Characteristics of Hospitalized COVID-19 Patients Discharged and Experiencing Same-Hospital Readmission — United States, March–August 2020

Coronavirus disease 2019 (COVID-19) is a complex clinical illness with potential complications that might require ongoing clinical care (1-3). Few studies have investigated discharge patterns and hospital readmissions among large groups of patients after an initial COVID-19 hospitalization (4-7). Using electronic health record and administrative data from the Premier Healthcare Database,* CDC assessed patterns of hospital discharge, readmission, and demographic and clinical characteristics associated with hospital readmission after a patient's initial COVID-19 hospitalization (index hospitalization). Among 126,137 unique patients with an index COVID-19 admission during March-July 2020, 15% died during the index hospitalization. Among the 106,543 (85%) surviving patients, 9% (9,504) were readmitted to the same hospital within 2 months of discharge through August 2020. More than a single readmission occurred among 1.6% of patients discharged after the index hospitalization. Readmissions occurred more often among patients discharged to a skilled nursing facility (SNF) (15%) or those needing home health care (12%) than among patients discharged to home or self-care (7%). The odds of hospital readmission increased with age among persons aged ≥65 years, presence of certain chronic conditions, hospitalization within the 3 months preceding the index hospitalization, and if discharge from the index hospitalization was to a SNF or to home with health care assistance. These results support recent analyses that found chronic conditions to be significantly associated with hospital readmission (6,7) and could be explained by the complications of underlying conditions in the presence of COVID-19 (8), COVID-19 sequelae (3), or indirect effects of the COVID-19 pandemic (9). Understanding the frequency of, and risk factors for, readmission can inform clinical practice, discharge disposition decisions, and public health priorities such as health care planning to ensure availability of resources needed for acute and follow-up care of COVID-19 patients. With the recent increases in cases nationwide, hospital planning can account for these increasing numbers along with the potential for at least 9% of patients to be readmitted, requiring additional beds and resources.

increases in cases nationwide, hospital planning can account for these increasing numbers along with the potential for at least 9% of patients to be readmitted, requiring additional beds and resources.
Data for this study were obtained from the Premier Healthcare Database, which includes discharge records from 865 nongovernmental, community, and teaching hospitals that contributed inpatient data during the study period. COVID-19 patients were identified through International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) discharge diagnosis code of U07.1 (COVID-19, virus identified) during April-July 2020 or B97.29 (Other coronavirus as the cause of disease classified elsewhere [recommended before the April 2020 release of U07.1] † ) during March-April 2020. Both codes were used for discharges during April. The patient's first hospitalization with a COVID-19 discharge diagnosis was defined as the index hospitalization. Any subsequent hospitalization occurring within 2 months of the index hospitalization discharge date through August 2020, whether for COVID-19 or other health complications, was considered a hospital readmission. § Hospital readmissions that occurred >2 months after the index hospitalization were excluded. In the Premier Healthcare Database, readmissions were only recorded if a patient returned to the same hospital where the index hospitalization occurred.
Demographic and clinical characteristics of patients at their index hospitalization were compared regarding discharge disposition and readmission status (none versus one or more). Presence of selected chronic conditions associated with a more severe COVID-19 clinical course were identified through ICD-10-CM diagnosis codes during or before the index COVID-19 hospitalization. Visits before the index hospitalization included all inpatient encounters for the cohort during calendar year 2020 only. Five chronic conditions that have been identified by CDC to increase or possibly increase the risk for severe COVID-19-associated illness (chronic obstructive pulmonary disease, heart failure, diabetes [type 1 or type 2, with chronic complications], chronic kidney disease, and obesity [body mass index ≥30 kg/m 2 ], including † https://www.cdc.gov/nchs/data/icd/Announcement-New-ICD-code-forcoronavirus-3-18-2020.pdf. § Two months was twice the period used by the Centers for Medicare & Medicaid Services for unplanned readmission measures as knowledge of COVID-19 has been evolving regarding acute and chronic sequelae. severe obesity, [body mass index ≥40 kg/m 2 ]) were mapped to ICD-10-CM codes using the Elixhauser Comorbidity Index (a method for classifying comorbidities based on ICD diagnosis codes found in administrative data; each comorbidity category is dichotomous [present or absent]) and implemented with the Elixhauser Comorbidity Software for ICD-10-CM (beta version; Agency for Healthcare Research and Quality) and R software (version 4.0.92020; The R Foundation) ¶ (10). The following three clinical severity indicators were defined using hospital chargemaster records (i.e., the comprehensive list of all items billable to a hospital patient or to a patient's insurance provider): intensive care unit (ICU) admission, invasive mechanical ventilation, and noninvasive ventilation. Time to readmission after the index hospitalization was calculated as the difference in days between date of readmission and date of discharge from the previous hospitalization. The primary discharge diagnosis for each hospitalization was categorized into Clinical Classification Software Refined Categories to approximate the primary reason for the hospital stay. A multivariable generalized estimating equation model assessed the odds of readmission, accounting for within-facility correlation.
Covariates included in the model were age, sex, race/ethnicity, presence of selected chronic conditions, discharge disposition category, and clinical severity indicators. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy.** During March-July 2020, a total of 126,137 patients within the Premier Healthcare Database were hospitalized for COVID-19. The majority of patients were admitted from a non-health care setting (81%), followed by transfer from another hospital, clinic, or SNF (18%) ( Table 1). During the index hospitalization, 15% of patients were admitted to an ICU, 13% required invasive mechanical ventilation, and 4% required noninvasive ventilation. At the time of the index hospitalization or at any time during 2020 before the hospitalization, 62% of patients had an ICD-10-CM diagnosis code for one or more of the following five chronic conditions: chronic obstructive pulmonary disease (21%), heart failure (16%), diabetes mellitus type 1 or type 2 (27%), chronic kidney disease (21%), or obesity (27%). Overall, 10,008 (8%) patients had been hospitalized at the same hospital in the 3 months preceding their index COVID-19 hospitalization. Approximately 15% of patients (19,594) died during the index hospitalization.
Among the 106,543 patients discharged from the index admission, 9,504 (9%) were readmitted, including 1,667 (1.6%) who were readmitted more than once. The median ¶ https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/peoplewith-medical-conditions.html. ** 45 C. interval from discharge to first readmission was 8 days (interquartile range = 3-20 days). Less than 0.1% of patients died during readmission (data suppressed for privacy). Among all patients who were discharged after the index hospitalization, 60% were discharged to home or self-care (to home without any additional professional services provided such as home nursing health care), 15% to a SNF, 10% to home with assistance from a home health organization, 4% to hospice, 4% to ongoing care, and 5% to other locations ( Table 2). Readmission was more common among patients discharged to a SNF (15%) or with home health organization support (12%), compared with patients discharged to home or self-care (7%). Median age, severity markers, time to readmission and length of stay differed by index hospitalization discharge disposition category.
When controlling for covariates, the odds of readmission increased with the presence of chronic obstructive pulmonary disease (OR = 1.4), heart failure (OR = 1.6), diabetes (OR = 1.2), and chronic kidney disease (OR = 1.6). Patients were more likely to be readmitted if they had been discharged from the index hospitalization to a SNF (OR = 1.4) or with home health organization support (OR = 1.3) than if they had been discharged to home or self-care. Compared with persons aged 18-39 years, the odds of readmission increased with age among persons aged ≥65 years (Table 3). Adjusted odds of readmission of patients with a hospitalization in the 3 months preceding their index hospitalization were 2.6 times the odds of those who were not hospitalized in the preceding 3 months. Non-Hispanic White persons were more likely to be readmitted than were those of other racial/ethnic groups. Common primary discharge diagnoses after readmission were infectious and parasitic diseases (primarily COVID-19; 45%) and diseases of the circulatory (11%) and digestive (7%) systems (Supplementary Table, https://stacks.cdc.gov/view/cdc/96391).

