key: cord-262022-kvezhyt5
authors: Kim, L.; Garg, S.; O'Halloran, A.; Whitaker, M.; Pham, H.; Anderson, E. J.; Armistead, I.; Bennett, N. M.; Billing, L.; Como-Sabetti, K.; Hill, M.; Kim, S.; Monroe, M. L.; Muse, A.; Reingold, A.; Schaffner, W.; Sutton, M.; Talbot, H. K.; Torres, S. M.; Yousey-Hindes, K.; Holstein, R. A.; Cummings, C.; Brammer, L.; Hall, A.; Fry, A.; Langley, G. E.
title: Interim Analysis of Risk Factors for Severe Outcomes among a Cohort of Hospitalized Adults Identified through the U.S. Coronavirus Disease 2019 (COVID-19)-Associated Hospitalization Surveillance Network (COVID-NET)
date: 2020-05-22
journal: nan
DOI: 10.1101/2020.05.18.20103390
sha: 
doc_id: 262022
cord_uid: kvezhyt5

Background: As of May 15, 2020, the United States has reported the greatest number of coronavirus disease 2019 (COVID-19) cases and deaths globally. Objective: To describe risk factors for severe outcomes among adults hospitalized with COVID-19. Design: Cohort study of patients identified through the Coronavirus Disease 2019-Associated Hospitalization Surveillance Network. Setting: 154 acute care hospitals in 74 counties in 13 states. Patients: 2491 patients hospitalized with laboratory-confirmed COVID-19 during March 1-May 2, 2020. Measurements: Age, sex, race/ethnicity, and underlying medical conditions. Results: Ninety-two percent of patients had at least 1 underlying condition; 32% required intensive care unit (ICU) admission; 19% invasive mechanical ventilation; 15% vasopressors; and 17% died during hospitalization. Independent factors associated with ICU admission included ages 50-64, 65-74, 75-84 and 85+ years versus 18-39 years (adjusted risk ratio (aRR) 1.53, 1.65, 1.84 and 1.43, respectively); male sex (aRR 1.34); obesity (aRR 1.31); immunosuppression (aRR 1.29); and diabetes (aRR 1.13). Independent factors associated with in-hospital mortality included ages 50-64, 65-74, 75-84 and 85+ years versus 18-39 years (aRR 3.11, 5.77, 7.67 and 10.98, respectively); male sex (aRR 1.30); immunosuppression (aRR 1.39); renal disease (aRR 1.33); chronic lung disease (aRR 1.31); cardiovascular disease (aRR 1.28); neurologic disorders (aRR 1.25); and diabetes (aRR 1.19). Race/ethnicity was not associated with either ICU admission or death. Limitation: Data were limited to patients who were discharged or died in-hospital and had complete chart abstractions; patients who were still hospitalized or did not have accessible medical records were excluded. Conclusion: In-hospital mortality for COVID-19 increased markedly with increasing age. These data help to characterize persons at highest risk for severe COVID-19-associated outcomes and define target groups for prevention and treatment strategies.

In December 2019, an outbreak of a novel coronavirus disease, termed coronavirus disease-80 2019 , was reported in China caused by a newly identified coronavirus, severe acute 81 respiratory syndrome coronavirus-2 (SARS-CoV-2). Since then, approximately 4.5 million cases of 82 COVID-19 have been reported globally (1). As of May 15, approximately 1.4 million cases, including 83 nearly 86,000 deaths, have been reported in the United States, and case counts continue to rise (1) with 84 evidence of widespread community transmission (2). 85

Previous reports from China, Italy, and New York City have demonstrated that hospitalized 86 patients are generally older and have underlying medical conditions, such as hypertension and diabetes 87 (3) (4) (5) . These studies have also found that older patients and those with certain underlying medical 88 conditions like diabetes were at higher risk for severe outcomes (3, 6, 7). Among cases reported to the 89 U.S. Centers for Disease Control and Prevention (CDC) from local and state health departments, the 90 prevalence of underlying medical conditions increased as severity of infections increased (8, 9) , although 91 findings were limited by missing or incomplete information. Questions remain about the independent 92 association of sex, race/ethnicity and specific underlying conditions with severe outcomes among 93 persons hospitalized with COVID-19, after adjusting for age and other important potential confounders. 94

