring intensive care at any time during hospitalization, and the patient requiring intubation and mechanical ventilation at any time during hospitalization. We used ordinal logistic regression to examine the associations between pre-existing characteristics (based on clinically relevant, non-missing data) and a primary outcome measure of illness severity, defined as an illness severity score. We constructed the illness severity score, with higher values assigned to needing more intensive levels of clinical care, based on the following stepwise categories: 0 = clinically deemed to not require admission; 1 = required hospital admission but never required intensive care; 2 = required intensive level care but never intubation; and, 3 = required intubation during hospitalization. We constructed age- and sex-adjusted models, from which significantly associated covariates (based on P<0.20) were selected for inclusion in the final multivariable-adjusted models, where appropriate (i.e. smaller sample sizes). Race was treated as a binary covariate: African American and non-African American. This approach was selected given the recently reported concerns of excess risk for African Americans [29], along with limited understanding of whether or not comorbidities contribute to this risk, in addition to the sample size for other race groups being too small for certain comparisons. Because hypertension and diabetes are not calculated as substantial contributors to the Elixhauser comorbidity index, we included each of these traits as separate additional covariates in all multivariable-adjusted analyses. We calculated the variance inflation factor (VIF) for each of the predictor variables to confirm absence of any substantial multicollinearity. In secondary analyses, we analyzed the associations of pre-existing patient characteristics with the distinct outcomes of needing any hospital admission (severe illness) and, in the cohort of all hospitalized patients representing an especially vulnerable population, the need for intensive care (critical illness) or intubation (respiratory failure). All analyses were performed using R, version 3.5.1 (R Foundation for Statistical Computing) and Stata, version 15 (StataCorp). For all final models, P values were 2-sided and considered significant at threshold level of 0.05.

Results

Regional analyses showed that patients presented to our healthcare system from across a broad geographic catchment area in Los Angeles County (S1 Fig). The demographic and clinical characteristics of all patients in our study sample are shown in Table 1. Of all patients with pharmacologically treated hypertension, a minority were taking ACE inhibitor or ARB class agents and a majority were taking anti-hypertensive medications from alternate classes. Overall, almost half of patients (N = 214, 48%) were clinically assessed to require hospital admission, of whom over a third (N = 77; 36%) required intensive care and almost a quarter (24.3%) required intubation. In unadjusted analyses, the patients who were more likely to require higher levels of care tended to be older, male, African American, and with known hypertension, diabetes mellitus, higher Elixhauser comorbidity index, and have prior myocardial infarction or heart failure (Table 1). The number of men with confirmed Covid-19 infection outnumbered women in nearly all age groups; this sex difference was more pronounced among patients requiring hospitalization and particularly among patients requiring intensive care or intubation (Fig 1). We also observed a consistently higher rate of greater illness severity among African Americans compared to persons of other racial groups (Fig 2).

Table 1

Demographic and clinical characteristics of all patients with Covid-19.

	Total	Covid-19 Illness Severity				P value*
		Not Admitted	Admitted, Non-ICU	ICU, Non-intubated	ICU, intubated
N	442	228	137	25	52
Age, years, mean ± sd	52.72 (19.65)	43.16 (15.49)	61.66 (19.54)	64.24 (18.81)	65.56 (15.16)	<0.001
Male sex, n (%)	256 (57.9)	121 (53.1)	78 (56.9)	16 (64.0)	41 (78.8)	0.007
Smoker, n (%)	16 (5.5)	13 (8.5)	3 (3.2)	0 (0.0)	0 (0.0)	0.104
Ethnicity, n (%)						0.012
Non-Hispanic	341 (77.1)	167 (73.2)	112 (81.8)	21 (84.0)	41 (78.8)
Hispanic	68 (15.4)	33 (14.5)	22 (16.1)	4 (16.0)	9 (17.3)
Race, n (%)						<0.001
White	283 (64.0)	136 (59.6)	99 (72.3)	17 (68.0)	31 (59.6)
African American	58 (13.1)	19 (8.3)	21 (15.3)	4 (16.0)	14 (26.9)
Asian	35 (7.9)	25 (11.0)	7 (5.1)	3 (12.0)	0 (0.0)
Other	37 (8.4)	23 (10.1)	8 (5.8)	1 (4.0)	5 (9.6)
Obesity, n (%)	71 (16.1)	27 (11.8)	27 (19.7)	4 (16.0)	13 (25.0)	0.059
Hypertension, n (%)	161 (36.4)	44 (19.3)	68 (49.6)	20 (80.0)	29 (55.8)	<0.001
Diabetes mellitus, n (%)	84 (19.0)	18 (7.9)	40 (29.2)	10 (40.0)	16 (30.8)	<0.001
Elixhauser comorbidity score, mean ± sd	6.32 (10.78)	1.31 (4.35)	10.62 (12.17)	17.52 (16.54)	11.62 (12.05)	<0.001
Prior myocardial infarction or heart failure, n (%)	49 (11.1)	4 (1.8)	27 (19.7)	8 (32.0)	10 (19.2)	<0.001
Prior COPD or asthma, n (%)	70 (15.8)	27 (11.8)	28 (20.4)	6 (24.0)	9 (17.3)	0.101
ACE inhibitor use, n (%)	31 (7.0)	11 (4.8)	15 (10.9)	1 (4.0)	4 (7.7)	0.15
Angiotensin receptor blocker use, n (%)	41 (9.3)	13 (5.7)	15 (10.9)	4 (16.0)	9 (17.3)	0.026

* P values are for between-group comparisons using the ANOVA test for continuous variables and the chi-square test for categorical variables.

A total of 29 patients were missing race data and 33 patients were missing ethnicity data.

Fig 1

Age and sex distribution of patients with Covid-19, stratified by admission status.

The frequency of laboratory confirmed Covid-19 was higher in males compared to females particularly among individuals requiring hospital admission, individuals with critical illness (requiring intensive care), and individuals with respiratory failure (requiring intubation).

Fig 2

Rates of clinical outcomes of all patients with Covid-19, stratified by race.

The frequency of African Americans manifesting more severe forms of Covid-19 illness, requiring higher levels of clinical care, was greater than that for other racial groups. *Rate was calculated as proportion of cases within each racial group.