Discussion
In a cohort of 106,543 patients discharged after an index COVID-19 hospitalization, 9% experienced at least one readmission to the same hospital within 2 months of discharge. More than one readmission occurred in 1.6% of cases. In this analysis, the odds of hospital readmission increased with age among persons aged ≥65 years, presence of one of five selected chronic conditions, hospitalization within the 3 months preceding the index hospitalization, and if discharge from the index hospitalization was to a SNF or to home with health care assistance. Although the proportions of patients in the Premier Healthcare Database cohort who were non-Hispanic Black (23%) or Hispanic (21%) were higher than those proportions in the U.S. Census (13% and 18%, respectively), their odds of readmission were lower than those of non-Hispanic White patients. The slight association of readmission with lengths of stay for hospitalized COVID-19 patients merits further study. These results are comparable to those of recently published analyses, which found a similar group of chronic conditions to be significantly associated with hospital readmission (6,7) and could be explained by the complications of underlying conditions in the presence of COVID-19 (8), COVID-19 sequelae (3), or indirect effects of the COVID-19 pandemic (9). Although only a small proportion of patients discharged to home or self-care were readmitted, 7% returned to the hospital within a median of 7 days. One explanation for their readmission is that approximately two thirds of these 4,406 patients had one or more of the selected chronic conditions. After hospitalization for COVID-19, the most common primary discharge diagnoses from hospital readmission were diseases of the circulatory, digestive, or respiratory systems. Future work will examine the detailed diagnoses recorded during readmissions to better understand COVID-19 sequelae or health conditions that require extended or ongoing care.
The findings in this report are subject to at least five limitations. First, COVID-19 diagnoses were determined by ICD-10-CM, not through laboratory confirmation, potentially leading to misclassification of cases. Second, chronic conditions were identified using ICD-10-CM diagnostic codes used at the index hospitalization or a previous encounter. If a patient had a chronic condition but the condition was not assigned a diagnostic code, that condition would not be recorded in this