Comprehensive data on U.S. patients with severe COVID-19 infections are needed to better 95 inform clinicians' understanding of groups at risk for poor outcomes and to inform current prevention 96 efforts and future interventions. We rapidly implemented population-based surveillance for laboratory-97 confirmed COVID-19-associated hospitalizations, collecting clinical data from hospitalized patients in 154 98 hospitals in 13 states since March 1, 2020. In this interim analysis restricted to patients who were 99 and/or review of hospital discharge records. Laboratory tests were ordered at the discretion of the 126 treating healthcare provider. 127

Medical chart reviews for demographic and clinical data were conducted by trained surveillance 128 officers using a standard case report form. Underlying medical conditions were categorized into major 129 groups (Appendix Table 1 ). Obesity and severe obesity were defined as a calculated body mass index 130 (BMI) ≥30 kg/m 2 and BMI ≥40 kg/m 2 , respectively. Chest radiograph results were obtained from the 131 radiology reports and not from review of the original radiograph. We defined severe outcomes as either 132 ICU admission or in-hospital mortality. We hypothesized that increasing age and underlying medical 133 conditions would be associated with an increased risk of ICU admission and in-hospital mortality. 134 135

After the exclusions noted above, we included adults hospitalized within 154 acute care 137 hospitals in 74 counties in 13 states with an admission date during March 1-May 2, 2020 who had either 138 been discharged from the hospital or died during hospitalization and had complete medical chart 139 abstractions. We calculated proportions using the number of patients with data available on each 140 characteristic as the denominator. 141

To construct multivariable models for ICU admission and in-hospital mortality, we first assessed 142 collinearity among underlying medical condition categories and outpatient use of ACE-inhibitors and 143

ARBs. We examined the association of demographic factors, underlying medical conditions, and 144 outpatient use of ACE-inhibitors and ARBs with ICU admission and in-hospital death using chi square 145 tests. Variables considered for inclusion in the final models included current or former smoker, 146 hypertension, obesity, diabetes, chronic lung disease (CLD), cardiovascular disease (CVD) (excluding 147 hypertension), neurologic disorders, renal disease, immunosuppression, gastrointestinal/liver disease, 148 hematologic conditions, rheumatologic/autoimmune conditions, and outpatient use of ACE-inhibitors or 149 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 ARBs. All multivariable models included age categorized into the following groups (18-39, 40-49, 50-150 64, 65-74, 75-84, and ≥85 years) , sex, and race/ethnicity. Other variables with p-values <0.10 in 151 bivariate analyses were included in the multivariable analyses. Log-linked Poisson generalized 152 estimating equations regression with an exchangeable correlation matrix (11, 12) , clustered by site, was 153 used to generate adjusted risk ratios (aRR), 95% confidence intervals (CI), and two-sided p-values for the 154 risk of ICU admission and in-hospital death. We also constructed separate multivariable models to 155 examine the association between the number of underlying medical conditions and ICU admission or in-156 hospital death. Two-sided p-values <0.05 were considered statistically significant. All analyses were 157 performed using the SAS 9.4 software (SAS Institute Inc., Cary, NC, USA). 158

These data were collected as part of routine public health surveillance and determined to be 159 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 from 43% (154/357) of the acute care hospitals included in COVID-NET surveillance across the 13 174 surveillance sites (Appendix Table 2 ). The percentage of facilities contributing data out of the total 175 number of facilities by site ranged from 19% to 100%. The median age of included and excluded 176 patients (62 vs. 63 years, respectively) was similar (Appendix Table 3 ). The highest proportion of 177 patients included in this analysis were from Minnesota (20%), Tennessee (20%), New York (12%), and 178