For the primary outcome of illness severity, categorized by escalating levels of care (i.e., hospitalization, intensive care, intubation), the pre-existing characteristics that demonstrated statistical significance in age- and sex-adjusted models included older age, male sex, African American race, obesity, hypertension, diabetes mellitus, and the Elixhauser comorbidity score (Table 2; Fig 3). The associations that remained significant in the fully-adjusted multivariable model included older age (odds ratio [OR] 1.49 per 10 years, 95% confidence interval [CI] 1.30–1.70, P<0.001), male sex (OR 2.01, 95% CI 1.34–3.04, P = 0.001), African American race (OR 2.13, 95% CI 1.19–3.83, P = 0.011), obesity (OR 1.95, 95% CI 1.11–3.42, P = 0.021), diabetes mellitus (OR 1.77, 95% CI 1.03–3.03, P = 0.037) and the comorbidity score (OR 1.77 per SD, 95% CI 1.37–2.28, P<0.001). We also observed a trend towards lower severity of illness among patients chronically treated with ACE inhibitor therapy, with OR 0.48 (95% CI 0.22–1.04; P = 0.06). Each estimated OR value represents the increment in higher (or lower) odds of a patient requiring a next higher level of care, for every unit difference in a continuous variable (e.g. per 10 years of age) or for presence versus absence of a given categorical variable (e.g. male sex). In effect, every 10 years of older age was associated with ~1.5-fold higher odds of requiring a higher level of care, and being male versus female was associated with a ~2-fold higher odds of requiring higher level care. We used the Brant method to test the proportional odds assumption for consistency of associations across our ordinal outcome; these analyses revealed no substantial qualitative violations, but did indicate that the Elixhauser score was predominantly associated with the specific outcomes of admission versus non-admission (OR 4.34, P<0.001) and need for intensive care versus no intensive care need (OR 1.55, P = 0.008) that with the less frequent outcome of needing intubation versus no need for intubation (OR 1.24, P = 0.25).

Table 2

Characteristics associated with overall Covid-19 illness severity* in the total sample (N = 442).

	Age- and Sex-Adjusted Models		Multivariable-Adjusted Model†
	OR (95% CI)	P value	OR (95% CI)	P value
Age, per 10 years	1.68 (1.52,1.87)	<0.001	1.49 (1.30,1.70)	<0.001
Male sex	1.87 (1.26,2.77)	0.002	2.01 (1.34,3.04)	0.001
African American race∫	2.46 (1.45,4.18)	<0.001	2.13 (1.19,3.83)	0.011
Hispanic ethnicity	1.54 (0.91,2.60)	0.11	1.39 (0.79,2.45)	0.26
Obesity	1.96 (1.19,3.24)	0.009	1.95 (1.11,3.42)	0.021
Hypertension	1.97 (1.27,3.05)	0.003	1.19 (0.71,1.99)	0.52
Diabetes mellitus	2.25 (1.41,3.57)	0.001	1.77 (1.03,3.03)	0.037
Elixhauser comorbidity score, per SD	1.63 (1.33,2.01)	<0.001	1.77 (1.37,2.28)	<0.001
Prior myocardial infarction or heart failure	1.72 (0.96,3.09)	0.07	0.56 (0.27,1.18)	0.13
Prior COPD or asthma	1.23 (0.75,2.03)	0.41	0.76 (0.44,1.31)	0.34
ACE inhibitor use	0.69 (0.35,1.38)	0.29	0.48 (0.22,1.04)	0.06
Angiotensin receptor blocker use	1.18 (0.63,2.19)	0.61	1.05 (0.54,2.06)	0.89

*The primary outcome of Covid-19 illness severity score in the total sample was defined as an ordinal variable wherein: 0 = referent, 1 = required admission but never ICU level care, 2 = required ICU level care but never intubated, 3 = required intubation.

† All listed covariates shown were included in the full multivariable-adjusted model.

∫The referent is non-African American race.

Fig 3

Characteristics associated with overall Covid-19 illness severity.

Results for the total sample of N = 442 admitted and non-admitted patients are shown in Panel A (all listed covariates shown were in the full multivariable-adjusted model). Results for the N = 214 admitted patients are shown in Panel B (to avoid model overfitting given the smaller sample size, covariates included in the multivariable model were selected from age- and sex-adjusted models based on significance with P<0.20).

For the specific outcome of needing any hospital admission, the pre-admission characteristics that demonstrated statistical significance included older age, male sex, African American race, obesity, hypertension, diabetes mellitus, the Elixhauser comorbidity index, and prior myocardial infarction or heart failure (S2 Table). In the multivariable model adjusting for all key covariates, the pre-existing traits that remained significantly associated with needing any hospital admission were older age, diabetes mellitus, and higher comorbidity index.

Among the patients whose illness severity required hospitalization, male sex was associated with the outcome of requiring further escalating levels of care (i.e., intensive care and intubation) (Table 3; Fig 3). In the multivariable model adjusting for key covariates, male sex remained the single most important risk marker of requiring higher-level care (OR 2.53, 95% CI 1.36–4.70, P = 0.003). The results for male sex were similar for the individual outcomes of requiring intensive care or intubation (S3 Table). We again observed a trend towards lower need for admission to the intensive care unit among patients chronically taking an ACE inhibitor (OR 0.38, 95% CI 0.13–1.17, P = 0.09), and greater need for intubation among African Americans patients (OR 2.14, 95% CI 0.99–4.64, P = 0.053).

Table 3

Characteristics associated with Covid-19 illness severity among all hospitalized patients.

	Age- and Sex-Adjusted Models		Multivariable-Adjusted Model†
	OR (95% CI)	P value	OR (95% CI)	P value
Age, per 10 years	1.13 (0.97,1.32)	0.12	1.14 (0.98,1.34)	0.09
Male sex	2.36 (1.29,4.34)	0.006	2.53 (1.36,4.70)	0.003
African American race	1.85 (0.92,3.71)	0.08	1.78 (0.88,3.58)	0.11
Hispanic ethnicity	1.31 (0.60,2.86)	0.49	-	-
Obesity	1.32 (0.66,2.65)	0.44	-	-
Hypertension	1.33 (0.72,2.46)	0.37	-	-
Diabetes mellitus	1.15 (0.63,2.09)	0.65	-	-
Elixhauser comorbidity score, per SD	1.04 (0.81,1.34)	0.74	-	-
Prior myocardial infarction or heart failure	0.95 (0.47,1.94)	0.89	-	-
Prior COPD or asthma	0.77 (0.38,1.54)	0.46	-	-
ACE inhibitor use	0.45 (0.15,1.34)	0.15	0.48 (0.16,1.43)	0.19
Angiotensin receptor blocker use	1.45 (0.65,3.21)	0.36	-	-

*The secondary outcome of Covid-19 illness severity score in hospitalized patients was defined as an ordinal variable wherein: 1 = referents required admission but never ICU level care, 2 = required ICU level care but never intubate, 3 = required intubation.

†To avoid model overfitting given the sample size, covariates included in the multivariable model were selected from age- and sex-adjusted models based on significance with P<0.20.

