Determinants of hospital outcomes for COVID-19 infections in a large Pennsylvania Health System

There is growing evidence that racial and ethnic minorities bear a disproportionate burden from COVID-19. Temporal changes in the pandemic epidemiology and diversity in the clinical course require careful study to identify determinants of poor outcomes. We analyzed 6255 individuals admitted with PCR-confirmed COVID-19 to one of 5 hospitals in the University of Pennsylvania Health System between March 2020 and March 2021, using electronic health records to assess risk factors and outcomes through 8 weeks post-admission. Discharge, readmission and mortality outcomes were analyzed in a multi-state model with multivariable Cox models for each transition. Mortality varied markedly over time, with cumulative incidence (95% CI) 30 days post-admission of 19.1% (16.9, 21.3) in March-April 2020, 5.7% (4.2, 7.5) in July-October 2020 and 10.5% (9.1,12.0) in January-March 2021; 26% of deaths occurred after discharge. Average age (SD) at admission varied from 62.7 (17.6) to 54.8 (19.9) to 60.5 (18.1); mechanical ventilation use declined from 21.3% to 9-11%. Compared to Caucasian, Black race was associated with more severe disease at admission, higher rates of co-morbidities and low-income resident zip code. Between-race risk differences in mortality risk diminished in multivariable models; while admitting hospital, increasing age, admission early in the pandemic, and severe disease and low blood pressure at admission were associated with increased mortality hazard. Hispanic ethnicity was associated with fewer baseline co-morbidities and lower mortality hazard (0.57, 95% CI: 0.37, .087). Multi-state modeling allows for a unified framework to analyze multiple outcomes throughout the disease course. Morbidity and mortality for hospitalized COVID-19 patients varied over time but post-discharge mortality remained non-trivial. Black race was associated with more risk factors for morbidity and with treatment at hospitals with lower mortality. Multivariable models suggest there are not between-race differences in outcomes. Future work is needed to better understand the identified between-hospital differences in mortality.