Maryland (10%), and Connecticut (9%) (Appendix Table 3 (Figure 2A ). Prevalence of CLD, neurologic conditions, obesity, and renal 192 disease varied between males and females (p<0.05, Figure 2B ). CVD, CLD, and neurologic conditions 193 were more prevalent among non-Hispanic whites, while diabetes, hypertension, obesity and renal 194 disease were more common among non-Hispanic blacks (p<0.05, Figure 2C ). 195

Cough (75%), fever or chills (74%), and shortness of breath (70%) were commonly 196 documented symptoms at admission (Table 2 and Appendix Table 4 ). Gastrointestinal symptoms, 197 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 including nausea, vomiting, and diarrhea, were documented in almost 30% of patients. Median length 198 of hospitalization was 6 days (IQR, 3-11), and median days from symptom onset to hospital admission 199 was 6 days (IQR, (3) (4) (5) (6) (7) (8) . Median values of initial vital signs were within normal range, except for elevated 200 blood pressure (Table 2) . Thirty-three individuals had a pathogen detected from positive blood cultures 201 (Appendix Table 5 ). Viral co-detections from respiratory specimens were rare among those who were 202 tested (n=38/1549, 2.5%) (Appendix Table 6 ). Among 1932 patients with chest radiograph 203 performed, 92% (n=1769) were documented as abnormal with infiltrate or consolidation (n=1574/1932, 204 81%) documented most frequently (Appendix Table 7 ). Ninety-five percent (n=540/566) of patients with 205 chest computerized tomography (CT) had abnormal findings, and ground glass opacity 206 was documented in 62% (n=350/566) (Appendix Table 7 ). 207

Forty-five percent (n=1125/2482) of patients received investigational medication regimens for 208 COVID-19 during hospitalization ( Table 2 ). The most common regimens included hydroxychloroquine 209 (n=1065/2479, 43%) and the combination of azithromycin and ≥1 COVID-19 treatment (n=725/2479, 210 29%) (non-mutually exclusive categories). The most frequent discharge diagnoses recorded in hospital 211 discharge summaries were pneumonia (n=1395/2485, 56%), acute respiratory failure (n=999/2487, 212 40%), acute renal failure (n=456/2,485, 18%), and sepsis (n=443/2,479, 18%). 213

Thirty-two percent (n=798/2490) of patients required ICU admission, with a median length of 214 ICU stay of 6 days (range, 1-41; IQR, 2-11) ( Table 2 ). Median days from symptom onset to ICU 215 admission was 7 days (range, 0-25; IQR, 4-10), and median days from hospital admission to ICU 216 admission was 1 day (range, 0-19; IQR, 0-2). Among 2,489 hospitalized patients, the highest respiratory 217 support received was invasive mechanical ventilation in 19% (n=462), bilevel positive airway pressure 218 (BIPAP) or continuous positive airway pressure (CPAP) in 3% (n=82), and high flow nasal cannula (HFNC) 219 in 7% (n=170). Fifty-three percent (n=246/462) of patients that received invasive mechanical ventilation 220 died in-hospital (median age, 71 years; IQR, 62-79); the proportion of patients receiving mechanical 221 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint ventilation who died increased with age (p<0.0001). Vasopressors were used in 15% (n=373/2486) of 222 patients, while renal replacement therapy was used in 5% (n=115/2487). As age increased, so did 223 the proportion of patients who required ICU admission, invasive mechanical ventilation, and 224 vasopressors (p<0.05, Figure 3A ). Males were admitted to the ICU and treated with invasive mechanical 225 ventilation, HFNC, or vasopressors more frequently than females (p<0.05) ( Figure 3B ). Non-Hispanic 226 whites more frequently received BIPAP, CPAP or HFNC (p<0.05, Figure 3C ). 227