In secondary analyses, we used multiplicative interaction terms to assess for effect modification for associations observed in the main analyses (S4 Table). While considered exploratory or hypothesis generating analyses, we found several interactions of potential interest (Fig 4). In particular, the associations of Hispanic ethnicity, obesity, diabetes, and Elixhauser comorbidity index with the primary outcome appeared paradoxically more pronounced in younger compared to older individuals (S5 Table). By contrast, the primary outcome was more pronounced among older compared to younger African Americans. Also paradoxically, hypertension appeared associated with greater risk in non-obese patient and with lower risk in obese patients. We repeated all main analyses with additional adjustment for smoking status in the subset of patients with available data on smoking; in these models, all significant results remained unchanged (S6 and S7 Tables).

Fig 4

Associations with overall Covid-19 illness severity, stratified by subgroups.

Relative risks associated with illness severity score are shown for all associations observed in the total sample (N = 442), stratified by subgroups defined by age (younger vs. older than median age 52 years), sex, and obesity (BMI ≥30 kg/m2). *The primary outcome of Covid-19 illness severity score in the total sample was defined as an ordinal variable wherein: 0 = referent, 1 = required admission but never ICU level care, 2 = required ICU level care but never intubated, 3 = required intubation. **P for interaction values were calculated from likelihood ratio test between models with and without the interaction term. For each variable in the list, age (

Discussion

We examined the pre-existing characteristics associated with severity of Covid-19 illness, as observed thus far in our healthcare system located in Los Angeles, California. We found that almost half of patients presenting for evaluation and then confirmed to have Covid-19 were clinically assessed to require hospital admission. These higher risk individuals were more likely to be older, male, African American, obese, and have diabetes mellitus in addition to a greater overall burden of medical comorbidities. Notably, chronic use of an ACE inhibitor appeared related to lower illness severity, in the absence of a similar finding for ARB use. Among all individuals requiring inpatient care for Covid-19, male patients had a greater than 2.5-fold odds of needing intensive care and a 3.0-fold odds of needing intubation. All of our findings were observed even after accounting for co-existing risk factors and chronic medical conditions.

Recognizing that patients with Covid-19 illness can present with clinical features that vary based on timing of the index encounter, we sought to identify the pre-existing traits that render some individuals at highest risk for developing the more severe forms of Covid-19 illness once contracted. In our U.S. based metropolitan community, we observed that both obesity and diabetes mellitus are predisposing factors associated with a greater odds of needing hospital admission for Covid-19 but not of requiring further escalation of care; this finding is consistent with emerging reports of obesity and diabetes mellitus each being associated with a greater risk for pneumonia due to Covid-19 as well as other community-acquired viral agents—particularly in areas of the world where obesity is prevalent [30-35]. Also consistent with worldwide reports, we observed that older age is a significant predisposing risk factor for greater Covid-19 illness severity in multivariable-adjusted models; this finding may represent an age-related immune susceptibility that is not completely captured by even a comprehensive comorbidity measure such as the Elixhauser index. Notwithstanding an overall age association in the expected direction of risk, we also found a paradoxical age interaction for certain key correlates. In effect, presence of obesity, diabetes, or an elevated overall comorbidity index were each associated with greater Covid-19 illness severity in younger (i.e. <52 years) compared to older age groups. While unexpected, this finding is actually consistent with the known reduction of ACE2 expression with advancing age, a phenomenon that has been proposed as a major contributor to the broad susceptibility to Covid-19 seen in younger to middle aged individuals across the population at large [36].

Consistent with worldwide reports, we found that the association of male sex with greater odds for every metric of Covid-19 illness severity was especially prominent—and this was not explained by age variation, risk factors, or comorbidities [37]. Reasons for the male predominance of illness severity remain unclear. Although ACE2 genetic expression is on the X chromosome, evidence to date would suggest relatively comparable expression levels between sexes [38, 39], albeit with some potential for variation in relation to differences in sex hormones; select animal studies have shown increased ACE2 activity in the setting of ovariectomy and the opposite effect with orchietomy [40, 41]. While there remains scant data currently available to explain sex differences for Covid-19, male sex bias was also observed for SARS and MERS [42, 43]. Similar to the findings in our study, this increase risk was not attributable to a greater prevalence of smoking among men. Notably, prior murine studies have also demonstrated male versus female bias in susceptibility to SARS-CoV infection, which may be related to the effects of sex-specific steroids and X-linked gene activity on modulation of both the innate and adaptive immune response to viral infection [44]. Further research specific to the sexual dimorphism seen in SARS-CoV-2 susceptibility is needed.

In our U.S. based metropolitan community, we also observed racial and ethnic patterns of susceptibility to greater Covid-19 illness severity. Specifically, we found that African Americans were at greater risk for needing higher levels of care overall, and this vulnerability appeared more pronounced in older age and among men. Although an overall risk association was not seen for Hispanic ethnicity, there was a trend towards greater Covid-19 illness severity in younger aged compared to older aged Hispanic/Latino persons. A recent national report from the CDC also suggests overall higher rates of Covid-19 susceptibility in African Americans, and our findings confirm this trend exists even after adjusting for age, risk factors, and comorbidities [45]. In addition to the effects of unmeasured socioeconiomic and healthcare access variables, racial/ethnic disparities in Covid-19 illness severity may relate to yet unidentified host-viral susceptibility factors that could also be contributing to heterogeneity of community transmission seen across regions worldwide and populations at large [29].

The use of ACE inhibitor or angiotensin receptor blocker (ARB) medications has been a focus of attention given that these agents may upregulate expression of ACE2, the viral point of entry into cells [46] and alveolar type 2 epithelial cells in particular [47]. Alternatively, these agents may confer benefit, given that SARS-CoV-2 appears to reduce ACE2 activity and lead to potentially unopposed excess renin-angiotensin-aldosterone activation [36, 46, 48]. Although we observed a non-significant trend in association of chronic ACE inhibitor treatment with lower Covid-19 illness severity, we found evidence of neither risk nor benefit with ARBs. Together, our findings are supportive of current recommendations to not discontinue chronic ACE inhibitor or ARB therapy for patients with appropriate indications for these medications.