Introduction
subsequently hospitalized individuals, including demographics, severity of illness amongst those 82 infected, and changes in practice patterns further complicate disentangling the independent 83 factors that contribute to patient outcomes. 84 The disease course of COVID-19 varies considerably across individuals, even among those 85 severe enough to be hospitalized, with some hospitalized patients recovering after a brief or 86 extended stay, others dying in hospital, while still others experience a worsening of disease and 87 possible death or readmission after initial discharge. In the large body of work appearing in the 88 medical literature to date that examines COVID-19 outcome, multivariate analyses of patient Philadelphia, of which two were tertiary referral centers and all had major training programs; 115 two were in outlying non-urban areas. Characteristics of these hospitals are presented in S1 116 Appendix. We abstracted demographic information, vital sign measurements, body mass index 117 (BMI), recent comorbidity diagnoses, and hospitalization outcomes. Socioeconomic status was 118 assigned according to the median household income for the patient's residential zip code from 119 the U.S. Census Bureau's American Community Survey (ACS) for 2014-2018 [11]. Prior to 120 study start, this study was approved and deemed exempt by the University of Pennsylvania 121 Institutional Review Board (UPenn IRB). A limited dataset was generated from the EHR using 122 the minimum necessary protected health information that was authorized by the UPenn IRB. 123 124 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint Baseline comorbidities designated by the CDC as underlying conditions that may increase risk of 125 severe illness from COVID-19 [12] were abstracted from the EHR for the year prior to 126 admission and classified using the International Classification of Diseases, 10 th Revision (ICD-127 10). We considered diabetes (Type 1 or Type 2), obesity, chronic kidney disease, chronic liver 128 disease, chronic respiratory disease, cardiovascular disease, cancer, and immune deficiency. 129 More specific categorizations were considered for cardiovascular disease (coronary artery 130 disease, hypertension, congestive heart failure, and other) and chronic respiratory disease 131 (COPD, asthma, and chronic oxygen requirement). Further details are in the S1 Appendix. was selected as the baseline value. Blood pressure was classified as low if systolic blood pressure 138 (SBP) < 90 millimeters of mercury (mm Hg) or diastolic blood pressure (DBP) < 60 mm Hg; 139 otherwise, subjects were classified as having high blood pressure if SBP > 130 mm Hg or DBP > 140 80 mm Hg. Associations between patient baseline characteristics were evaluated using the chi-141 squared test. 142 143 Patient outcomes included hospital discharge, readmission, and mortality within 56 days of first 144 hospital admission. Cumulative incidence curves for mortality and discharge for the first hospital 145 stay were computed [13]. Discharge and death were analyzed as competing risks, as death 146 following COVID recovery is presumed to be occurring at a different rate than for patients 147 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint requiring hospitalization. For this analysis, the discharge outcome of interest was defined as 148 discharge followed by survival to day 56; patients who remained in hospital for longer than 56 149 days were censored at day 56. 150 A multi-state model was created to investigate the determinants of risk for each of the potential 151 outcome (states) patients experience throughout the disease course and incorporated data from all 152 events on a patient within the follow-up period. Separate   admitted at a much a higher proportion to the urban Philadelphia sites (sites 1, 3, 5), which were 304 associated with an overall lower hazard of death relative to the other two sites (Fig 2). Because 305 the urban sites included the majority of patients and had a higher proportion of Black patients, 306 the unadjusted comparison of outcomes by race shows overall better outcomes for Blacks 307 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint compared to Whites (Fig 4). In our base multivariable model, there was no association between 308 Black race and outcome for any of the transitions after adjustment for admitting hospital (Figs 2  309 and 3, S3 and S4 Figs). Adjusting for comorbidities in model 2 yielded similar results, although a 310 small beneficial effect for black race remained marginally significant for the hospital to 311 discharge and discharge to re-hospitalization transitions after adjusting for admitting hospital 312 increasingly reserved for a more severely ill patient population. In parallel, increased community 324 spread over time led to admissions of younger and overall healthier individuals, which 325 contributed to the drop in mortality rates mid-way through the pandemic; however, older age 326 groups made up a larger percent of admissions during both winter periods. With potential new 327 waves of infection, changes in circulating virus variants and expanded vaccine uptake will likely 328 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint precipitate further shifts in the demographics and clinical outcomes of those acquiring infection 329 and experiencing severe illness. 330 In addition to exploring trends over time, our analysis employed a novel multi-state model to 331 investigate determinants of patient outcomes including death, discharge and hospital re-332 admission. This approach allowed us to capture the movement of patients between these states 333 with more detail compared to the traditional approaches that have dominated the COVID-19 334 literature, which presented regression analyses for a single dichotomized (two-state) outcome 335 such as 30-day mortality. Unlike the traditional two-state Cox or logistic model approach, the 336 multi-state analysis also allows for data on multiple outcomes for a single patient, further 337 allowing for nuanced assessments, such as directly modeling the number of prior admissions or 338 risk factors affecting negative outcomes after discharge. The simplified, single outcome 339 strategies have been used in similar COVID-19 cohort studies to show that age, severe illness at 340 admission, existing comorbidities and socioeconomic factors may impact hospital outcomes 341 [2,3,18]. Our analysis largely corroborated these results but also highlighted the nuanced 342 associations with death, discharge, and re-hospitalization rates. There was appreciable variation 343 in mortality rates by admitting hospital that could not be explained by the observed patient 344 characteristics. These results are consistent with those of a recent study of nearly 40,000 patients 345 from 955 US hospitals, which found that standardized 30-day mortality rates varied from 9.1 to 346 15.6% [8]; however, there were few identified factors associated with these differences. 347 Our analyses found that, at time of hospital admission, Black patients had more risk factors 348 associated with poor outcomes compared to other races, including a higher proportion with EHR-349 documented comorbidities, residence in a low-income zip code, and requirement of ICU-level 350 care at admission. Despite this finding, unadjusted comparisons of the mortality rates in Black 351 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint patients compared to White patients found that Black race was associated with better outcomes. 352 After adjusting for a wide set of demographic and clinical risk factors, however, we did not find 353 an effect of race on death. There were no strong associations in multivariate models for the other 354 outcomes (transitions). Notably, we found that ignoring admitting hospital in our analyses led to 355 qualitatively different results regarding the impact of race. This is an example of Simpson's 356 COVID-19 to one of 1,188 hospitals across the US, which also found that unadjusted differences 364 in mortality by race were significant but were undetectable after adjusting for admitting hospital 365 [20]. In contrast to our study, however, Asch et al. [20] found that in their Medicare population, 366 Black patients were admitted in larger proportion to hospitals with poorer outcomes. These 367 investigators hypothesized that if Blacks were admitted to better performing hospitals, they 368 would have better outcomes overall. Our results support this hypothesis. Further research is 369 needed to understand whether public insurance status was also affecting the two racial groups 370 differently. 371 Our results contribute to delineating the complex role of race and racial health disparities in 372 patient exposures, pre-existing comorbidities, and outcomes from COVID-19. Several authors 373 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint have discussed that structural factors are critical driving forces behind COVID-19 disparities and 374 that system level reforms and interventions are required to fully address them [21,22]. 375 In contrast to race, Hispanic/Latinx ethnicity was associated with a decreased mortality hazard. 376 This result should be interpreted with caution, however, given the relatively small number of 377 Hispanic/Latinx individuals who were younger and had fewer comorbidities in our cohort. A key 378 limitation of our analysis is the reliance on EHR, which could introduce confounding if the level 379 of completeness was disproportionally affected by ethnicity. 380 Our findings corroborate the impact of COVID-19 in the elderly seen broadly. Patients over age 381 75 were at greatest risk for death and/or readmission. Improved strategies for post-discharge care 382 of COVID-19 patients are needed to address the sustained vulnerability of these patients [23]. 383 Passive data collection through routine electronic health records facilitates rapid research; 384 however, the data are subject to a number of limitations. Records of comorbidities at hospital 385 admission may be incomplete. Further, comorbidities diagnosed and treated outside the Penn 386 health care system may not have been captured. Information regarding discharge destination (e.g. 387 independent living or nursing home facility) was not available. Penn EHR may also have missed 388 readmissions or deaths that occurred outside the Penn system. Linkage with federal and state 389 registries will provide more complete data on survival, but these registries take time to be 390 updated. Future work is needed to fully capture long-term outcomes of COVID-19. 391 392 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