Overall, seventeen percent (n=420/2490) of patients died during hospitalization ( Table 2 ). The 228

proportion of patients who died increased with increasing age groups, ranging from 3% among 18-49 229 years to 10% among 50-64 years to 29% among ≥65 years ( Figure 3A ). Males died more frequently 230 compared to females (p<0.05) ( Figure 3B ), as did non-Hispanic whites compared to other 231 race/ethnicities (p<0.05, Figure 3C ). 232

Among 420 patients who died, median age was 76 years (range, 24-97; IQR, 66-85); 58% 233 (n=244) were male; 71% (n=299) were admitted to the ICU; and 59% (n=246) received invasive 234 mechanical ventilation. The median length of hospitalization among patients who died was 7 days 235 (range, 0-40; IQR, 4-12). 236 237

Factors independently associated with ICU admission included age 50-64 years (adjusted risk 239 ratio (aRR) = 1.53; 95% confidence interval (CI), 1.28 to 1.83); 65-74 years (aRR = 1.65; CI, 1.34 to 2.03); 240 75-84 years (aRR = 1.84; CI, 1.60 to 2.11); ≥85 years (aRR = 1.43; CI, 1.00 to 2.04); male sex (aRR = 1.34; 241 CI, 1.20 to 1.50); obesity (aRR = 1.31; CI, 1.16 to 1.47); diabetes (aRR = 1.13; CI, 1.03 to 1.24); and 242 immunosuppression (aRR = 1.29; CI, 1.13 to 1.47) (Table 3A) . 243

Independent factors associated with in-hospital mortality included age 50-64 years (aRR = 3.11; 244 CI 1.50 to 6.46); age 65-74 years (aRR = 5.77; CI, 2.64 to 12.64); age 75-84 years (aRR = 7.67; CI, 3.35 to 245 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint 17.59); age ≥85 years (aRR = 10.98; CI, 5.09 to 23.69); male sex (aRR = 1.30; CI, 1.14 to 1.49); diabetes 246 (aRR = 1.19; CI, 1.01 to 1.40); CLD (aRR = 1.31; CI, 1.13 to 1.52); CVD (aRR = 1.28; CI, 1.03 to 1.58); 247 neurologic disorders (aRR = 1.25; CI, 1.04 to 1.50); renal disease (aRR = 1.33; CI, 1.10 to 1.61); and 248 immunosuppression (aRR = 1.39; CI, 1.13 to 1.70) (Table 3B) . 249

Having ≥3 underlying medical conditions was significantly associated with higher risk of 250 ICU admission and death after adjusting for age group, sex, and race/ethnicity (Appendix Table 8 ). 251 252 DISCUSSION 253

Using a geographically diverse, multi-site, population-based U.S. surveillance system, we found 254 that among adults hospitalized with laboratory-confirmed COVID-19, almost one-third required ICU 255 admission, 19% received invasive mechanical ventilation, and 17% died during hospitalization. About 256 75% of patients were ≥50 years, and >90% had underlying medical conditions. Older age, being male, 257 and the presence of certain underlying medical conditions were associated with a higher risk of ICU 258 admission and in-hospital mortality. Race/ethnicity was not independently associated with either 259 outcome among hospitalized patients. This information can alert healthcare providers to patients at 260 greatest risk of severe outcomes and help target prevention strategies and future interventions. 261

In a published COVID-NET analysis, we found that when comparing the racial/ethnic distribution 262 of residents of the surveillance catchment areas to the racial/ethnic distribution of COVID-19-associated 263 hospitalizations, non-Hispanic blacks were disproportionately hospitalized with COVID-19 compared to 264 non-Hispanic whites (10). In this analysis, however, we found that once hospitalized, non-Hispanic 265 blacks did not have increased risk of poorer outcomes compared to other race/ethnicities after adjusting 266 for age and underlying conditions. In a preprinted article of U.S. Veterans seeking care at VA Hospitals, 267