Several limitations of our study merit consideration. Our cohort included all individuals who underwent laboratory testing for Covid-19 and not individuals who did not undergo testing; thus, our study results are derived from individuals presenting with symptoms that were deemed severe enough to warrant testing. All data including past medical history data were collected from the EHR and, thus, subject to coding bias and variations in reporting quality. To minimize the potential effects of these limitations that are inherent to EHR data, we performed iterative quality checks on the dataset and conducted manual chart review to verify values for key variables. We recognize that the illness severity outcomes defined as clinically ascertained need for hospital admission, ICU level care, and intubation, may vary from practice to practice. As in many other U.S. medical centers affected by the Covid-19 pandemic, our clinical staff have been practicing under institutional guidance to conserve resources and we anticipate that the thresholds for escalating care are likely comparable; thresholds for admission, transfer to intensive care, and intubation may be different in more resource constrained environments. Given the relatively small number of observed in-hospital deaths (N = 11), and thus limited statistical power to detect associations, we deferred analyses of pre-existing characteristics and mortality risk to future investigations. The modest size of this early analysis of our growing clinical cohort may have limited our ability to detect potential additional predictors of Covid-19 illness severity, as well as potential interactions or effect modification relevant to the outcomes; thus, further investigations are needed in larger sized samples. Finally, our results are derived from a single healthcare system, albeit multi-center and serving a large catchment of the diverse population of Los Angeles, California. Additional studies are needed to examine the extent to which our findings are generalizable to other populations affected by Covid-19.

In summary, we found that among patients tested and managed for laboratory confirmed Covid-19 in our healthcare system to date, approximately half require admission for inpatient hospital care. These individuals are more likely to be older, male, African American, obese, and with known diabetes mellitus as well as a greater overall burden of medical comorbidities. Well over a third of hospitalized patients require intensive care, with a substantial proportion needing intubation and mechanical ventilation for respiratory failure. Among hospitalized patients, the highest risk individuals were more likely to be predominantly men of any age or race—for reasons not explained by comorbidities. Further investigations are needed to understand the mechanisms underlying these associations and, in turn, determine the most optimal approaches to attenuating adverse outcomes for all persons at risk.

Elixhauser comorbidity index and van Walraven weights.

(DOCX)

Click here for additional data file.

Characteristics associated with need for any hospitalization in all patients with Covid-19.

(DOCX)

Click here for additional data file.

Characteristics associated with distinct outcomes in patients hospitalized for Covid-19.

(DOCX)

Click here for additional data file.

Age, sex, and body mass index interactions with characteristics associated with overall Covid-19 illness severity in the total sample.

(DOCX)

Click here for additional data file.

Age, sex, and obesity stratified associations with overall Covid-19 illness severity in the total sample.

(DOCX)

Click here for additional data file.

Characteristics associated with overall Covid-19 illness severity* in the total sample.

(DOCX)

Click here for additional data file.

(DOCX)

Click here for additional data file.

Los Angeles county regional distribution of all patients with Covid-19.

The patients treated in our healthcare system for Covid-19 illness presented from across a diverse regional distribution of residential locations across Los Angeles County. The map shown was generated using ArcGIS software by Esri.

(TIF)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

15 May 2020

PONE-D-20-12746

Pre-Existing Traits Associated with Covid-19 Illness Severity

PLOS ONE

Dear Dr. Cheng,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Both reviewers raised several concerns especially regarding collection, analysis, and presentation of data. The authors need to be effectively respond to their comments in the revision.

We would appreciate receiving your revised manuscript by Jun 28 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Yu Ru Kou, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your ethics statement in the Methods section and in the online submission form, please provide additional information about the data used in your retrospective study. Specifically, please ensure that you have discussed whether all data were fully anonymized before you accessed them and/or whether the IRB or ethics committee waived the requirement for informed consent. If patients provided informed written consent to have data from their medical records used in research, please include this information.

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

4. Please amend the manuscript submission data (via Edit Submission) to include authors Joseph E. Ebinger, MD, MS, Natalie Achamallah, MD, MS, Hongwei Ji, MD, Brian L. Claggett, PhD, Nancy Sun, MPS, Patrick Botting, MSPH, Trevor-Trung Nguyen, BS, Eric Luong, MPH, Elizabeth H. Kim, BA, Eunice Park, BS, Yunxian Liu, MS, PhD, Ryan Rosenberry, PhD, Yuri Matusov, MD, Steven Zhao, MD, Isabel Pedraza, MD, Tanzira Zaman, MD, Michael Thompson, BS, MS, Koen Raedschelders, PhD, Anders H. Berg, MD, PhD, Jonathan D. Grein, MD, Paul W. Noble, MD, Sumeet S. Chugh, MD, C. Noel Bairey Merz, MD, Eduardo Marbán, MD, PhD, Jennifer E. Van Eyk, PhD, Scott D. Solomon, MD, Christine M. Albert, MD, MPH and Peter Chen, MD.

5. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

6. We note that Supplemental Figure 1 in your submission contain map image which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

6.1    You may seek permission from the original copyright holder of Supplemental Figure 1 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

6.2.   If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors investigated the association between patients’ preexisting traits and the coronavirus disease (COVID-19) severity as defined by the level of required care, including the needs for hospital admission, intensive care, and intubation, in patients with laboratory-confirmed COVID-19. The study was conducted in the Cedars-Sinai Health System (Cedars-Sinai Medical Center, Marina Del Rey Hospital, and affiliated clinics) in Los Angeles, California. The present study concluded that the COVID-19 severity was greater in patients who were older, male, and African American, and had obesity, diabetes, and greater overall comorbidity burden. This study evaluated an important clinical issue and was well conducted, and the writing of the manuscript is coherent and well organized. I only have a few questions and comments as follows:

1. In the Data Collection section, the authors mentioned that they obtained data on length of stay, death, and vital signs/laboratory diagnostics assessed within 1 week prior to presentation. I did not find any analytical results related to these variables (as shown in Table 1, baseline characteristics, or adjustment in the regression models). How were these data utilized in the study?

2. As this study was conducted on the basis of electronic health records, I assumed that some missing data inevitably existed. I suggest briefly summarizing how much data were missing and how the authors treated the missing data in their analyses.

3. What model was chosen for performing the ordinal logistic regression analysis? If the proportional odds model was selected, did the authors check the proportional odds assumption?

4. Was there collinearity between any of the predictor variables (e.g., ethnicity and race, which were treated as two different variables) listed in Table 1? As all the variables listed in Table 1 were included in the multivariable regression models and each OR was calculated and interpreted for each independent variable in this study, the authors should consider checking the assumption of no multi-collinearity, which is one of the assumptions when performing ordinal logistic regression analyses.

5. In Table 2, why did you calculate only the OR of the African American race? There are other races such as Asian and others (I assume the white race as reference) shown in Table 1.

6. As the authors claimed a trend toward lower illness severity among the patients chronically treated with angiotensin-converting enzyme inhibitor (ACEI) therapy, with an OR of 0.48 (P = 0.06), I suggest describing how they defined “chronically treated with drugs (ACEI and ARB)” in the Methods section, as these findings are interesting and clinically relevant.