404
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review) preprint
The copyright holder for this this version posted September 13, 2021. ;  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review) preprint
The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Section 1.2 ICD-10 Code comorbidity definitions
Data on pre-existing conditions in patients was gathered from ICD-10 codes documented in the past year based on the patient's electronic health record in Penn Medicine. Conditions were grouped according to the conventions below. Respiratory diseases and chronic oxygen requirement were only included if the diagnosis occurred at least one month before the patient was admitted to the hospital with COVID-19 in order to avoid categorizing COVID symptoms as pre-existing conditions. Bold content is the title of the recommended sections as specified by the ICD10 tabular.

Condition ICD10 Codes Notes
Diabetes E08 -E13 "Diabetes mellitus." These are all the codes for diabetes mellitus excluding neonatal and gestational diabetes. Diabetes Type I E10 "Type 1 diabetes mellitus" Diabetes Type II E11 "Type 2 diabetes mellitus" Obesity E66 (excluding E66.3) "Overweight and Obesity." All sections of E66 code for obesity besides E66.3 which codes for overweight. We may also use given height and weight data or BMI (coded in Z68) Chronic Kidney Disease N18, N19 "Chronic kidney disease (CKD)" and "Unspecified kidney failure" Chronic Liver Disease "Coronary atherosclerosis." These are the codes that the ICD10 says code for coronary atherosclerosis. Hypertension I10 "Essential (primary) hypertension." Excludes primary hypertension of the brain (I60-I69) and eye (H35.0). Also excludes hypertensive heart (I11), chronic kidney (I12), heart and chronic kidney (I13), secondary hypertension (I15), and hypertensive crisis (I16). Congestive Heart Failure   I50 "Heart failure." The ICD10 doesn't specify congestive, but this seems to cover congestive heart failure. Includes all types of heart failure excluding cardiac arrest.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint S1 Table Comorbidity at COVID-19 hospital admission by race. N=6255. Percentages may not add up to 100% due to missing data in median household income or rounding elsewhere. P-values from chisquared test. Any ICD-10 comorbidity included diabetes, cardiovascular disease, respiratory disease, kidney disease, liver disease, immune deficiency, and chronic oxygen requirement. See Supplemental S1 Appendix for more details. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ;

S2 Table Comorbidity at COVID-19 hospital admission by Hispanic/Latinx ethnicity.
N=6255. Percentages may not add up to 100% due to missing data in median household income or rounding elsewhere. P-values were for the chi-squared test. Any ICD-10 comorbidity included diabetes, cardiovascular disease, respiratory disease, kidney disease, liver disease, immune deficiency, and chronic oxygen requirement. See Supplemental S1 Appendix for more details. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.    . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Characteristic
Level HR (95% CI) p−Value 0.5 1 5 10 50100 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted September 13, 2021. ; https://doi.org/10.1101/2021.09.08.21263311 doi: medRxiv preprint