Rentsch et al. found no association between black race and ICU admission (13). Similarly, a large study 268 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint of patients hospitalized in New York City did not find race/ethnicity to be associated with ICU admission 269 or death (4). 270 COVID-19-associated hospitalizations, ICU admissions, and deaths have been shown to occur 271 more frequently with increasing age (6, 9, 14) . In our study, age ≥65 years was the strongest 272 independent predictor of ICU admission and in-hospital mortality. Persons aged 75-84 years had the 273 highest the risk of ICU admission compared to 18-49 years old, and those ≥85 years experienced 11 274 times the risk of death. These findings are similar to other studies from China, Europe, and the United 275

States (4, 9, (14) (15) (16) (17) . Our data provide support that older persons are particularly vulnerable to severe 276 COVID-19 disease and should be targeted for aggressive preventive measures (8). 277 Being male was associated with a higher risk of ICU admission and death after adjusting for age, 278 race/ethnicity and underlying conditions. Other studies have similarly shown male sex to be associated 279 with COVID-19-associated hospitalizations (4, 18), ICU admissions (19) This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint neurologic disease, renal disease and immunosuppression associated with in-hospital death, and 293 diabetes, obesity, and immunosuppression associated with ICU admission. While hypertension was 294 highly prevalent in our patient population, it was not associated with ICU admission or death. Additional 295 studies are needed to determine whether hypertension, which is also highly prevalent in the U.S. 296 population, increases the risk of COVID-19-associated hospitalizations and whether the duration of 297 hypertension and the degree to which it is controlled impact the risk for severe Similarly, the associations between the duration and degree of glycemic control in diabetes and severity 299 of COVID-19 disease require further investigation. Obesity, which was also highly prevalent in this 300 cohort, imparted increased risk for ICU admission, but not death. This finding may, in part, be explained 301 by a trend of decreasing obesity prevalence with increasing age, which was a strong risk factor for 302 mortality. Healthcare providers should be aware of these findings to appropriately triage and manage 303 patients with high-risk conditions that may either increase risk for hospitalization or poorer outcomes 304 once hospitalized (22, 23). 305

We collected data on initial symptoms, vital signs and laboratory values to characterize disease 306 severity at admission. While approximately 70% of patients had shortness of breath at admission, the 307 median oxygen saturation at admission was 94% on room air. Other admission vital signs and laboratory 308 values were also largely within normal ranges. Because we did not collect data on vital signs or 309 laboratory values during the hospital course, we may not have fully captured the onset of clinical 310 deterioration that has been reported during the second week after illness onset (24). We limited our 311 analysis to patients that had either been discharged or died in-hospital and found that 15% of patients 312 received vasopressor support, and 19% received invasive mechanical ventilation. Other U.S. studies 313 have found that up to 32% of hospitalized patients have received vasopressors and 29-33% have 314 received invasive mechanical ventilation (19, 20), though some of these studies included patients who 315

were still hospitalized at the time of analysis. In our study, 53% of patients requiring mechanical 316 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint ventilation died, which is higher than the 36% reported in a recent study from New York City (4). In 317 general, the proportion of patients with severe outcomes was higher in the United States than reported 318 from China (6). Differences between patients' outcomes in the United States and China may reflect 319 differences in clinical practices or varying thresholds for hospitalization (18). These proportions of 320 severe outcomes among U.S. patients are also generally higher than those found in U.S. adults 321 hospitalized with seasonal influenza (25, 26). Our findings may help to inform resource planning and 322 allocation in healthcare facilities during the COVID-19 pandemic. 323

There are several limitations to our analysis. First, it is likely that not all COVID-19-associated 324 hospitalizations were captured because of the lack of widespread testing capability during the study 325 period and because identification of COVID-19 patients was largely reliant on clinician-directed testing. 326