7. On page 10, first paragraph: “…ACE inhibitor (OR 0.38, 95% CI 0.13-0.17, P=0.09)…,” I believe there was an error in the confidence interval. Please correct it. Please check the accuracy of all the statistical values presented in this manuscript.

Reviewer #2: This study investigates the association between pre-existing diseases (comorbidities) and the severity of COVID19 (n=442). The authors defined the outcome as 0, 1, 2, 3 which is an ordinal scale reflecting the severity of COVID19. Based upon the multivariate ordinal logistic regression analysis, they found older age, male, race, obese, DM and high ECI were significantly associated with the severity of COVID 19. The topic is important for the COVID 19 pandemic, but I have some concerns about the methods:

Major comment:

1. Recent studies indicate that SARS-CoV-2 might enter host cells by binding angiotensin-converting enzyme 2 (ACE2). However, the theme of this study is to investigate the risk of severity in relation to pre-existing traits rather ACE2. In this sample, the age and race differences exist among the 4 groups. If ACEI or ARB prescriptions are related to age or race, this might cause the bias. Further, the current sample size of ACEI and ARB is too few to confirm ACE2 hypothesis. I recommend the authors should focus on the theme of this study. Particularly, the total percentage of ACEI (n=31) and ARB (n=41) is 16% which is much lower than hypertension 36%. Why is it?

2. Because there are different forms of ordinal logistic regression models to take care of ordinal outcome (0= not require admission; 1=required hospitaladmission without intensive care; 2 = required intensive level care without intubation; and, 3 = required intubation during hospitalization), the authors should clearly describe which model is used and examine whether the assumption (e.g., proportional odds) holds or not.

3. The primary analysis is ordinal logistic regression (the outcome is 0,1,2,3), and the secondary analysis is logistic regression analysis (the focus is for specific outcome, a binary variable). The authors need to explain the meaning of the estimated odds ratio from each analysis. Also, since ECI depends on 31 comorbidities including DM, Hypertension and obesity, do the authors assess the collinearity between the covariates?

4. The authors need to provide the details of interaction models. Because of small sample size (n=442, particularly n=52 for patients required intubation), the power to detect 11 interaction terms might be low (in Suppl. Tables 4-5). Please explain it better.

5. In Table1: according to the Covid-19 Illness Severity outcome ( i.e., 0, 1, 2, 3), the authors should compare the sample characteristics among the 4 disjoint groups to fit their goal. Smoking status could be added although the data might not be complete. Remove “unknown” category for ethnicity and race.

Minor comments:

6. Methods: Data collection

The details of the data should be provided. For example, what is the definition of obesity (e.g., BMI>30) or smoking status (e.g., current, ever or non-smoker).

Statistical Analyses:

7. Statistical testing methods for Table 1 should be described.

8. Figure 1 is bar chart not histogram. For age groups, use thresholds e.g., 40, and 70 might be enough for grouping. For Fig 1 A-D, Y-axis should be rate (%) for fair comparison among these groups. Fig 1 C, the sample size of “Patients Needing ICU Level Care” is 52 or 77? For Fig 1E, please remove “unknown” category.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

16 Jun 2020

We appreciate the valuable feedback from the Editors and Reviewers. Herein, we detail the changes to our manuscript that we have made in response to the helpful comments and suggestions provided. We believe that the manuscript has been improved as a result.

Comments from the Editors

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Reply: We thank the Editor for this suggestion and have ensured adherence to the style requirements of PLOS ONE.

2. In your ethics statement in the Methods section and in the online submission form, please provide additional information about the data used in your retrospective study. Specifically, please ensure that you have discussed whether all data were fully anonymized before you accessed them and/or whether the IRB or ethics committee waived the requirement for informed consent. If patients provided informed written consent to have data from their medical records used in research, please include this information.

Reply: This has been clarified in the methods section to reflect that the IRB waived the requirement for informed consent.

Data Collection (Page 7): “The CSMC institutional review board approved all protocols for the current study and waived the requirement for informed consent.”

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

Reply: We appreciate this query and can clarify that even after identifiers are removed, the primary dataset includes individual patient level information on the age, sex, race, ethnicity, and date of admission to one of two named hospitals. Given that we specify admissions occurred after March 8, 2020 and over the course of the subsequent several weeks in the setting of the local COVID-19 pandemic, even removal of date of admission would allow possible individual-level identification of patients who live locally within our communities. Given that our data contain this type of potentially sensitive information, there is an ethical restriction to sharing this information publicly.

4. Please amend the manuscript submission data (via Edit Submission) to include authors Joseph E. Ebinger, MD, MS, Natalie Achamallah, MD, MS, Hongwei Ji, MD, Brian L. Claggett, PhD, Nancy Sun, MPS, Patrick Botting, MSPH, Trevor-Trung Nguyen, BS, Eric Luong, MPH, Elizabeth H. Kim, BA, Eunice Park, BS, Yunxian Liu, MS, PhD, Ryan Rosenberry, PhD, Yuri Matusov, MD, Steven Zhao, MD, Isabel Pedraza, MD, Tanzira Zaman, MD, Michael Thompson, BS, MS, Koen Raedschelders, PhD, Anders H. Berg, MD, PhD, Jonathan D. Grein, MD, Paul W. Noble, MD, Sumeet S. Chugh, MD, C. Noel Bairey Merz, MD, Eduardo Marbán, MD, PhD, Jennifer E. Van Eyk, PhD, Scott D. Solomon, MD, Christine M. Albert, MD, MPH and Peter Chen, MD.

Reply: We thank the Editor for this guidance and have now updated the submission data, as directed.

5. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

Reply: We appreciate this comment. As suggested, we have now added these results to the manuscript supplement, and specifically in Supporting Table 6 and Supporting Table 7.

6. We note that Supplemental Figure 1 in your submission contain map image which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

Reply: We appreciate this query and can confirm that the map was generated by our team using ArcGIS software by Esri and is not copy written. We have now added to the manuscript a statement regarding the use of this software to the revised Figure Legend, as shown below:

Supporting Figure 1 Legend (Page 26): “The map shown was generated using ArcGIS software by Esri.”

Comments from Reviewer #1

1. In the Data Collection section, the authors mentioned that they obtained data on length of stay, death, and vital signs/laboratory diagnostics assessed within 1 week prior to presentation. I did not find any analytical results related to these variables (as shown in Table 1, baseline characteristics, or adjustment in the regression models). How were these data utilized in the study?

Reply: We thank the Reviewer for the helpful comments and suggestions provided. We especially appreciate this important query regarding the variables that were captured during initial clinical assessment and hospital admission. We recognize, as have prior publications, that many of these clinical variables (e.g. temperature, respiratory rate, oxygen saturation, white blood cell count, interleukin-6, d-dimer, ferritin, chest x-ray findings, etc) directly reflect the severity of illness upon initial clinical presentation – and this degree of illness severity is often dependent on the time between infection onset and initial presentation to medical attention (i.e. the longer the wait before presenting for medical care, the greater the severity of illness on initial presentation). Therefore, because the presenting clinical measures (e.g. vital signs, laboratory values, imaging diagnostics) can be highly variable, we elected to focus this analysis on the pre-existing clinical characteristics that may predispose to Covid-19 illness severity in a manner that is much less dependent on timing of presentation. In addition to removing mention of these more highly variable measures from the Data Collection section, we have also added clarification regarding the main motivation behind the study design:

Data Collection (Page 6): “Because presenting clinical measures such as vital signs and laboratory values can be highly variable, based on timing of the original clinical presentation, we elected to focus on pre-existing traits that may predispose to Covid-19 illness severity in a manner less dependent on the timing of patients presenting to medical care.”

2. As this study was conducted on the basis of electronic health records, I assumed that some missing data inevitably existed. I suggest briefly summarizing how much data were missing and how the authors treated the missing data in their analyses.

Reply: We thank the Reviewer for this very important point. We also recognize the limits of data gathered from the electronic health record. For this analysis, given the importance of each prioritized variable including in each model, we completed manual chart reviews to extract and verify information for any data variables that originally appears to be missing or outliers in value. This comprehensive approach allowed us to resolve potential missingness. As helpfully suggested by the Reviewer, we have now added clarification of our approach to the revised manuscript:

Data Collection (Page 6): “We conducted iterative quality control and quality assurance analyses on all information extracted from the EHR; all data variables included in the main analyses were verified for completeness and accuracy through manual chart review, to avoid variable missingness or potential impact of inappropriate outliers in statistical modeling.”

3. What model was chosen for performing the ordinal logistic regression analysis? If the proportional odds model was selected, did the authors check the proportional odds assumption?

Reply: We appreciate this important question from the Review. As suggested, we have conducted analyses to test the proportional odds assumption. Specifically, we used the Brant test with the results displayed in the Table R1.3.a below. We observed two predictor variables that demonstrated significant deviation from the proportional odds assumption: hypertension and Elixhauser score. Thus, we further investigated these deviations by conducting sensitivity analyses wherein each component of the primary outcome (an ordinal variable) was treated as a separate binary outcome in separate logistic regression models. The three separate binary outcomes were defined as: Outcome 1 (‘admitted to floor’) = (outcome >=1) vs (outcome =0); Outcome 2 (‘admitted to ICU’) = (outcome >=2) vs (outcome <=1); and Outcome 3 (‘intubation’) = (outcome =3) vs (outcome <=2). As shown in Table R1.3.b below, sensitivity analyses revealed that hypertension is not significantly associated with any of the binary outcomes in the sub-analyses. These findings indicate that potential deviation from the proportional assumption does not substantially impact interpretable results for hypertension, which was also not associated with the primary ordinal outcome in the main analyses. For the Elixhauser score, which was positively associated with the primary ordinal outcome in the main analyses, we observed that it was also consistently positively associated with all three binary outcomes in the sensitivity analyses. Thus, interpretation of an overall positive association with the primary outcome is not threatened. However, sensitivity analyses did reveal that the Elixhauser score appears to have a stronger association with outcome 1 (admitted vs not admitted) than with outcome 3 (intubation vs no intubation). For this reason, we have now added details to the Results section to clarify this important finding:

Results (Pages 10-11): “We used the Brant method to test the proportional odds assumption for consistency of associations across our ordinal outcome; these analyses revealed no substantial qualitative violations, but did indicate that the Elixhauser score was predominantly associated with the specific outcomes of admission versus non-admission (OR 4.34, P<0.001) and need for intensive care versus no intensive care need (OR 1.55, P=0.008) that with the less frequent outcome of needing intubation versus no need for intubation (OR 1.24, P=0.25)."

Table R1.3.a. Results of the Brant test of the proportional odds assumption.

Variables Chi-square Probability

Age 5.967 0.051

Male sex 4.958 0.084

African American race 3.628 0.163

Hispanic ethnicity 0.923 0.630

Obesity 2.184 0.336

Hypertension 9.030 0.011

Diabetes mellitus 1.857 0.395

Elixhauser comorbidity score 16.537 0.000

Prior myocardial infarction or heart failure 0.024 0.988

Prior COPD or asthma 0.857 0.651

ACE inhibitor use 2.426 0.297

Angiotensin receptor blocker use 0.333 0.847

Table R1.3.b. Results of analyses treating each outcome as a binary outcome.

Outcome 1:

Admitted to Floor Outcome 2: Admitted to ICU Outcome 3: Intubation

Variables OR P value OR P value OR P value

Age 1.55 <0.001 1.37 0.001 1.59 <0.001

Male sex 1.64 0.054 2.86 0.001 4.07 <0.001

African American race 1.66 0.190 2.16 0.046 3.35 0.004

Hispanic ethnicity 1.16 0.673 1.60 0.236 1.85 0.180

Obesity 1.99 0.059 1.65 0.189 2.44 0.039

Hypertension 1.14 0.690 1.46 0.291 0.66 0.325

Diabetes mellitus 2.81 0.006 1.59 0.194 1.47 0.359

Elixhauser comorbidity score 4.34 <0.001 1.55 0.008 1.24 0.249

Prior myocardial infarction or heart failure 0.69 0.601 0.61 0.297 0.61 0.360

Prior COPD or asthma 0.80 0.539 0.60 0.194 0.48 0.128

ACE inhibitor use 0.42 0.119 0.30 0.037 0.49 0.264

Angiotensin receptor blocker use 0.85 0.718 1.11 0.813 1.17 0.749

4. Was there collinearity between any of the predictor variables (e.g., ethnicity and race, which were treated as two different variables) listed in Table 1? As all the variables listed in Table 1 were included in the multivariable regression models and each OR was calculated and interpreted for each independent variable in this study, the authors should consider checking the assumption of no multi-collinearity, which is one of the assumptions when performing ordinal logistic regression analyses.

Reply: We thank the Reviewer for this very thoughtful suggestion. We have now assessed for multicollinearity by calculating the Variance Inflation Factor (VIF) value for each of the variables, all of which were <5, suggesting no multicollinearity. We have now added report of these findings to the revised manuscript.

Statistical Analyses (Pages 7-8): “We calculated the variance inflation factor (VIF) for each of the predictor variables to confirm absence of any substantial multicollinearity.”

Table R1.4. Results of testing for collinearity between predictor variables.