Second, clinical practices and availability of specific interventions may differ across hospitals, which 327 might have influenced findings. Third, COVID-NET is an ongoing surveillance system, and only 15% of 328 the 16,318 COVID-19 hospitalized patients were included, representing those who were discharged or 329 died in-hospital during March 1-May 2, 2020 and for whom medical records were available and chart 330 abstractions were completed. These restrictions may have resulted in selection bias. However, there 331 was no difference in the age and sex distribution between cases included and excluded from the 332 analysis. The geographic distribution of cases included versus excluded from this analysis differed, 333 which may have impacted the racial and ethnic distribution of cases included in this analysis as 334 compared to the racial and ethnic distribution of the surveillance catchment population; however, as we 335

do not yet have complete data on race/ethnicity for all identified cases, we were not able to assess this 336 further. Nevertheless, COVID-NET encompasses a large geographic area with multiple hospitals and 337 likely offers a more racially and ethnically diverse patient population compared to other single-center or 338 state-based studies. Lastly, small counts limited our ability to determine risk factors for severe 339 outcomes among all racial and ethnic groups. COVID-NET data will become more robust as additional 340 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 medical chart reviews are completed and may allow further investigation within these racial and ethnic 341 groups over time. 342

Based on preliminary findings from this multi-site, geographically diverse study, a high 343 proportion of patients hospitalized with COVID-19 received aggressive interventions and had poor 344 outcomes. Increasing age was the strongest predictor of in-hospital mortality. Prevention strategies, 345 such as social distancing and rigorous hand hygiene, are key to minimizing the risk of infection in high-346 risk patients. These data help to characterize persons at highest risk for severe COVID-19-associated 347 disease in the United States and to define target groups for future prevention and treatment strategies 348 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint Excluded 1 site's cases (no clinical data available) n=74

Excluded cases due to incomplete medical chart reviews n=13,652

for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint *p-value <0.05 CVD = Cardiovascular disease (excluding hypertension); HTN = hypertension; CLD = chronic lung disease. * The underlying medical condition categories are not mutually exclusive. Patients can have more than once underlying medical condition. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020. . https://doi.org/10.1101/2020.05.18.20103390 doi: medRxiv preprint *p-value <0.05 ICU = intensive care unit; BIPAP = bilevel positive airway pressure; CPAP = continuous positive airway pressure; HFNC = high flow nasal cannula; RRT = renal replacement therapy * For mechanical ventilation, BIPAP/CPAP, and HFNC, patients are assigned based on the highest level of respiratory support required during hospitalization (i.e. invasive mechanical ventilation followed by BIPAP or CPAP, followed by high flow nasal cannula). This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 22, 2020 . . https://doi.org/10.1101 /2020 

Comorbidity and its impact on 1590 383 patients with Covid-19 in China: A Nationwide Analysis

Factors associated with 385 hospitalization and critical illness among 4,103 patients with COVID-19 disease

COVID-19 Surveillance Group. Characteristics of COVID-19 patients dying in Italy: report based 388 on available data on

Clinical Characteristics of Coronavirus 390 Disease 2019 in China

Covid-19 in Critically Ill 392 Patients in the Seattle Region -Case Series

Among Patients with Coronavirus Disease 2019 -United States

Severe Outcomes Among Patients with Coronavirus Disease

United States

Hospitalization Rates 399 and Characteristics of Patients Hospitalized with Laboratory-Confirmed Coronavirus Disease 400 2019 -COVID-NET, 14 States

Longitudinal Data Analysis Using Generalized Linear Models

Comparison 404 of two different approaches for the analysis of data from a prospective cohort study: an 405 application to work related risk factors for low back pain

Hospital Admission, and Intensive Care Among 2,026,227 United States Veterans Aged

Clinical characteristics of 113 deceased 411 patients with coronavirus disease 2019: retrospective study

Clinical course and risk factors for mortality of adult 413 inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Case-Fatality Rate and Characteristics of Patients Dying in 416 Relation to COVID-19 in Italy

Risk Factors Associated With Acute Respiratory 418

Distress Syndrome and Death in Patients With Coronavirus Disease

Presenting 421 Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With