VIF

Age 1.68

Male sex 1.04

African American race 1.15

Hispanic ethnicity 1.10

Obesity 1.21

Hypertension 1.83

Diabetes mellitus 1.37

Elixhauser comorbidity score 2.09

Prior myocardial infarction or heart failure 1.77

Prior COPD or asthma 1.12

ACE inhibitor use 1.18

Angiotensin receptor blocker use 1.22

5. In Table 2, why did you calculate only the OR of the African American race? There are other races such as Asian and others (I assume the white race as reference) shown in Table 1.

Reply: We appreciate this query and can confirm that we elected to compare risks for the African American race to risks for all other races combined, for two reasons: (1) recently reported concerns of excess risk for African Americans along with limited understanding of whether or not comorbidities contribute to this risk; and, (2) the sample size for certain individual race groups was too small for certain comparisons and particularly for the less common outcome categories such as intubation. We have now added more clarification regarding the approach to analyses of risk by race:

Statistical Analyses (Page 7): “Race was treated as a binary covariate: African American and non-African American. This approach was selected given the recently reported concerns of excess risk for African Americans[29], along with limited understanding of whether or not comorbidities contribute to this risk, in addition to the sample size for other race groups being too small for certain comparisons.”

Table 2, footnote (Page 11): “The referent is non-African American race.”

6. As the authors claimed a trend toward lower illness severity among the patients chronically treated with angiotensin-converting enzyme inhibitor (ACEI) therapy, with an OR of 0.48 (P = 0.06), I suggest describing how they defined “chronically treated with drugs (ACEI and ARB)” in the Methods section, as these findings are interesting and clinically relevant.

Reply: We appreciate this helpful suggestion and can confirm that chronic treatment with these therapies refers to verification of stable ongoing use of the medication. We carefully adjudicated chronic medication use based on the following two criteria: (1) presence of documented use of the medication in a provider clinic note, and (2) active prescription for standing use of the medication in the medication ordering system. We have now added details regarding our approach to adjudicating medication use in the methods section:

Data Collection (Page 6): “Chronic use of ACE or ARB medications was verified by confirming presence of documented ongoing medication use in an outpatient provider’s clinic note along with presence of an active outpatient prescription for the medication, both dated from prior to Covid-19 testing.”

7. On page 10, first paragraph: “…ACE inhibitor (OR 0.38, 95% CI 0.13-0.17, P=0.09)…,” I believe there was an error in the confidence interval. Please correct it. Please check the accuracy of all the statistical values presented in this manuscript.

Reply: We thank the Reviewer for identifying this typographical error. Indeed, as the Reviewer astutely observed, the correct confidence interval values for this estimate should be 0.13 and 1.17. We have now made these corrections to the revised manuscript (Page 12). In addition, we have carefully rechecked all values in the manuscript against the original statistical programming output from R.

Comments from Reviewer #2

1. Recent studies indicate that SARS-CoV-2 might enter host cells by binding angiotensin-converting enzyme 2 (ACE2). However, the theme of this study is to investigate the risk of severity in relation to pre-existing traits rather ACE2. In this sample, the age and race differences exist among the 4 groups. If ACEI or ARB prescriptions are related to age or race, this might cause the bias. Further, the current sample size of ACEI and ARB is too few to confirm ACE2 hypothesis. I recommend the authors should focus on the theme of this study. Particularly, the total percentage of ACEI (n=31) and ARB (n=41) is 16% which is much lower than hypertension 36%. Why is it?

Reply: We thank the Reviewer for the helpful comments and suggestions provided. We agree that the results are certainly not definitive regarding the potential associations of ACEI or ARB treatment with illness severity. Thus, we completely agree with the Reviewer that our findings should be considered hypothesis generating. As helpfully advised by the Reviewer, we have now substantially shortened the discussion of these findings and have re-emphasized the need for additional studies for further investigation. The Reviewer also raises an important question regarding the prevalence of ACEI and ARB medication therapy in the context of prevalent hypertension in our study sample. We can verify that the majority of patients with hypertension who were taking anti-hypertensive medications that were not from the ACEI or ARB classes were being prescribed medications from alternate classes, the most common being diuretics, beta-blockers, and calcium channel blockers. We have now added clarification on the distribution of anti-hypertensive medication use to the revised manuscript:

Discussion (Pages 16-17): “The use of ACE inhibitor or angiotensin receptor blocker (ARB) medications has been a focus of attention given that these agents may upregulate expression of ACE2, the viral point of entry into cells[46] and alveolar type 2 epithelial cells in particular[47]. Alternatively, these agents may confer benefit, given that SARS-CoV-2 appears to reduce ACE2 activity and lead to potentially unopposed excess renin-angiotensin-aldosterone activation[36, 46, 48]. Although we observed a non-significant trend in association of chronic ACE inhibitor treatment with lower Covid-19 illness severity, we found evidence of neither risk nor benefit with ARBs. Together, our findings are supportive of current recommendations to not discontinue chronic ACE inhibitor or ARB therapy for patients with appropriate indications for these medications.”

Results (Page 8): “Of all patients with pharmacologically treated hypertension, a minority were taking ACE inhibitor or ARB class agents and a majority were taking anti-hypertensive medications from alternate classes.”

2. Because there are different forms of ordinal logistic regression models to take care of ordinal outcome (0= not require admission; 1=required hospital admission without intensive care; 2 = required intensive level care without intubation; and, 3 = required intubation during hospitalization), the authors should clearly describe which model is used and examine whether the assumption (e.g., proportional odds) holds or not.

Reply: We appreciate this very thoughtful and important point, which was also raised by Reviewer 1. As shown in Table R1.3.a above, we used the Brant method to test the proportional odds assumption. Because hypertension and the Elixhauser score demonstrated significant deviation from the proportional odds assumption, we conducted sensitivity analyses wherein each component of the primary outcome (an ordinal variable) was treated as a separate binary outcome in separate logistic regression models: Outcome 1 (‘admitted to floor’) = (outcome >=1) vs (outcome =0); Outcome 2 (‘admitted to ICU’) = (outcome >=2) vs (outcome <=1); and Outcome 3 (‘intubation’) = (outcome =3) vs (outcome <=2). As shown in Table R1.3.b above, hypertension was not significantly associated with any of the binary outcomes in the sub-analyses – indicating that potential deviation from the proportional assumption does not substantially impact interpretable results for hypertension, which was also not associated with the primary outcome in the main analyses. For the Elixhauser score, which was positively associated with the primary ordinal outcome in the main analyses, we observed that it was also consistently positively associated with all three binary outcomes in the sensitivity analyses. Thus, interpretation of an overall positive association with the primary outcome is not threatened. However, sensitivity analyses did reveal that the Elixhauser score appears to have a stronger association with outcome 1 (admitted vs not admitted) than with outcome 3 (intubation vs no intubation). For this reason, we have now added details to the Results section to clarify this important finding:

Results (Pages 10-11): “We used the Brant method to test the proportional odds assumption for consistency of associations across our ordinal outcome; these analyses revealed no substantial qualitative violations, but did indicate that the Elixhauser score was predominantly associated with the specific outcomes of admission versus non-admission (OR 4.34, P<0.001) and need for intensive care versus no intensive care need (OR 1.55, P=0.008) that with the less frequent outcome of needing intubation versus no need for intubation (OR 1.24, P=0.25)."

3. The primary analysis is ordinal logistic regression (the outcome is 0,1,2,3), and the secondary analysis is logistic regression analysis (the focus is for specific outcome, a binary variable). The authors need to explain the meaning of the estimated odds ratio from each analysis. Also, since ECI depends on 31 comorbidities including DM, Hypertension and obesity, do the authors assess the collinearity between the covariates?

Reply: We thank for the Reviewer for these astute comments. As suggested, we have now added a clearer explanation of the interpretation of estimated odds ratios for each of the main analyses:

Results (Page 10): “Each estimated OR value represents the increment in higher (or lower) odds of a patient requiring a next higher level of care, for every unit difference in a continuous variable (e.g. per 10 years of age) or for presence versus absence of a given categorical variable (e.g. male sex). In effect, every 10 years of older age was associated with ~1.5-fold higher odds of requiring a higher level of care, and being male versus female was associated with a ~2-fold higher odds of requiring higher level care.”

We also appreciate the important point regarding the ECI and possible collinearity between covariates. We had carefully considered the constituents of the EIC, which is calculated based on weights with each component comorbidity being assigned a weight ranging from -7 to 11 according to the van Walraven algorithm. Given that both diabetes and hypertension have weights of 0 in the ECI calculation, we elected to include these covariates separately in the main models. We have now added clarification regarding this approach to the revised manuscript:

Statistical Analyses (Pages 7-8): “Because hypertension and diabetes are not calculated as substantial contributors to the Elixhauser comorbidity index, we included each of these traits as separate additional covariates in all multivariable-adjusted analyses. We calculated the variance inflation factor (VIF) for each of the predictor variables to confirm absence of any substantial multicollinearity.”

4. The authors need to provide the details of interaction models. Because of small sample size (n=442, particularly n=52 for patients required intubation), the power to detect 11 interaction terms might be low (in Suppl. Tables 4-5). Please explain it better.

Reply: We appreciate this comment and completely agree with the Reviewer that statistical power to detect potential associations in stratified analyses is often limited when the sample size is modest and the outcome event rate is infrequent. For this reason, we elected to use the ordinal multi-level outcome as the primary outcome, rather than a single infrequent outcome event alone, to maximize statistical power. Although we recognize that addition of an interaction term added to the main model is a relatively statistically stable approach, we agree that subsequent models stratifying results by subgroup are more limited in power and thus should be considered hypothesis generating. As suggested, we have now added more details to the revised manuscript to clarify these important points.

Results (Page 13): “In secondary analyses, we used multiplicative interaction terms to assess for effect modification for associations observed in the main analyses (S4 Table). While considered exploratory or hypothesis generating analyses, we found several interactions of potential interest (Fig 4).”

Discussion (Page 17): “The modest size of this early analysis of our growing clinical cohort may have limited our ability to detect potential additional predictors of Covid-19 illness severity, as well as potential interactions or effect modification relevant to the outcomes; thus, further investigations are needed in larger sized samples.”

5. In Table1: according to the Covid-19 Illness Severity outcome ( i.e., 0, 1, 2, 3), the authors should compare the sample characteristics among the 4 disjoint groups to fit their goal. Smoking status could be added although the data might not be complete. Remove “unknown” category for ethnicity and race.

Reply: We thank the Reviewer for these helpful suggestions. We have now made all of these recommended revisions to Table 1.

6. The details of the data should be provided. For example, what is the definition of obesity (e.g., BMI>30) or smoking status (e.g., current, ever or non-smoker).

Reply: We thank the Reviewer for this helpful suggestion. As advised, we have now added to the revised manuscript details regarding definitions of key covariates.

Methods (Pages 5-6): “For all patients considered to have Covid-19, based on direct or documented laboratory test result and suggestive signs and/or symptoms, we obtained information from the electronic health record (EHR) and verified data for the following demographic and clinical characteristics: age at the time of diagnosis; sex; race; ethnicity; smoking status defined as current versus prior, never, or unknown; comorbidities, including obesity, as clinically assessed and documented by a provider with ICD-10 coding; and, chronic use of angiotensin converting enzyme (ACE) inhibitor or angiotensin II receptor blocker (ARB) medications. Chronic use of ACE or ARB medications was verified by confirming presence of documented ongoing medication use in an outpatient provider’s clinic note along with presence of an active outpatient prescription for the medication, both dated from prior to Covid-19 testing.”

7. Statistical testing methods for Table 1 should be described.

Reply: We appreciate this suggestion and have now added details regarding the statistical test results displayed in Table 1:

Table 1, footnote (Page 9): “P values are for between-group comparisons using the ANOVA test for continuous variables and the chi-square test for categorical variables.”

8. Figure 1 is bar chart not histogram. For age groups, use thresholds e.g., 40, and 70 might be enough for grouping. For Fig 1 A-D, Y-axis should be rate (%) for fair comparison among these groups. Fig 1 C, the sample size of “Patients Needing ICU Level Care” is 52 or 77? For Fig 1E, please remove “unknown” category.

Reply: We appreciate these helpful suggestions. For the age groups definitions, we evaluated the display using less refined cutpoints and found the result to more sparsely display the same information (as shown in Figure R2.8. below); thus, we have elected to retain the original version and would be happy to include the additional figure in the Supplement if requested by the Editors and Reviewer. For Figure 1, we have adjusted the Y axis labels as suggested. We have also added clarification that the sample size for patients needing ICU level care is N=77. In addition, we have removed the “unknown” category as suggested. Thank you again for these helpful suggestions.

Figure R2.8. Age and Sex Distribution of All Patients with Covid-19, Stratified by Admission Status

Submitted filename: Comments for PLoS Paper 6-3-2020_sc_JE_mh.docx

Click here for additional data file.

6 Jul 2020

Pre-existing traits associated with covid-19 illness severity

PONE-D-20-12746R1

Dear Dr. Cheng,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Yu Ru Kou, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for inviting me to review this revised manuscript. The authors have well addressed my previous questions and comments.

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

16 Jul 2020

PONE-D-20-12746R1

Pre-existing traits associated with covid-19 illness severity

Dear Dr. Cheng:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Yu Ru Kou

Academic Editor

PLOS ONE